Which AI Search Agencies Measure Citations per Engine for Clients? (2026)

46 AI search agencies checked for per-engine citation measurement: the 2026 table

None found means not found on the pages we checked. AI Overviews means Google AI Overviews. Tier 1A shows one client across two or more engines side by side; Tier 1B shows a client number for one engine. Every reading comes from the agency's own site on 17 October 2026.

How did we test 46 agency sites?

We asked every agency one question: on its own website, read today, does it show that it measures AI citations or mentions for clients, per engine?

The roster came from the engines. We took every agency that ChatGPT, Perplexity or Google AI Overviews named two or more times when asked which agencies measure citations for clients, then added the AI search agencies most often ranked in B2B SaaS lists, such as First Page Sage, Siege Media, NoGood and Directive. Adatek and BAMS Digital were skipped because we could not confirm their domains. Tools sit outside the roster. Wellows calls itself a platform, but ChatGPT names it as an agency here, so it stays with a label.

For each agency we read the home page, the AI search or GEO service page, any measurement or methodology page and up to two AI visibility case studies, about six pages in all. Then we sorted:

1. Tier 1: at least one client number tied to a named engine (1A: one client across two or more engines; 1B: one engine per client).
2. Tier 2: names two or more engines plus a tool or metric for client reporting, but no per-engine client number. A sample dashboard, an illustrative table, the agency's own brand or a blended figure counts as no.
3. Tier 3: describes measuring AI visibility but names fewer than two engines, or no tool or metric.
4. Tier 4: no measurement statement found.

Tier 1 is a low bar on purpose: one card with one engine clears it. This test differs from our pipeline reporting audit, which asked who names a client whose leads came from AI search, and from our 90-day citation study, which counted who the engines cite.

What do the 15 Tier 1 agencies publish per engine?

LoudFace

The GEO agency page describes "A live Peec panel across ChatGPT, Perplexity, and Google AI Overviews, reported per engine every week." The Toku case study gives Toku's visibility per engine for a 30-day spring 2026 window: 35% on Google AI Overviews, 11% on ChatGPT and 10% on Perplexity. In the 30 days to 19 August 2026, Toku's visibility read 39.3% on Google AI Overviews, the second most-visible brand on that engine. The methodology page says "Share of answers and citations are tracked per engine, then read against search demand, signups, booked demos and captured leads."

Cintra

Its reporting guide shows a wide engine gap for one client: "We had a client with 40% citation share in ChatGPT and 8% in Perplexity for the same prompts." The guide says "We build these reports for clients every month", with results in "a leads dashboard updated in real time". The client is anonymised.

NoGood

The SteelSeries case study gives before and after scores per engine: "Gemini Visibility Score: 62 → 85 (+37%)", "ChatGPT Visibility Score: 51 → 73 (+43%)" and "Perplexity Visibility Score: 44 → 77 (+75%)". Its AEO page says it tracks "continuously, not just at quarterly check-ins." The client is named, so you can ask about it on a reference call.

Single Grain

The LS Building Products case study splits a 12-month result by engine, "specifically 174 more mentions on AI Overviews and 113 on ChatGPT". Tracking runs on "custom dashboards".

Thrive Internet Marketing

Client cards read "+876% Engaged Sessions From ChatGPT +749% Microsoft Copilot and Partner Citations" for a shoe company and "+2,658% Google AI Overview Citations" for a medical group. The Gemini and ChatGPT lifts elsewhere on the page are Thrive's own site. Its AI Overviews page promises "live metrics and detailed monthly reports".

Breaking B2B

The AEO page names one result: "We took RB2B to #1 in both Google and ChatGPT for 'Leadfeeder alternatives'". One query and one named client, with ChatGPT the AI engine in it.

Cubitrek

"We measure citations across ChatGPT, Perplexity, Claude, Gemini, Bing Copilot, and Google AI Overviews every week." The Keyper case study adds "86% of citations came from ChatGPT, which is a planning fact and a concentration risk." That is the right way to read a lopsided split.

Digital Authority Partners

The case-study index shows "47% AI Overviews Citing Asimily", and a home page card reads "#1 in ChatGPT for ABA in 8 of 9 Chicago neighborhoods". The GEO page names a tracking step but no tool.

Discovered Labs

A client quote on the tracker page: "One month in, our referrals from ChatGPT are up 29%." Its tracker is described as "the instrument behind a service".

iPullRank

Two anonymised clients, one engine each. A vehicle-sales platform's card reads "661% growth in ChatGPT Visibility". In AI Overviews, a telecom brand went "From 712 to 3,235 AI Overview inclusions YoY".

Kern Media

The only per-engine revenue figure in Tier 1: "revenue per ChatGPT visit rose from $1.05 to $6.33" for an anonymised eCommerce brand. Each AI tool is "reported separately."

Percepture

"We track AI mentions, referral traffic, and sentiment monthly." The engine result reads "Another telecom company saw a 3000% boost in Google AI Overviews mentions within three months", phrased after a client example, so client status is implied.

Rampiq

The Avassa case study reports "2900% AI Overview visibility growth for an edge computing SaaS".

Siege Media

Zapier gets "47.2K citations in AI Overviews, and 96% visibility in LLMs for industry-specific queries". The first half is per engine; the second is blended. Its GEO page lists "LLM share of voice in your category prompts, tracked monthly".

Wellows

A self-described platform that ChatGPT names as an agency. It "runs your tracked prompts across all five platforms daily". The Articos guide "entered the same ChatGPT answer twice, at positions 2 and 4."

Tier 2: engines and metrics named, no per-engine client number found

Thirty-one agencies describe a real measurement setup and stop short of a client number per engine. A Tier 2 page is a fine reason to book a call. Ask for a redacted client report before you sign.

Per-engine numbers for the agency itself. Verticality writes that "Since 20 September 2026 ChatGPT has named Verticality in answers to 8 of the 56 AI-search buying questions we track". DoodleWeb "recorded 246 visits from ChatGPT over 84 days in GA4" for its own site. CitationLab, Clear Cited and TheProjectSEO show their own numbers too, and Revlift says it plainly: "These are first-party results on products we operate, not client case studies." These prove the method runs. They don't show a client result.

Labelled samples. Search Agency's per-engine panel is headed "AI_VISIBILITY / DASHBOARD EXAMPLE". Point Visible labels the AI-citation section of its sample report "SAMPLE DATA". Tenten GEO's growth engine page reads "Figures are illustrative; your weekly dashboard is the source of truth". The labels are honest, and each one previews the report format.

Blended client results. Over The Top SEO reports an "AI citation rate: 87% (up from 12%)." Graphite: "94% share of voice growth from LLMs (increase from 33% to 64%)". Omniscient Digital reports that "OnBoard appeared in 88% of category prompts". Optimist: "8x growth in conversions from LLM referrals within 8 months (ChatGPT, Perplexity, Claude)". First Page Sage tracked a client's "share of recommendations across ChatGPT, Claude, and Google's AI Overviews using Profound", reported as one figure that "grew from 8% at kickoff to 61% by the end of Year 2". RevvGrowth and Broadcastwell also report combined figures. A blend can hide one engine falling while another climbs.

Method stated, no client figure found. 1Digital Agency is the most specific: "Three numbers roll up weekly per engine", with "server-log evidence of AI crawler activity". Traffiy says "The monthly report shows citation counts for each engine separately". The Enough Agency reports engines "separately because each engine cites and recommends differently." Winston Digital Marketing, Momentence, Powered by Search, Grailstar, Animalz, Skale, RevenueZen, 5WPR and Go Fish Digital name engines and metrics. Directive names Scrunch, and Crackle PR's "recusal policy keeps clients out of scored positions in the Index."

Sharply Labs states a method on its GEO page: "Use a defined question set to record visible mentions and citations where a product shows them." The same page lists six "Surfaces to inspect for relevant buyer questions", from ChatGPT to Microsoft Copilot, and its named tool is "Search Console's available AI reporting". No agency landed in Tier 3 or Tier 4.

What do mention, citation and share of voice numbers mean?

Momentence names the problem: "One vendor reports mentions, another reports citations, a third reports a proprietary visibility score with no published formula."

Start with two definitions from the LoudFace methodology page: "A citation is the engine listing your URL as a source." and "A mention is the engine naming your brand in the answer text a buyer reads." You can be mentioned without a link, or cited while the answer recommends a rival.

Search Agency publishes the ratios. Mention rate: "Prompts where the answer names you, divided by prompts tracked." Citation share: "Answers that cite your domain, divided by answers carrying any citation, scored against named competitors." Share of voice: "Your mentions divided by every brand mention in the category, yours and your competitors together." 1Digital Agency uses "citation share (prompts with a citation ÷ total prompts)".

Visibility, in Peec's sense, is how often a brand is mentioned across tracked responses. Share of voice reads lower, because it divides your mentions by everyone's. Semrush's AI Visibility Score is relative to "the median number of mentions for your top industry competitors."

So when an agency shows a percentage, ask which one it is and over how many prompts. A 40% citation share and a 40% visibility are different claims.

What do Peec, Profound, Otterly, Semrush and Ahrefs measure?

Several agencies in the table name a third-party tool. In the tools' own words:

• Peec AI: "Peec AI runs your prompts across AI platforms like ChatGPT, Gemini, and Copilot daily". "The basis for all these metrics is Sources".
• Profound: "Citation Analytics Sources cited across answers" and "Visibility Score by Platform".
• Otterly.AI: "Analyze how AI search engines mention, and cite your brand".
• Semrush AI Visibility Toolkit: "Prompt tracking: updated daily." and "Brand Performance reports: updated weekly."
• Ahrefs Brand Radar: "Each prompt is checked daily on every selected platform."

LoudFace runs its panel on Peec. Percepture lists Semrush, Search Console, BrandLight and Profound, and First Page Sage credits Profound in a case study. Directive names Scrunch. Over The Top SEO writes "We recommend Profound or Otterly.AI for dedicated GEO monitoring". Discovered Labs draws the line: "Profound, Peec.ai, and Otterly are SaaS dashboards you operate yourself."

A tool samples prompts on a schedule. The agency chooses the prompts, reads each engine, and ties movement to the business. Some describe their own evidence layer: 1Digital Agency reads server logs, Cubitrek records every citation by source URL, and Kern Media and DoodleWeb track clicks in GA4. If the report is a tool export with a logo on it, buy the tool. For tying the numbers to revenue, see how to measure AEO agency ROI.

What does Google Search Console show for AI Overviews?

Google says "sites appearing in AI features (such as AI Overviews and AI Mode) are included in the overall search traffic in Search Console". A June 2026 Search Central post added dedicated generative AI performance reports: "As of August 31, 2026, we've rolled out these insights to all websites worldwide."

The metric is impressions: "How often URLs from your site appeared in generative AI features in Search and Discover." Google mentions "adding additional metrics over time". Search Console tells you Google's AI features showed your URLs. It says nothing about ChatGPT or Perplexity, or which rival the answer recommended.

How should you read an agency's citation report?

A per-engine report should let you rebuild every number from one row. Ask for this layout:

Then ask:

1. Is each engine reported separately? A blended score hides the engine you are losing.
2. Is the prompt set fixed? Momentence, RevvGrowth and Powered by Search describe fixed sets. Shifting prompts make trends meaningless.
3. Mention or citation? Ask for the formula, as Search Agency and 1Digital Agency publish it.
4. Is the example a client? Own-brand numbers and sample dashboards preview the format. Ask for a redacted client report.
5. What does it connect to? Kern Media's revenue per ChatGPT visit and LoudFace's demos and leads are two answers.

The full pre-hire checklist is in how to verify an AEO agency's results before you hire one.

What can't this check tell you?

We read about six pages per agency. A Tier 2 agency may publish a per-engine client number on a case study we didn't open, so every absence here means not found on the pages we checked. If yours does, send us the page.

Everything is self-reported, as served on 17 October 2026. We did not audit the numbers or check archived copies, anonymised clients can't be verified from outside, and several Tier 1 entries rest on one card or one quote. Tier 1 also says nothing about price, which our AI search agency pricing table covers.

Measurement is not results. An agency can report citation share per engine every week and still show flat lines. The table tells you who will show you the number. Judging the number is on you.