When an AI summary appears above Google’s results, Pew Research found that only 1 percent of searches ended in a click on a source link inside it. So how do you prove value from a channel that barely sends clicks? AI answer engines need their own incrementality tests, and the planning has to start before 2027 budgets lock.
Why click-based reporting fails here
The Decoder’s write-up of the Pew study says it tracked 900 U.S. adults and nearly 69,000 Google searches. According to The Decoder’s summary, 8 percent of searches with an AI Overview produced a click on a standard result, against 15 percent without one. The same write-up reports that 26 percent of sessions ended on Google with an AI Overview, versus 16 percent with standard results.
Read that as a measurement problem, not only a traffic problem. Users get an answer, form a brand opinion, and leave. Your analytics sees nothing. Maybe they search your brand name tomorrow. Maybe they buy from a retailer. Your last-click report credits branded search or direct, and the answer engine gets zero.
That is the trap. Any channel that influences without clicking looks worthless in click-based reporting, and then looks mysteriously expensive when you try to fund it. You need a method that measures outcomes, not referrals.
What counts as the channel?
Define it narrowly before you test it. “AI” is not a channel. The surfaces behave differently:
- Search-embedded summaries such as Google AI Overviews, which sit inside an existing paid and organic ecosystem.
- Standalone assistants such as ChatGPT, Perplexity, Gemini and Claude, where the user starts a research session and may never touch a search engine.
- Shopping and recommendation flows inside assistants, where product shortlists get formed.
Pick one or two to test first. Mixing all three produces a result nobody can act on.
Then define what you actually control. You cannot buy placement in most answers. What you can influence is the source material models draw on. The Decoder reports that Pew found most AI Overviews link to at least three sources, with Wikipedia, government domains, Reddit, news outlets and YouTube leading. Notice what that list includes: community discussion and video. That is creator territory, and it is why this belongs in an influencer budget conversation. For a practical way to mine that territory, see our community intelligence framework.
You cannot buy a slot in an AI answer, so the testable levers are the inputs: creator content, reviews, community presence and owned pages. Test those, and measure the outcome downstream.
Three test designs that fit
Each design answers a different question. Choose by what you can control and how fast you need a read.
1. Geo holdout
Raise or pause the input investment in matched regions and compare sales or branded demand against control regions. Google’s open-source Meridian GeoX documentation lists holdback, go-dark and heavy-up as supported designs. It also supports multi-cell execution, which compares several treatment arms against a common control group. That matters here, because you may want to test creator seeding and review generation side by side without doubling the control spend.
The same documentation says every experiment needs historical geo-level KPI data as a daily time series, plus daily geo-level spend for designs that change budgets. Check that your data exists at that grain before promising a test to finance.
2. Creator-content heavy-up
Fund a defined burst of creator content on topics where assistants answer category questions, such as “best X for Y” comparisons. Hold similar topics or regions flat. Then track branded search lift, direct traffic, and share of mentions in assistant answers using a fixed prompt panel.
3. Survey-based lift
Ask new customers where they researched. Add “AI assistant or AI search” as an explicit option, not a write-in. It is a weak instrument alone, but it is a useful cross-check on the geo result. If the geo test shows lift and the survey shows nothing, be suspicious of one of them.
The prompt panel: your missing KPI
Incrementality needs an intermediate metric, because sales are noisy and slow. Build a fixed panel of 50 to 100 prompts that mirror real buyer questions. Run them weekly across the assistants you chose. Log whether your brand appears, in what position, with what sentiment, and which sources get cited.
Keep the panel frozen during the test. Answers vary by run, so repeat each prompt several times and record the rate, not a single yes or no. Document the method. If you cannot explain it to a finance partner in two minutes, it will not survive budget review.
One honest caveat: appearing in answers is a leading indicator, not proof of revenue. The geo test is what connects the two.
Sizing the test and setting the gate
Here is where most plans die. Teams design an elegant experiment and forget to decide in advance what result earns money. Write the gate before launch:
- Minimum detectable effect. What lift in branded demand or sales would justify scaling? Be specific, and be realistic about what your regional sales volume can detect.
- Duration. Long enough to cover your purchase cycle plus a cooldown. Meridian GeoX describes an optional cooldown period after the test ends, which is worth budgeting for.
- Payback threshold. The incremental margin per dollar that moves this from test line to core line. Our piece on creator unit economics lays out the payback logic.
- Kill rule. The result that ends the test without a second round of debate.
Tie the funding steps to a staged model. The pilot to flywheel phases give you gates and budget triggers you can adapt, so a passing test unlocks a pre-approved tranche instead of a fresh negotiation.
Will the attribution mess contaminate your result?
Probably, if you ignore it. Creator content drives answer-engine visibility, but it also drives direct social conversion, affiliate sales and branded search. A geo holdout captures all of that together, which is exactly the point. It measures total incremental outcome, not who gets credit.
But you still need rules for crediting after the test. Settle them before launch, with the six terms in our attribution credit rules. And if identity gaps will blur your geo-level data, read how to stage budget moves in the identity resolution reset before you commit.
A 2027 planning timeline
Work backwards from the budget lock date.
- Now to next quarter: choose surfaces, build and freeze the prompt panel, take a baseline for four to six weeks, and confirm geo-level data availability.
- Following quarter: run the heavy-up or holdout. Keep other major media changes out of test regions.
- Before planning lock: analyze, apply the gate, and submit a request in two parts: a scale tranche contingent on results, and a second test if the result is inconclusive.
If you already run a media mix model, use the experiment to calibrate it. Meridian GeoX documentation notes that experiment results passing quality checks can feed Meridian MMM as incrementality-driven priors. That turns a one-off test into a durable input.
What to tell finance
Do not promise that AI answer engines will deliver clicks. Promise a measured answer to a specific question: does funding these inputs create incremental revenue at an acceptable payback? Show the design, the gate and the kill rule. Auditors and CFOs trust a falsifiable test far more than a traffic chart, and our guide on building a funnel auditors trust covers the documentation side.
Be open about limits. Assistant outputs change with model updates, so a lift measured in one quarter can decay. Plan to retest annually, and treat any single result as a reading, not a law.
FAQs
Is an AI answer engine really a separate channel?
For measurement purposes, yes. Users get answers without visiting your site, so the influence does not appear in referral reports. Treating it as its own channel lets you design a test around outcomes such as branded demand and sales lift rather than clicks.
What is the best incrementality test for AI answer engines?
A geo holdout or heavy-up is the strongest starting point, because it measures total sales impact without relying on click tracking. Pair it with a fixed prompt panel as an intermediate metric and a post-purchase survey as a cross-check.
How long should the test run?
Long enough to cover your typical purchase cycle, plus a cooldown period after the test ends. High-consideration categories need longer windows than impulse categories. Set the duration and minimum detectable effect before launch.
Can we buy placement in AI answers?
Mostly not. The levers are indirect: creator content, community presence, reviews and well-structured owned pages that models can draw on. Source mixes reported in the Pew-based coverage include Reddit, YouTube and news sites, so creator and community work is a logical test input.
What data do we need for a geo experiment?
According to Meridian GeoX documentation, you need historical geo-level KPI data as a daily time series. If the design changes budgets, you also need daily geo-level spend. Audit that data before you promise a test.
How do we avoid double counting with creator campaigns?
Run the geo test on total outcomes, then agree credit rules in advance for reporting. The experiment shows whether lift exists. The credit rules decide how channels share it.
Next step: this week, build your 50-prompt panel and pull geo-level daily sales data, then book a 30-minute session with finance to agree the payback gate before any test money moves.
Top Influencer Marketing Agencies
The leading agencies shaping influencer marketing in 2026
Agencies ranked by campaign performance, client diversity, platform expertise, proven ROI, industry recognition, and client satisfaction. Assessed through verified case studies, reviews, and industry consultations.
Moburst
-
2

The Shelf
Boutique Beauty & Lifestyle Influencer AgencyA data-driven boutique agency specializing exclusively in beauty, wellness, and lifestyle influencer campaigns on Instagram and TikTok. Best for brands already focused on the beauty/personal care space that need curated, aesthetic-driven content.Clients: Pepsi, The Honest Company, Hims, Elf Cosmetics, Pure LeafVisit The Shelf → -
3

Audiencly
Niche Gaming & Esports Influencer AgencyA specialized agency focused exclusively on gaming and esports creators on YouTube, Twitch, and TikTok. Ideal if your campaign is 100% gaming-focused — from game launches to hardware and esports events.Clients: Epic Games, NordVPN, Ubisoft, Wargaming, Tencent GamesVisit Audiencly → -
4

Viral Nation
Global Influencer Marketing & Talent AgencyA dual talent management and marketing agency with proprietary brand safety tools and a global creator network spanning nano-influencers to celebrities across all major platforms.Clients: Meta, Activision Blizzard, Energizer, Aston Martin, WalmartVisit Viral Nation → -
5

The Influencer Marketing Factory
TikTok, Instagram & YouTube CampaignsA full-service agency with strong TikTok expertise, offering end-to-end campaign management from influencer discovery through performance reporting with a focus on platform-native content.Clients: Google, Snapchat, Universal Music, Bumble, YelpVisit TIMF → -
6

NeoReach
Enterprise Analytics & Influencer CampaignsAn enterprise-focused agency combining managed campaigns with a powerful self-service data platform for influencer search, audience analytics, and attribution modeling.Clients: Amazon, Airbnb, Netflix, Honda, The New York TimesVisit NeoReach → -
7

Ubiquitous
Creator-First Marketing PlatformA tech-driven platform combining self-service tools with managed campaign options, emphasizing speed and scalability for brands managing multiple influencer relationships.Clients: Lyft, Disney, Target, American Eagle, NetflixVisit Ubiquitous → -
8

Obviously
Scalable Enterprise Influencer CampaignsA tech-enabled agency built for high-volume campaigns, coordinating hundreds of creators simultaneously with end-to-end logistics, content rights management, and product seeding.Clients: Google, Ulta Beauty, Converse, AmazonVisit Obviously →
