The last year in paid media,
kept current.

What changedGoogle announced that effective June 15, 2026, all Google Ads data controls will be unified under the Consent Mode framework, centralizing how consent and data-use signals govern Google Ads measurement and personalization across regions.

Why it mattersConsent Mode stops being an EEA-only compliance chore and becomes the single control plane for what data Google can use. Operators in all regions, not just the EU, need a clean, correctly wired consent signal to keep full measurement and bidding capability.

Do thisTreat Consent Mode as global infrastructure, not an EU checkbox. Make sure your CMP and tag setup are ready to govern all Google Ads data use ahead of June 15, 2026.

What changedEffective June 1, 2026, Google Ads applies new data retention limits: hourly, daily, and weekly (sub-monthly) granular reporting data is retained for only 37 months; monthly, quarterly, and annual data for 11 years; and reach & frequency metrics for 3 years. After those windows the data is no longer accessible via the Google Ads interface or APIs, and Google Ads API / Scripts queries requesting granular segments (e.g., segments.date or segments.week) for ranges older than 37 months will return a date-range error.

Why it mattersLong-range, day-level year-over-year and seasonality analysis silently breaks once data ages past 37 months, and any reporting pipeline, dashboard, or data warehouse that pulls granular historical segments via the API will start failing rather than warning. For operators who rely on multi-year granular history for audits, forecasting, and bidding diagnostics, the data simply disappears unless archived.

Do thisBefore June 1, 2026, export granular historical Google Ads data (manual download, the Google Ads API, or a linked GA property) into your own warehouse so multi-year day/week-level history survives. Update any API or Scripts reporting that requests granular segments older than 37 months to fall back to monthly-or-coarser segments so jobs don't error out.

What changedGoogle began rolling out the May 2026 broad core update on May 21, 2026, announced via its Search Central account with the note that the rollout 'may take up to 2 weeks to complete.' Google described it as the second confirmed core update of 2026 and published no companion blog post or specific guidance. It is the first confirmed broad core update after a long stretch of relative core-update calm, and it landed in parallel with Google's I/O 2026 AI Search expansion.

Why it mattersCore updates are the single biggest source of organic-traffic swings for client sites. A B2B SaaS site that moves in rankings in late May or early June 2026 should attribute it to this rollout, not to its own recent changes, and should avoid knee-jerk content edits while the update is still in flight.

Do thisPull GSC organic clicks for the May 21 onward window per client, annotate any step-change as core-update-driven, and hold off on other large changes mid-rollout so you can isolate cause. For losers, audit helpfulness and E-E-A-T rather than chasing quick fixes, and wait for the rollout to finish before judging impact.

What changedAt Google I/O 2026 (May 19, 2026), Google said AI Mode has surpassed one billion monthly users about a year after launch, with queries more than doubling each quarter. It unveiled its biggest Search box redesign in over 25 years — dynamically expanding, accepting multimodal input (text, images, files, videos, Chrome tabs), and offering AI-powered suggestions beyond autocomplete — made Gemini 3.5 Flash the default model powering AI Mode globally, and previewed agentic features (24/7 information agents, expanded agentic booking and calling, real-time generative UI dashboards, and Personal Intelligence across nearly 200 countries and 98 languages).

Why it mattersAI Mode is now a billion-user default surface, cementing the conversational, agentic SERP as the primary top-of-funnel for B2B SaaS buyers. Classic rank tracking is increasingly blind to where buyers actually research, and longer planning/conversational queries are growing fastest, so being citable across a multi-turn buyer journey matters more than ranking #1 for a single head term.

Do thisTrack your brand and category answer share inside AI Mode as a core KPI alongside organic rank, and structure content to answer multi-turn, conversational buyer questions in self-contained, quotable blocks. Plan for continued erosion of classic blue-link clicks on informational queries and protect bottom-funnel pages where intent still drives action.

What changedFor AI-powered search experiences, Google's Search Terms report can now display a normalized or inferred meaning/intent of a search rather than the exact words the user typed. Reported terms may be queries the user never literally entered. Separately, a 2025 change began surfacing previously hidden misspelled queries (grouped under the correct spelling), adding roughly 9% more search-term data.

Why it mattersSearch-term analysis is the operator's core lever for negatives and intent QA. If reported terms are AI interpretations rather than literal queries, negative-keyword decisions and intent audits get fuzzier, and it is harder to police exactly what AI Max/broad match is matching against.

Do thisTreat the Search Terms report as directional intent rather than literal queries going forward. Lean more on conversion-quality and search-category signals (and offline conversion feedback) to police match quality, and keep adding negatives but expect less precision than the literal-query era.

What changedAfter the June 2025 core update finished July 17, 2025, the only confirmed broad action in the window was the September 2025 spam update. Multiple algorithm-tracking timelines show no further confirmed core update through the current date (May 2026), even as third-party tools logged periodic unconfirmed volatility — consistent with Google's stated shift toward more continuous, rolling ranking changes rather than discrete named core updates.

Why it mattersOperators waiting for a 'next core update' to recover lost rankings may be waiting on an event that does not arrive in a tidy named package. Ranking shifts increasingly come from continuous systems, so recovery and monitoring should be ongoing rather than event-driven.

Do thisStop framing recovery around 'the next core update'; monitor volatility continuously (GSC 24-hour view + a rank tracker), and treat unconfirmed-volatility days as normal noise unless a sustained multi-week trend appears.

What changedGoogle published an official guide on optimizing for its AI search features (AI Overviews and AI Mode), stating plainly that 'optimizing for generative AI search is optimizing for the search experience, and thus still SEO' — i.e., AEO and GEO are not separate disciplines. The guide explicitly debunks five tactics as unnecessary or ineffective for earning Google AI citations: machine-readable files like llms.txt, AI-specific content chunking, AI-specific content rewriting, inauthentic/bought brand mentions, and excessive structured-data focus. Google treats manipulated citations and bought brand mentions as a violation, comparing them to paid links its systems detect and ignore.

Why it mattersThis is Google's official position on the GEO playbook, and it kills several tempting shortcuts. Operators should not spend budget on llms.txt, AI-only content reformatting, or paid 'AI mention' services to win Google AI citations; the durable path remains quality SEO plus genuine earned mentions. The caveat is that this is scoped to Google citations specifically and may not govern how autonomous AI agents outside Google's ecosystem consume content.

Do thisCut GEO spend on llms.txt and AI-mention-buying tools, and redirect it to citable, well-structured content and authentic earned coverage. Keep doing strong technical and content SEO as the foundation for AI visibility, and treat any vendor selling guaranteed 'AI mentions' as a paid-link-style risk.

What changedGoogle announced that legacy Search automation tools are being folded into AI Max: Dynamic Search Ads, automatically created assets (ACA), and the campaign-level broad match setting are being automatically upgraded/migrated to AI Max, with the migration scheduled around September 2026. AI Max came out of beta and became Google's default path for Search automation.

Why it mattersThis makes AI Max non-optional for anyone relying on DSA or ACA. The familiar DSA workflow disappears, and broad-match-on campaigns inherit AI Max's keywordless expansion and text customization, changing matching and creative behavior whether or not you opted in.

Do thisAudit every campaign using DSA, automatically created assets, or the campaign-level broad match setting now. Pre-configure AI Max brand controls, URL inclusions/exclusions, text guidelines, and negatives before the auto-migration so you control the transition instead of inheriting defaults.

What changedStarting April 2026, advertisers can send user-provided data simultaneously via website tags, Data Manager, and the API; by June 2026 Enhanced Conversions for web and leads collapse into one on/off toggle, removing the need to pick an implementation method. Existing users migrate automatically.

Why it mattersSimplifies a historically fiddly setup and lets multiple data sources (tag + offline/CRM via Data Manager + API) reinforce each other. Better match rates mean better-fed Smart Bidding for B2B pipelines that close offline.

Do thisPlan to feed Enhanced Conversions from multiple sources at once (tags + Data Manager/CRM + API) once the unified toggle lands, rather than relying on a single pixel-based path.

What changedA Similarweb analysis found ChatGPT referrals to news sites grew roughly 25x (about 2,500%) over approximately a year, but the absolute volume remains small relative to Google and does not offset broader declines in traditional search referral traffic.

Why it mattersIt frames the real magnitude of AI referral traffic for operators: the growth rate is explosive but off a tiny base. AI answer engines are a rising channel worth building for, not yet a replacement for organic search volume.

Do thisTreat AI referrals as a high-growth, low-absolute channel: set up tracking now to capture the trend early, but don't reallocate major budget off search on the strength of percentage growth alone.

What changedGoogle announced it is winding down a large set of Privacy Sandbox advertising APIs, including Topics, Protected Audience, and Attribution Reporting, on Chrome and Android, ending its push for a single Chrome-led replacement for third-party cookies.

Why it mattersAny ad-tech roadmap that bet on Topics/Protected Audience for targeting or on the Attribution Reporting API for conversion measurement is now a dead end. The industry reverts to cookies plus first-party/server-side measurement plus modeled conversions as the durable stack.

Do thisAudit your measurement and targeting stack for any dependency on Privacy Sandbox APIs and remove it. Consolidate around Enhanced Conversions, server-side tagging, and first-party data instead.

What changedGoogle began serving ads within and below AI Mode conversational responses in the US. Ads are eligible automatically from existing Search, Shopping, and Performance Max campaigns, and Google uses the context of the entire conversation (not just a single query) to choose ads, with limited AI-Mode-specific advertiser controls at launch.

Why it mattersTargeting now keys off a multi-turn conversation rather than one keyword, which both expands and blurs intent signals. For B2B SaaS, this can surface your ad against a richer buyer journey, but it also reduces query-level control and makes negative keywords and exclusions less precise.

Do thisTighten campaign-level brand safety and audience signals since keyword-level control weakens in AI Mode. Monitor lead quality from AI-Mode-attributed conversions closely in the first weeks.

What changedGoogle announced it will use the content and context of users' AI Mode conversations as signals for ad targeting, turning natural-language exchanges (topics, questions, follow-ups) into intent data for serving ads.

Why it mattersConversational data is far richer than a keyword, giving advertisers access to deeper intent signals, but it raises privacy scrutiny and shifts the unit of targeting from query to dialogue. SaaS operators should expect new intent audiences and new compliance questions.

Do thisBrief your team and legal/privacy stakeholders that conversation-derived targeting is coming, and prepare to test the richer intent signals while documenting consent/privacy posture for regulated SaaS buyers.

What changedBy October 2025, Google began auto-enabling Enhanced Conversions at the account level for advertisers who had not configured it, selecting a method based on existing setup. Enhanced Conversions sends SHA-256 hashed first-party data (e.g., email) to Google to recover otherwise-lost conversions.

Why it mattersEnhanced Conversions is now effectively the default conversion-accuracy layer, and it directly feeds Smart Bidding. For lead-gen B2B SaaS, it's the single biggest free win for recovering measurable conversions in a cookie-restricted, consent-gated world.

Do thisConfirm Enhanced Conversions is on and actually receiving hashed first-party data (email at minimum) at the conversion point. Verify the diagnostic status in Google Ads rather than assuming the auto-upgrade configured it correctly.

What changedGoogle launched the September 2025 spam update on September 22, 2025, finishing October 7, 2025 (about two weeks). It was the first confirmed spam update since December 2024 and used Google's automated systems (SpamBrain) to enforce spam policies including scaled content abuse, expired-domain abuse, and link spam.

Why it mattersSpam updates are the enforcement teeth behind the scaled-content and expired-domain policies. Any client leaning on programmatic/AI-mass-produced content or acquired domains was exposed in this window, and a sudden drop in late September 2025 likely traces here rather than to a core update.

Do thisAudit each client for mass-produced or thin programmatic pages and any expired-domain plays; if visibility dropped late Sept-early Oct 2025, treat it as a spam-policy hit (recovery requires removing/fixing the offending content, not a core-update wait).

What changedAround September 10-14, 2025 Google disabled the &num=100 URL parameter that let tools pull 100 results on one page; searches are now capped at 10 results per page. Many sites saw large drops in Search Console desktop impressions and an apparent improvement in average position, because rank-tracking bots that had been inflating impressions from positions 11-100 could no longer load those results.

Why it mattersThis is a data-integrity event, not a traffic event. Year-over-year impression and average-position comparisons that cross September 2025 are now apples-to-oranges, and most rank-tracking tools got more expensive or shallower overnight. Misreading the impression drop as a ranking loss would send an operator chasing a phantom problem.

Do thisAnnotate every client's GSC at mid-September 2025; switch YoY reporting to clicks (not impressions/avg position) across that boundary, and confirm with your rank tracker how deep it still tracks and whether pricing/coverage changed.

What changedIn September 2025 Google began rolling out a Recommendations report in Search Console that surfaces personalized, property-specific suggestions based on a site's own data — highlighting content worth expanding, queries gaining traction, and relevant Search Console features to act on.

Why it mattersIt gives operators a Google-sourced, per-property prioritization layer that previously required manual GSC analysis. Useful as a fast first pass when onboarding a new client account or triaging where to focus limited SEO hours.

Do thisOn each managed property, review the Recommendations report during monthly audits as a triage starting point; cross-check its content-expansion and query suggestions against your own keyword strategy before acting.

What changedGoogle's site reputation abuse policy (anti 'parasite SEO') applies regardless of whether the rank-leveraging content is produced by a third party OR the host's own first party — the first-party exemption was removed, and enforcement moved from manual actions toward broader algorithmic detection through 2025, reinforced by the September 2025 spam update. The test is whether content exists primarily to exploit a domain's ranking signals rather than to serve the site's regular audience.

Why it mattersAny B2B SaaS using a high-authority domain (or a partner/media domain) to host loosely related, rank-chasing content — coupon hubs, 'best tools' roundups built only for SEO, sponsored sections — is now squarely in scope. This closes a tactic many growth teams quietly relied on.

Do thisAudit client domains (and any earned placements on partner/media sites) for sections that exist purely to capture rankings off the domain's authority; bring them under genuine editorial ownership tied to the core audience, or remove them.

What changedGoogle's John Mueller said Google does not use llms.txt, comparing it to the old keywords meta tag, a self-reported signal that became useless because site owners stuffed and lied in it. The proposed llms.txt standard (a file telling LLMs what content to read) thus has no Google support for its AI features.

Why it mattersIt kills a tempting GEO shortcut. Operators shouldn't spend effort on llms.txt expecting Google AI Overviews/AI Mode to honor it; the path to AI visibility remains quality, citable, well-structured content plus earned mentions, not a new metadata file.

Do thisDon't prioritize llms.txt for Google visibility. If you add one for other tools it's low-cost, but allocate GEO effort to citable content and third-party validation instead.

What changedServer-side tagging shifted from niche to default best practice through 2025-2026 as client-side tags get blocked on a large share of sessions (ad blockers plus Safari ITP). The common pattern is a GTM server container on Cloud Run behind a first-party subdomain (e.g., metrics.yourdomain.com), recovering a large portion of otherwise-lost conversion signal.

Why it mattersClient-side-only tracking now loses a meaningful chunk of conversions, which starves Smart Bidding and understates ROAS. Server-side tagging is the durable fix for the post-reversal reality where Safari/Firefox/ad-blockers, not Chrome cookies, are the real leak.

Do thisStand up a server-side GTM container on a first-party subdomain for any account where measured conversions look light. Route Google/Meta/LinkedIn conversions server-side to lift match rates and recover signal.

What changedData clean rooms (Google Ads Data Hub, Amazon Marketing Cloud, Meta Advanced Analytics, plus neutral vendors) saw broader adoption through 2025 as the privacy-compliant way to match first-party data with platform data for measurement and audience overlap without sharing raw user-level records.

Why it mattersClean rooms are how larger advertisers now reconcile cross-platform exposure and outcomes without cookies. For B2B operators with meaningful first-party data, they enable real cross-channel dedup and incrementality that platform UIs can't provide on their own.

Do thisIf first-party data volume justifies it, evaluate a clean room (or a clean-room-backed vendor) for cross-platform measurement and overlap analysis; for smaller accounts, prioritize server-side plus MMM first.

What changedPew Research Center analyzed 68,879 Google searches from 900 US adults (data collected March 2025). Users clicked a traditional result link in only 8% of visits when an AI Overview appeared, versus 15% when none did. AI Overviews showed on about 18% of all searches, and users clicked a source link inside the Overview just 1% of the time.

Why it mattersThis is hard, independent confirmation that AI Overviews suppress organic clicks, and that being cited inside the Overview almost never sends traffic (1% click rate). For SaaS demand gen, informational/top-of-funnel content can no longer be modeled on the old click-through math; the AI answer is increasingly the terminal destination.

Do thisReforecast organic traffic assumptions for informational queries downward and shift those content goals from clicks to brand citation/share-of-voice inside AI answers. Protect bottom-funnel pages (pricing, comparisons, demos) where intent still drives clicks.

What changedA Pew Research Center study (published July 22, 2025, based on March 2025 browsing data from 900 US adults and 68,879 searches) found users clicked a traditional result link on just 8% of searches that showed an AI summary, versus 15% when no AI summary appeared. Only 1% clicked a link inside the AI summary itself, and AI summaries appeared on about 18% of searches.

Why it mattersThis quantifies the organic-traffic tax of AI Overviews on informational queries with hard third-party numbers. For B2B SaaS, top-of-funnel educational content (the classic SEO play) now converts far fewer impressions into clicks, which changes the ROI case for ranking on informational keywords versus bottom-funnel/commercial intent.

Do thisRe-weight SEO content strategy toward commercial/comparison/bottom-funnel queries (where AI Overviews are less dominant and clicks still convert) and away from broad informational terms; reset client traffic expectations using the 8% vs 15% benchmark.

What changedGoogle tightened enforcement of its EU User Consent Policy: accounts not passing valid Consent Mode v2 signals (ad_storage, analytics_storage, ad_user_data, ad_personalization) for EEA/UK traffic lose conversion modeling, personalization, and remarketing audience population. Some misconfigured accounts saw conversions drop up to ~90% overnight.

Why it mattersFor B2B SaaS running EU/UK demand-gen, a banner that collects consent but doesn't transmit the signals silently breaks Smart Bidding inputs and remarketing. Attribution gaps look like a performance problem but are actually a tagging-compliance problem.

Do thisVerify Consent Mode v2 is actually transmitting all four parameters to Google (use Tag Assistant or a certified CMP) for any account with EEA/UK traffic. Confirm conversion modeling is active and remarketing lists are still populating.

What changedGoogle began surfacing an 'AI Max' match type label in the Search Terms report so advertisers can see which queries were matched by AI Max's keywordless expansion versus their traditional keywords, giving some transparency into what the new layer is actually matching.

Why it mattersWithout this label, AI Max expansion would be indistinguishable from your keyword-driven traffic, making it impossible to evaluate or QA. The label lets operators isolate AI Max-sourced queries and decide whether the incremental matching is high-quality or needs reining in with negatives and brand controls.

Do thisSegment the Search Terms report by the AI Max match type to evaluate AI Max-sourced queries separately. If those queries show poor intent or weak conversion quality, tighten brand controls, URL exclusions, and negatives, or pause AI Max on that campaign.

What changedMeta introduced GEM (Generative Ads Recommendation Model), a foundation-model approach using sequence learning to model user behavior for ads ranking and recommendations. Meta reported large gains in ad performance from the new architecture, framing it as a paradigm shift in how the system predicts which ads a user will respond to.

Why it mattersGEM, layered with Andromeda, means the ranking and prediction stack is now LLM-style and behavior-sequence-driven. The practical effect is the same as Andromeda: outcomes are increasingly determined by signal quality and creative variety rather than manual targeting, and the system gets better at finding buyers from broad pools.

Do thisPrioritize feeding the model clean, deduplicated conversion events (especially deeper-funnel signals like qualified-lead or trial-started) so GEM/sequence learning has high-quality outcomes to optimize toward.

What changedGoogle rolled out the June 2025 broad core update starting June 30, 2025, completing July 17, 2025 (about 17 days). Like all core updates it re-weighted Google's core ranking systems site-wide rather than targeting one tactic, and third-party trackers logged substantial volatility across the rollout. This was the second confirmed core update of 2025, following the March 2025 core update (March 13-27).

Why it mattersCore updates are the single biggest source of organic-traffic swings for client sites. A B2B SaaS site that lost or gained visibility over July 2025 should attribute it to this update, not to its own changes, and should hold off on knee-jerk content edits during a rollout window.

Do thisPull GSC organic clicks for the June 30-July 17 window per client; flag any step-change, annotate it as core-update-driven, and avoid making other large changes mid-rollout so you can isolate cause. For losers, audit helpfulness/E-E-A-T rather than chasing quick fixes.

What changedMark Zuckerberg stated Meta intends to let advertisers fully create and target ad campaigns with AI by the end of 2026: a business states a goal and a budget, connects a bank account, and Meta's AI generates the creative (image and video), writes the copy, picks the audience, and optimizes the spend end-to-end.

Why it mattersThis is the strategic direction the whole platform is being pulled toward. It signals that manual campaign craft is being commoditized and that the durable edge for a paid-media operator shifts to strategy, offer, creative direction, measurement integrity, and incrementality, not button-pushing inside Ads Manager.

Do thisReposition your value (and client conversations) around what AI won't do: offer/positioning, creative concept quality, conversion-signal hygiene, and incrementality measurement. Start documenting incremental results now so you can defend spend when 'just let Meta do it' becomes the default pitch.

What changedMeta rolled out an 'incremental attribution' setting that lets advertisers optimize and report on conversions Meta deems incremental (driven by the ad) rather than all attributed conversions, moving reporting and optimization toward causal lift instead of last-touch credit.

Why it mattersStandard Meta attribution over-credits conversions that would have happened anyway. Incremental attribution gives operators a built-in, lift-oriented view, useful for defending Meta's true contribution in a B2B mix where Meta is often a demand-gen/assist channel, not last-click.

Do thisTest the incremental attribution conversion setting on a Meta campaign and compare reported volume/CPA against standard attribution to understand Meta's real incremental contribution before scaling budget.

What changedAnnounced at Google Marketing Live 2025, Smart Bidding Exploration lets tROAS campaigns flex below your strict ROAS target (generally a 10-30% band) to bid on valuable queries the fixed target was suppressing. It changes bidding, not targeting. Google reports campaigns using it average an 18% increase in unique search-query categories with conversions and a 19% increase in conversions; it later expanded into Performance Max and Shopping in beta.

Why it mattersIt loosens how tightly your ROAS target constrains the system, so the same target now allows more experimentation. That can surface net-new converting demand but also widens CPA/ROAS variance and loosens the cost discipline operators relied on, which can be misread as a tracking bug or account problem.

Do thisEnable on a campaign with healthy conversion volume and watch widened CPA/ROAS variance during a learning period. Judge it on net-new converting query categories and incremental conversions, not just blended efficiency, before keeping it on across the account.

What changedAt Google Marketing Live 2025, Google announced it is expanding ads in AI Overviews to desktop (after a mobile U.S. start) and bringing ads to AI Mode, its Gemini-powered conversational Search experience. Existing Search, Shopping, and PMax ads became eligible to appear in these AI answer surfaces, with new formats like conversational/Highlighted ad experiences.

Why it mattersAI answer surfaces are absorbing informational and zero-click queries that organic used to own. Google placing ads inside them means paid is increasingly how brands stay visible in AI results, and it shifts which top-of-funnel queries convert and how clicks attribute.

Do thisDo not assume AI Overview/AI Mode placements behave like classic Search. Monitor impression share, CTR, and CPC for shifts, keep responsive assets and feeds strong (these surfaces auto-assemble from your existing assets), and watch for top-funnel/informational traffic creeping into Search campaigns.

What changedAt Google Marketing Live 2025, Google announced its Veo (video) and Imagen (image) generative models are being built into Google Ads and Merchant Center, enabling advertisers to generate and edit images and produce/extend/resize video assets (including auto-filling missing aspect ratios) directly inside the platform for PMax, Demand Gen, and Search.

Why it mattersCreative is now the primary lever in AI-driven campaigns, since PMax and Demand Gen optimize on the assets you feed them. Native generation lowers the cost of producing the volume and variety these campaign types need, which historically bottlenecked lean B2B teams.

Do thisUse native generation to expand creative variety into PMax/Demand Gen asset groups, but keep brand-controlled hero assets and human-reviewed copy for B2B credibility. Treat generated assets as volume the algorithm can test, not your primary message, and watch for off-brand outputs.

What changedGoogle removed the Labs opt-in gate and began rolling AI Mode out to all US users as a dedicated tab in Google Search and the Google app, powered by a custom version of Gemini 2.5. It had previously been limited to Labs testers.

Why it mattersAI Mode moved from an experiment to a default surface for the entire US market essentially overnight. For a B2B SaaS operator, that means a growing share of branded and category research now happens in a conversational pane where traditional rank tracking is blind, and where the answer (and its citations) is what the buyer sees first.

Do thisStart logging how your brand, top competitors, and category are described inside AI Mode for US queries. Treat AI Mode answer share as a new tracked KPI alongside organic rank.

What changedAt Google Marketing Live, Google announced testing of Search and Shopping ads inside AI Mode and expanded AI Overviews ads to desktop (previously mobile-only), placing them at the top and bottom of Overviews in the US in English. Ads draw from existing Search, Shopping, and Performance Max campaigns automatically.

Why it mattersGoogle is monetizing its AI answer surfaces using campaigns advertisers already run, so a SaaS account can start serving inside AI experiences without building anything new. It also means ad placement is now coupled to AI surfaces you can't fully control or see in standard placement reports.

Do thisConfirm whether your Search/PMax campaigns are opted into AI surfaces and review search terms/placement data for AI-driven impressions. Don't blanket-exclude; instead watch CPA by surface as the inventory scales.

What changedAt Google I/O, Google said AI Overviews now reach over 1.5 billion users monthly across 200+ countries and 40+ languages, and introduced AI Mode (a full conversational search experience built on a custom Gemini 2.5) using a 'query fan-out' technique that breaks one question into many simultaneous sub-queries. AI Mode added Deep Search, agentic task completion (Project Mariner), personalization via Gmail, and agentic shopping checkout.

Why it mattersThis is the structural shift behind everything else: the dominant top-of-funnel surface for B2B SaaS buyers is now an AI answer, not a list of blue links. Query fan-out means a single buyer question can pull from dozens of pages, so a SaaS site has to be citable across many sub-topics, not just rank #1 for one head term.

Do thisAudit your highest-intent commercial and comparison queries inside AI Mode and AI Overviews, not just classic SERPs. Map which sub-questions Google fans out to and make sure your site answers each one in a directly quotable, self-contained block.

What changedGoogle launched AI Max for Search, an optional one-click optimization layer (not a new campaign type) you toggle on inside an existing Search campaign. Its two pillars are 'search term matching' (broad-match + keywordless expansion that learns from your keywords, assets, and URLs to match queries you never added) and 'asset optimization' (AI text customization of headlines/descriptions plus final-URL expansion). Google claims ~14% more conversions at similar CPA/ROAS, and up to 27% for campaigns still mostly on exact/phrase. Beta opened late May 2025 with global rollout reaching all advertisers by early Q3 2025.

Why it mattersFor B2B SaaS accounts built on tight keyword and match-type control, AI Max blurs Search into PMax-style automation. It can capture incremental converting queries but also pulls in irrelevant and competitor-adjacent traffic, so it demands strong negatives, brand controls, and a hard down-funnel conversion goal before you trust it.

Do thisPilot AI Max on one well-instrumented Search campaign with a real down-funnel conversion (demo/SQL or offline closed-won, not raw form fills) as the bid target. Lock in brand controls, URL inclusions/exclusions, and a robust shared negative list first, then compare search-term quality and CPA against an untouched control before scaling.

What changedMeta expanded its Advantage+ creative generative AI suite, broadening access to image generation, text/copy variations, image expansion, and background generation, and adding more full-creative generation capabilities so advertisers can produce and resize creative variants natively in Ads Manager.

Why it mattersCreative volume is now the main lever in an Andromeda/GEM world, and Meta is lowering the cost of producing variants to near zero. For a lean B2B SaaS team, this means you can test more angles faster, but AI-generic output risks blending in, so human creative direction and brand-correct concepts become the differentiator.

Do thisUse Meta's generative tools for fast variant production (backgrounds, resizes, copy alternates) on top of strong human-made hero concepts, not as a replacement. Feed the system many variants but keep brand voice and offer clarity human-controlled.

What changedMeta rolled out AI video generation features for advertisers, including the ability to generate and animate video creative (e.g., turning images into short video and producing video variations) within its Advantage+ creative tools, extending generative AI from static images into motion.

Why it mattersVideo is the highest-performing format across Reels and feed, and Meta is removing the production bottleneck. B2B operators who previously skipped video for cost reasons can now generate motion creative to test, expanding inventory access (Reels, Stories) without a production budget.

Do thisTest AI-generated video variants for Reels and Stories placements on your best-performing static concepts. Measure whether motion lifts CTR and CPL before committing to a video-first creative strategy.

What changedMeta rebranded Conversions API Gateway to Signals Gateway, broadening it from purely conversion events to a unified server-side signals pipeline. The rebrand reflects Meta consolidating first-party data sharing (web, app, CRM, offline) into one signal infrastructure that feeds its AI optimization systems.

Why it mattersSignal quality is now the primary input that determines whether Andromeda/GEM can find buyers. A consolidated, server-side signals setup directly improves match rates and optimization, which matters even more for B2B where the conversion (qualified lead, demo, trial) is lower-volume and signal loss hurts disproportionately.

Do thisAudit your server-side setup and migrate to Signals Gateway (or a robust CAPI implementation) if you're still relying on the pixel alone. Send deeper-funnel CRM events back to Meta so the AI optimizes toward revenue-quality leads, not just form fills.

What changedThrough 2025 LinkedIn continued enhancing the Conversions API (CAPI) for better match rates and conversion matching, including support for additional user identifiers, and maintained a monthly/quarterly Marketing API versioning cadence (versions roughly supported about a year before sunset) with rolling tier migrations and deprecations.

Why it mattersBrowser-based tracking keeps degrading, so server-side CAPI is now the backbone of reliable LinkedIn conversion signal, and it's what feeds Accelerate, Predictive Audiences, and the Revenue Attribution Report. The versioning/deprecation cadence also means integrations silently break if not maintained, which can corrupt optimization without obvious warning.

Do thisMake sure CAPI (ideally plus offline conversions) is live and sending hard, revenue-proximate events with strong identifiers, and put the LinkedIn API version on a maintenance watch so deprecations don't silently kill your conversion feed.

What changedWith user-level attribution degraded by cookie restrictions and consent gating, geo-based holdout experiments (geo-lift) and platform conversion-lift tests became the standard way to validate true incremental impact, and are now used to calibrate MMMs like Meridian and Meta's models.

Why it mattersPlatform-reported ROAS increasingly overstates value because of modeling and over-attribution. Geo-lift is the cleanest causal read an operator can run on Google/Meta/LinkedIn spend, and it's the credible way to prove or disprove a channel's contribution to a founder.

Do thisBuild a recurring geo-lift / holdout testing cadence into the measurement plan and use the results to calibrate budget splits and any MMM, rather than trusting in-platform attribution alone.

What changedGoogle pushed Google Ads Data Manager as the central hub to connect and activate first-party data (CRM, offline conversions, warehouse connectors) for measurement and audiences, increasingly feeding Enhanced Conversions and conversion imports from one place.

Why it mattersFor B2B SaaS where the real conversion (MQL to closed-won) happens in the CRM weeks later, Data Manager is the bridge that gets offline/pipeline outcomes back into Google to train Smart Bidding on revenue, not just form fills.

Do thisConnect the CRM to Google via Data Manager so closed/qualified outcomes flow back as conversions, and optimize bidding toward pipeline-quality events instead of raw lead volume.

What changedAnthony Chavez (VP, Privacy Sandbox) announced Chrome will NOT deprecate third-party cookies and will not ship a standalone choice prompt; cookies stay on by default, managed through Chrome's existing Privacy and Security settings. This ended the multi-year cookie-deprecation countdown.

Why it mattersThe cookieless-future panic that drove years of measurement re-architecture is effectively over for Chrome. Operators no longer need to treat full cookie loss as imminent, but Safari and Firefox still block by default (~30-40% of traffic), so first-party and server-side measurement remain necessary, not optional.

Do thisStop treating cookie deprecation as a hard deadline. Keep investing in first-party data and server-side tracking for the Safari/Firefox/ad-blocker gap, but don't burn budget on Privacy Sandbox API integrations.

What changedGoogle rolled out a 24-hour view in Search Console Performance reports in April 2025, showing hour-by-hour search data from the last 24 hours with only a few hours' lag, versus the traditional 2-3 day delay.

Why it mattersFaster feedback loops let an operator gauge whether a new page, title rewrite, or technical fix is landing within hours instead of days. For agile SEO testing on client sites, this compresses the iterate-measure cycle dramatically.

Do thisUse the 24-hour view to validate new-page indexing/performance and title/meta tests same-day; build it into client reporting for launches and time-sensitive content.

What changedMeta began testing Business AI: conversational AI agents that can talk to prospects and customers across Messenger, Instagram, and WhatsApp, answering questions, qualifying interest, and helping convert from ads into chats. The agents are trained on a business's own info and tied into click-to-message ad flows.

Why it mattersThis pushes the conversion point from landing pages into AI-handled conversations, a meaningful shift for lead capture. For B2B SaaS, AI agents could pre-qualify leads from click-to-message campaigns, but they also introduce a new surface to monitor for accuracy and lead quality before scaling spend behind them.

Do thisPilot click-to-message campaigns with a Business AI agent for top-of-funnel qualification, but keep human review on transcripts early. Don't route high-value B2B leads entirely through an AI agent until you've verified qualification accuracy.

What changedOn April 3, 2025, LinkedIn launched the Wire Program, enabling brands to run in-stream video ads within professionally produced video content from trusted publishers. Launch partners included Bloomberg, Business Insider, Forbes, The Wall Street Journal, and NBCUniversal, with plans to expand the network.

Why it mattersWire extends LinkedIn video monetization to publisher in-stream placements, giving advertisers another premium, brand-safe video channel with professional targeting. Combined with BrandLink and CTV, LinkedIn now has a real video network beyond the feed, which changes how an operator can run upper-funnel video.

Do thisTreat Wire, BrandLink, and feed video as distinct video placements with distinct creative and KPIs. Use Wire/BrandLink for credible top-of-funnel reach to buyers; reserve feed and conversion campaigns for capture, and unify measurement through Revenue Attribution.

What changedGoogle completed the deprecation of Enhanced CPC for Search and Display the week of March 31, 2025. The option was removed for new campaigns in October 2024, and campaigns still on eCPC were migrated to Manual CPC in mid-March 2025, pushing advertisers fully onto Maximize Conversions / Maximize Conversion Value (with optional tCPA/tROAS) Smart Bidding.

Why it mattersThe last semi-manual bidding bridge is gone. Any B2B account leaning on Manual CPC + eCPC must commit to full Smart Bidding, which makes accurate conversion tracking and good conversion-value inputs non-negotiable rather than a nice-to-have.

Do thisAudit legacy Search/Display campaigns for eCPC or fallback Manual CPC remnants, migrate to Maximize Conversions with a tCPA (or Max Conversion Value with tROAS) tied to real pipeline value, and verify conversion tracking and value rules are solid before switching.

What changedPerformance Max gained self-serve campaign-level negative keywords (previously you had to ask Google to add them), and in March 2025 the negative keyword list limit jumped to 10,000 per campaign. Google also added full search-term reporting/insights to PMax for Search and Shopping placements, plus increased search themes from 25 to 50 per asset group and added device and age targeting controls.

Why it mattersThese close PMax's two biggest gaps for B2B: you can now exclude unprofitable, competitor, and irrelevant queries at scale, and you can see the actual search terms driving spend. It makes PMax controllable enough to run alongside a tightly managed non-brand Search program.

Do thisOn every PMax campaign, deploy a shared negative keyword list (now up to 10,000 terms), review the new search-term report monthly to add negatives, and apply device/age controls to steer away from low-quality segments.

What changedDemand Gen (which replaced Discovery campaigns) added channel-level controls letting advertisers choose exactly where ads serve across YouTube, Discover, Gmail, and Display, including YouTube Shorts-only targeting. Google also added full Google Display Network inventory and AI image/video enhancements, and Video Action Campaigns were mandatorily upgraded into Demand Gen by mid-2025.

Why it mattersDemand Gen is Google's Meta/LinkedIn-style demand-creation product. For B2B SaaS chasing pipeline above bottom-funnel Search, channel controls make it far less of a black box, and the VAC absorption means your YouTube performance video now lives inside Demand Gen's auction and reporting.

Do thisStand up a Demand Gen prospecting campaign fed by strong first-party audiences (Customer Match, site visitors) and lookalikes, use channel controls to favor YouTube/Discover and exclude weak placements, and measure it on view-through-assisted pipeline rather than last-click.

What changedAn Ahrefs analysis (reported by Search Engine Journal) found the #1 organic position suffers roughly a 34% drop in click-through rate when an AI Overview is present for the query, comparing CTR for position one with and without the Overview across a large keyword dataset.

Why it mattersEven if you win the top organic spot, the presence of an AI Overview taxes the click you used to get. For SaaS SEO this changes the ROI math on chasing #1 for informational keywords and reinforces prioritizing queries that don't trigger Overviews or that sit deeper in the funnel.

Do thisSegment your keyword targets by AI Overview presence. Deprioritize content bets on Overview-heavy informational terms; double down on intent-rich, lower-Overview-frequency commercial queries.

What changedMeta deployed Andromeda, a deep-learning ad-retrieval engine running on NVIDIA Grace Hopper superchips that sits at the first (retrieval) stage of the ad-ranking funnel. It scans a far larger pool of candidate ads per impression and uses a much larger neural model to pick which ads advance to final ranking. Meta reported roughly an 8% improvement in ad quality (recall of relevant ads) in tested segments and a meaningful lift in ads recommendations performance.

Why it mattersAndromeda is why broad targeting plus strong creative now outperforms tightly hand-built audiences: the machine surfaces the right ad to the right person at the retrieval layer, before your targeting choices even matter much. For a B2B SaaS operator, the lever shifts from audience micro-segmentation to feeding the system more (and better) creative variants and clean conversion signals.

Do thisStop over-segmenting audiences and instead give Andromeda more creative to choose from. Run broad or Advantage+ audiences with 4 to 6 distinct creative angles per ad set so the retrieval engine has real variety to optimize against.

What changedMeta launched Opportunity Score, a 0-to-100 score in Ads Manager that rates how well a campaign is set up against Meta's recommended best practices and lists specific, applicable optimizations (e.g., enable Advantage+ audience, consolidate ad sets, add creative). Advertisers can apply recommendations directly from the score panel.

Why it mattersIt's Meta's nudge engine pushing accounts toward automation and consolidation, and it surfaces concrete fixes operators can triage quickly. But many recommendations bias toward turning on Ai defaults, so it's a useful checklist that still needs human judgment, especially for B2B where Meta's 'best practice' may favor volume over lead quality.

Do thisUse Opportunity Score as a fast diagnostic, but vet each recommendation against your own quality goals before bulk-applying. Treat low scores as a prompt to investigate, not an order to enable every default.

What changedOn March 5, 2025, LinkedIn launched BrandLink, a video advertising program that lets brands run pre-roll ads alongside premium video content from a curated network of publishers and business creators, including Bloomberg, Forbes, The Wall Street Journal (Dow Jones), TED, and Yahoo Finance.

Why it mattersIt opens a brand-safe, premium video supply that pairs LinkedIn's professional targeting with trusted business content, a meaningfully different inventory type from native feed ads. For SaaS brands building category awareness with executive buyers, it's a top-of-funnel video play with credibility-by-adjacency.

Do thisIf you have quality video creative and a brand/awareness budget line, test BrandLink for executive reach, but keep it separate from lead-gen and measure with brand lift or influenced-pipeline (via Revenue Attribution), not CPL.

What changedGoogle continued consolidating around data-driven attribution (DDA) as the default conversion model and, with consent-gated traffic, leans on conversion modeling to fill non-consented gaps. Rule-based models (last-click, linear, position-based, time-decay) have been retired as primary options in Google Ads and GA4.

Why it mattersOperators can no longer hide behind clean last-click numbers; reported conversions are increasingly modeled and DDA-credited. Understanding what's measured vs. modeled is now essential to read performance and set realistic CPA targets for Smart Bidding.

Do thisReframe internal reporting around DDA plus modeled conversions and pair platform numbers with at least one independent lift signal (geo experiment or MMM). Don't benchmark against legacy last-click expectations.

What changedOpenAI removed the account requirement for ChatGPT Search, making it available to anyone with no login. The feature launched for ChatGPT Plus in October 2024, expanded to logged-in free users in December 2024, then went fully open in February 2025.

Why it mattersChatGPT became a real, friction-free answer engine that competes with Google for top-of-funnel SaaS research. Being cited in ChatGPT's web answers is now a distinct visibility channel an operator has to monitor and optimize for, separate from Google.

Do thisRun your category, brand, and 'best [tool] for [use case]' prompts in ChatGPT to see if and how you're cited. Add ChatGPT/OpenAI referrers to analytics segmentation if not already tracked.

What changedMeta removed the separate 'Advantage+ shopping/sales campaign' choice and made Advantage+ settings the default campaign setup across sales, leads, and app objectives. Instead of choosing a campaign type, advertisers now start in an Advantage+ flow and must actively toggle OFF the AI settings (audience, placements, etc.) to run a manual/original campaign.

Why it mattersThe default path is now AI-automated, which changes onboarding for every new campaign. For B2B SaaS, the leads objective inheriting Advantage+ defaults means audience controls and placements are AI-driven unless you deliberately constrain them, which can burn budget on low-intent placements if you don't manage it.

Do thisFor every new campaign, consciously decide whether to keep or disable each Advantage+ default. For B2B leads, set audience suggestions/exclusions and review placement defaults rather than accepting the AI flow blindly.

What changedMeta restructured campaign creation so the campaign-type selection screen is gone and Advantage+ automation settings (Advantage+ audience, Advantage+ placements, budget optimization) are pre-enabled inside the standard creation flow. Manual control still exists but is now opt-out rather than opt-in.

Why it mattersThis is the operational mechanics behind the 'Advantage+ default' headline. It quietly changes what an account looks like at scale: if an operator doesn't audit defaults on every build, the whole account drifts to AI-managed audiences and placements, making manual control discipline a deliberate workflow step.

Do thisBuild an internal campaign-setup checklist that forces a review of each Advantage+ toggle at creation time, so defaults are a choice, not an accident.

What changedAdvantage+ audience (AI-driven audience finding, where 'detailed targeting' becomes a suggestion rather than a hard constraint) was promoted to the default audience setting inside the new campaign flow. The original/manual audience controls remain available but are now the secondary, opt-out path.

Why it mattersThis is the targeting-layer manifestation of Meta's automation push. For B2B SaaS, Advantage+ audience treats your job titles, industries, and interests as a starting hint, not a fence, which can be powerful for prospecting but risky for niche, high-ACV targeting where audience precision matters.

Do thisFor broad-appeal B2B offers, test Advantage+ audience against manual targeting head-to-head. For narrow, high-ACV ICPs, use audience suggestions plus exclusions and watch lead quality closely before trusting the AI to define the audience.

What changedGoogle rolled out channel performance reporting in Performance Max, letting advertisers see impressions, interactions, conversions, and conversion value broken out by channel (Search, YouTube, Discover, Gmail, Display, Search partners, Maps) instead of a single aggregate. It expanded across 2025 with ROI columns, a cost toggle, segmentation, bulk downloads, and Google Ads API access in early 2026.

Why it mattersPMax's biggest objection for B2B was the black box. Channel reporting finally shows whether a 'great' PMax campaign is harvesting cheap brand Search or dumping budget into low-intent Display/Gmail, which is essential before trusting PMax with meaningful B2B budget. Note: it is visibility, not direct per-channel budget control.

Do thisPull channel-level reporting on every active PMax campaign. If spend concentrates in Display/Gmail with weak downstream pipeline, restructure toward feed/Search-leaning setups, tighten asset quality toward the channels you want, and move budget back to standard Search where appropriate.

What changedGoogle expanded brand exclusion controls so advertisers can apply brand exclusions to just Search text ads while keeping branded traffic for Shopping ads, alongside existing account-level negative keyword lists and brand lists that keep PMax off your own and competitor brand terms. This refines controls first introduced in 2024.

Why it mattersBrand exclusions stop PMax from cannibalizing cheap branded Search and double-paying for traffic your dedicated brand campaign already wins, while the format-level split lets retail/Shopping setups keep branded Shopping coverage. For B2B it protects clean attribution between brand and non-brand programs.

Do thisApply brand exclusions to your own brand on PMax (so credit and budget stay in your dedicated brand Search campaign) and add competitor brand lists where appropriate, while keeping any wanted Shopping brand coverage on if you run feeds.

What changedAccelerate, LinkedIn's AI-powered campaign type (launched in beta October 2024 on Microsoft's Azure OpenAI tech), moved to broad global availability in early 2025. It auto-builds an entire campaign in roughly five minutes, automating targeting/audience, creative recommendations, bidding, budget allocation, and structure from a single URL and a few inputs. It initially optimized for website conversions and lead generation, with objective support expanding through 2025.

Why it mattersThis is LinkedIn's Performance Max equivalent. For a B2B SaaS operator it trades manual audience/bid control for an opaque, machine-run setup. It can outperform on lead-gen efficiency but is a black box, so it demands hard conversion signal (CAPI/offline conversions) and guardrails to avoid wasting Series A budget on low-intent reach.

Do thisTest Accelerate as a parallel lead-gen line against a manually structured campaign, but only after CAPI/offline conversions are feeding clean revenue-proximate signal. Cap budget, watch lead quality (not just CPL), and keep a manual ABM campaign running as the control.

What changedLinkedIn rolled Predictive Audiences out broadly to advertisers in 2025. It uses AI to build a high-propensity audience from a brand's own data sources, conversion data, Lead Gen Form submissions, and uploaded contact lists, then finds similar professionals likely to convert, rather than relying purely on manual firmographic/job-title targeting.

Why it mattersIt's LinkedIn's answer to the death of granular manual targeting and gives operators a way to scale beyond a narrow ABM list. Quality of the seed (clean conversion data or a tight customer list) directly drives output quality, which favors operators who already have good first-party signal wired up.

Do thisFeed Predictive Audiences a high-quality seed: closed-won customer lists or hard-conversion events, not raw MQLs. Layer it on top of, not instead of, your core firmographic ABM audience and compare CPL and downstream pipeline.

What changedLinkedIn expanded Thought Leader Ads so brands can promote posts from any consenting member, including rank-and-file employees and subject-matter experts, rather than only company leadership. This widens the pool of personal-profile content that can be amplified as paid ads.

Why it mattersPerson-to-person content consistently outperforms brand-page ads on engagement and trust in B2B. Opening Thought Leader Ads to any employee gives operators a much larger library of authentic, founder/expert-led creative to scale paid behind, which is one of the few formats that still beats rising LinkedIn CPMs.

Do thisBuild a roster of consenting employees (founders, product/eng SMEs, top sellers), identify their best-performing organic posts, and promote them as Thought Leader Ads. Use this format for the cheavy-lifting top and mid-funnel creative rather than polished brand-page units.

What changedAs part of the Accelerate and AI tooling push in 2025, LinkedIn surfaced AI-assisted creative and copy generation inside the campaign-building flow, recommending and generating ad creative and headlines automatically rather than requiring fully manual asset creation.

Why it mattersFor a solo or lean operator, AI creative generation lowers the cost of producing the volume of variants LinkedIn's machine learning needs to optimize. The risk is generic, off-brand output, so it shifts the operator's job from writing every asset to curating and steering the AI's output.

Do thisUse AI creative as a first-draft variant generator to feed testing volume, but enforce brand voice and a human edit pass before anything goes live. Never ship raw AI copy that carries generic LLM tells into a B2B buyer's feed.

What changedGoogle released Meridian, an open-source Bayesian marketing mix modeling library, as its recommended MMM for measuring cross-channel impact and incrementality without user-level tracking. It supports reach/frequency inputs, calibration with incrementality experiments, and budget optimization.

Why it mattersMMM is becoming the privacy-durable answer to cross-channel measurement now that user-level attribution is degraded. Meridian gives a credible, free framework to estimate channel contribution and reallocate budget, even below classic enterprise MMM spend levels.

Do thisFor accounts with enough spend history, evaluate Meridian (or a vendor built on it) as a top-down cross-channel check against platform-reported numbers, and calibrate it with geo-lift experiments.

What changedGoogle shifted the campaign-level 'broad match keywords' setting to be enabled by default when a new Search campaign uses a Smart Bidding strategy (previously off by default). With it on, keywords are treated as broad match while still prioritized like exact match in the auction, and Google's query understanding considers landing page, other keywords, and account context. Misspelled-variation handling for negatives also improved.

Why it mattersBroad match is no longer the budget-incinerator it once was, but enabling-by-default means new campaigns can quietly run broad unless you check. For expensive, low-volume B2B conversions, blindly going broad without strong conversion signals and negatives still risks waste on adjacent intent.

Do thisOn every new Search campaign, check the broad-match campaign setting deliberately rather than accepting the default. Only run broad with a hard down-funnel conversion goal plus Smart Bidding, gate it with a strong shared negative list and brand controls, and review search terms weekly during ramp.

What changedNew data reported by Search Engine Land showed AI referral traffic (from ChatGPT, Perplexity, Gemini, and other AI tools) grew more than 500% over the prior six months, with ChatGPT as the dominant driver of AI-sourced referrals.

Why it mattersConfirms a steep, sustained ramp in AI-sourced visits across engines, signaling that GEO is becoming a real acquisition channel rather than a curiosity. For SaaS, it's a prompt to stand up measurement before competitors capture the citations.

Do thisBuild an 'AI referrers' channel grouping in your analytics (ChatGPT, Perplexity, Gemini, Copilot) and report it monthly so you can see the trend and tie GEO efforts to actual sessions and pipeline.

What changedFoundation Marketing analyzed 10,000+ AI citations and found that statistics, original research, structured/listicle content, and clear direct answers are cited most; authority and trust signals, cited credible sources, expert quotes, and fresh content increase citation likelihood; and Wikipedia and Reddit are among the most-cited sources across engines, with brand mentions mattering even without links.

Why it mattersIt gives operators a practical GEO playbook: earned third-party validation and community presence (Reddit, review sites) can matter more than on-site SEO for getting cited. Unlinked brand mentions becoming a ranking-style signal is a meaningful shift for SaaS PR and community strategy.

Do thisInvest in original data/research content, seed credible third-party mentions (Reddit threads, review sites, Wikipedia-eligible references), and structure pages with direct, quotable answers to win AI citations.

What changedMeta first publicly unveiled Andromeda as the AI system powering its advertising retrieval stack, describing it as a new machine-learning architecture purpose-built to expand the candidate ad pool and improve personalization at scale. It was framed as a foundational piece of Meta's move toward AI-run ad selection.

Why it mattersThis was the early signal that Meta was rebuilding the plumbing of its ad system around AI, not just adding AI features on top. Operators who understood this in late 2024 got ahead of the 2025 shift to broad, creative-led, automation-first campaigns.

Do thisTreat Meta's account as a creative-and-signal feeding operation rather than a manual-levers operation. Audit whether your account is still structured around legacy detailed targeting and plan a migration to broad + Advantage+ structures.

What changedLinkedIn launched the Revenue Attribution Report, which connects a company's CRM data (named integrations include Salesforce, HubSpot, and Microsoft Dynamics 365) to LinkedIn campaign data to show which campaigns, audiences, and content influence pipeline and closed revenue, not just clicks and leads. Announced at B2Believe in November 2024 and rolled out to advertisers through 2025.

Why it mattersThis is the single most important measurement change for a B2B operator: it moves LinkedIn reporting from CPL/CTR to influenced pipeline and revenue, which is the language the founder and board actually care about. It also gives a native answer to the long-standing 'LinkedIn is too expensive' objection by showing deal influence.

Do thisConnect the CRM (Salesforce/HubSpot/Dynamics) to LinkedIn and stand up the Revenue Attribution Report so reporting leads with influenced pipeline and revenue. Use it to reallocate budget toward the campaigns/audiences that touch closed-won, and to defend LinkedIn's high CPLs to leadership.

What changedLinkedIn rebuilt the Companies Hub into a dedicated account-based marketing workspace where marketers identify and prioritize high-value target accounts, see account-level engagement and buying signals, and manage ABM campaigns in one place. Announced alongside the Revenue Attribution Report in November 2024 and expanded to advertisers through 2025.

Why it mattersFor B2B SaaS the account is the unit of value, not the individual lead. A native ABM hub with engagement and intent signals lets an operator concentrate spend on in-market accounts and align paid with sales, which is exactly the motion a Series A team needs to make limited budget count.

Do thisLoad the target-account list into Companies Hub, prioritize by engagement/intent signals, and run account-targeted campaigns against the hottest tier. Share the account engagement view with sales so paid and outbound hit the same accounts.

What changedLinkedIn rolled out Connected TV advertising (announced November 14, 2024), letting B2B advertisers extend campaigns to streaming TV screens using LinkedIn's professional audience data, with inventory through partners including Paramount, Roku, and Samsung Ads. The video network continued expanding through 2025 via BrandLink and Wire.

Why it mattersCTV with B2B professional targeting is new ground: it lets a SaaS brand reach decision-makers on the living-room screen rather than only in-feed. For most Series A budgets it's an awareness experiment, but it signals LinkedIn is building a full cross-screen video stack an operator now has to factor into upper-funnel planning.

Do thisPark CTV as an awareness test for later stages or larger budgets; for most Series A SaaS, prioritize feed and Wire/BrandLink video first. If you test CTV, measure with brand lift and influenced pipeline, never direct response.

What changedThe academic paper 'GEO: Generative Engine Optimization' (Aggarwal et al., originally submitted Nov 2023, revised June 2024, published at KDD 2024) formalized GEO and introduced GEO-BENCH. It found that GEO methods (adding citations, quotations, and statistics) can boost a source's visibility in generative-engine answers by up to 40%.

Why it mattersThis is the evidentiary backbone for why AEO/GEO works and what specifically moves the needle: citing sources, adding quotes, and using statistics make content more quotable by LLMs. It gives operators concrete, research-backed levers rather than guesswork.

Do thisRewrite key SaaS pages to lead with quotable statistics, cite credible third-party sources, and include expert quotes, the exact moves shown to raise generative-engine citation rates.