GEO reporting for agency clients: what to include

GEO is a new discipline and most clients do not yet have a mental model for it. Effective GEO reporting starts with education, builds through clear metrics, and closes with proof of impact. Here is a framework that works.

What should a GEO report include?

A GEO report needs six sections, in this order: business context, current GEO score, what changed this period, real citation examples, AI referral traffic, and next actions. The order matters more than the contents, because most clients have no mental model for AI search yet and will misread the metrics if you lead with them.

The single most persuasive item on that list is section four, the citation examples. A screenshot of a competitor being recommended by ChatGPT when the client is not does more work than any score.

Key Takeaway

Effective GEO reporting for clients requires a six-section structure: business context, current GEO Score, changes since last period, real citation examples, AI referral traffic data, and specific next actions - because most clients need education before they can interpret raw metrics.

Why GEO reporting is different from SEO reporting

SEO reporting is familiar territory - rankings, traffic, conversions. Clients understand the metrics and the story they tell. GEO reporting is new territory for most clients, which means your report has to do double duty: educate and demonstrate value simultaneously.

The temptation is to lead with technical detail. Resist it. Lead with the business context.

Report structure: the six sections

1. Why this matters (context slide)

Open every GEO report with a brief reminder of why AI search visibility matters for this client specifically. One or two statistics about AI search adoption. One sentence about what being cited (or not cited) means for their business. Keep it brief - one slide or half a page.

2. Current GEO score

Show the overall score and the category breakdown as a visual. The score should be the centrepiece of the report. Use colour coding - red for low-scoring categories, green for strong ones. Make the score trend visible: current vs last period vs baseline.

3. What changed this period

Specific actions taken and their impact on signal scores. “We implemented FAQPage schema on 12 pages - Structured Data category score improved from 34 to 67.” Keep this concrete and causal.

4. Citation examples

This is the most powerful section. Screenshots of the client’s website appearing as a cited source in ChatGPT, Perplexity or AI Overviews. For clients who are not yet appearing - screenshots of competitors appearing, to illustrate the opportunity. One or two examples are enough.

5. AI referral traffic

Even if the numbers are small, show the trend. A chart showing AI referral traffic growing month-on-month is compelling evidence that GEO work is translating into business impact.

6. Next actions

Three to five specific actions planned for the coming period, with expected impact. This gives the client confidence that momentum is building and sets expectations for the next report.

How to actually measure AI referral traffic

Section five is where most agencies wave their hands, because AI referrals are not a channel in Google Analytics by default. They arrive as ordinary referral traffic and get scattered across a dozen hostnames. You have to group them yourself.

In GA4, create a segment or a comparison filtered on session source containing any of these hosts:

- chatgpt.com and chat.openai.com - ChatGPT - perplexity.ai - Perplexity - claude.ai - Claude - gemini.google.com - Gemini - copilot.microsoft.com and bing.com/chat - Copilot - grok.com and x.com/i/grok - Grok

Three warnings worth passing to the client before they see the numbers.

The volume will look tiny. AI referral traffic is a small fraction of organic for almost every site. Report it as a trend line, not a total, and never put it next to organic sessions on the same axis.

Much of the impact is invisible. When an engine names a client in an answer and the user does not click, no referral is recorded at all. The visit may arrive weeks later as direct or branded search. This is the honest limitation of referral data and you are better off saying it than being caught by it.

Attribution is inconsistent. Some engines strip the referrer, some pass it, and the behaviour changes without notice. Treat month-on-month movement of a few percent as noise.

This is exactly why the report needs both citation data and referral data. Citations show whether you are being recommended. Referrals show the fraction of that which converted into a visit you can see. Reporting only one of the two gives the client a distorted picture. There is more detail in how to measure AI search visibility.

What to leave out

Padding a GEO report is tempting when the channel is young and the numbers are small. These four items cost you credibility.

Branded citation rates. If you ask an engine “what is Acme Ltd” and it describes Acme Ltd, you have measured nothing. Almost every brand scores well on branded questions. Report the non-branded rate, meaning how often the client is named when the question describes the problem rather than the company, and say explicitly which one you are showing. A blended figure that hides a zero non-branded rate is the most common way GEO reports mislead.

Single-check results. AI answers vary between sessions for the same question. One reading is an anecdote. Report a trend across at least four weekly checks before drawing any conclusion.

Score movement without attribution. “Your GEO score rose from 48 to 55” invites the obvious question. Pair every movement with the specific change that caused it, or leave it out.

Borrowed industry statistics. Third-party projections about AI search adoption age badly and are frequently unsourced. One or two are useful framing in section one. Any more and the report reads as a pitch. Client-specific data always beats a market forecast.

Handling the “is this real?” question

Some clients will question whether AI search is a significant enough channel to invest in. Three responses work well:

- Demonstration - open ChatGPT in the meeting and ask their business’s key questions. Show them the results live.

- Growth data - share adoption statistics for ChatGPT and Perplexity. Hundreds of millions of monthly active users searching for products and services is a concrete business reality.

- First-mover framing - “Your competitors haven’t done this yet. The question isn’t whether to do it, it’s whether to do it before or after them.”

Where the data comes from

You need three inputs, and they are not all available from the same place.

Readiness data is a scan of the client’s own site: whether it is crawlable, structured and quotable. A free SearchScore audit covers this across 250+ signals and 23 categories, returns in about 60 seconds, and needs no account. This is what feeds sections two and three of the report.

Citation data requires putting real questions to the engines, which no site scan can do. SearchScore Tracker runs a fixed question set against six engines, ChatGPT, Claude, Gemini, Perplexity, Grok and DeepSeek, and reports who was cited and who was named instead. Starter is £20 a month and Pro is £79. This feeds section four, and the competitor set it returns is usually the most useful slide in the deck.

Referral data comes from GA4, filtered as described above, and feeds section five.

Worth being precise when a client asks: a site scan measures readiness, meaning whether the site is built to be quoted. Only querying the engines measures citation, meaning whether they actually quote it. Any tool that claims to report AI visibility without querying engines is reporting the former. Saying so plainly tends to build trust rather than cost you the sale.

Cadence

Monthly is right for the report itself. Weekly is right for the citation checks that feed it, because AI answers move faster than rankings and a monthly snapshot cannot distinguish a real decline from ordinary variation. Full technical re-audits are quarterly, since readiness only changes when the site changes.

For clients not on a GEO retainer, a one-page GEO summary inside the standard monthly digital report is worth including anyway. It is the cheapest way to make the case for the retainer, and it is a better conversation than a cold pitch.

Back to pillar

- How to Measure and Track Your GEO Performance → - How to measure AI search visibility → - AI rank checker: check where you appear in AI answers →

Ronnie Huss

GEO Research & Analysis

The SearchScore editorial team researches and writes about generative engine optimisation, AI search visibility and the signals that determine whether your website gets cited by ChatGPT, Perplexity and Google AI Overviews.

Sources & Further Reading

- SearchScore – Scoring methodology (250+ signals)

- SearchScore – The State of AI Visibility Index (SAVI)

- Google – Search Console overview

Frequently asked questions

What should a GEO report include?

A GEO report should include: context on AI search and why it matters, the current GEO audit score with category breakdown, progress since last report, citation examples (screenshots of the client appearing in AI answers), AI referral traffic trend, and next actions planned.

How do you justify GEO work to a sceptical client?

The most effective approach is demonstration. Show the client their website being asked about in ChatGPT or Perplexity - and whether they appear. If competitors appear and they do not, the conversation writes itself. Supplement with data on AI search growth and user behaviour shifts.

How often should agencies deliver GEO reports?

Monthly reporting including citation monitoring and AI traffic is appropriate for active GEO clients. Full audit reports are typically quarterly. Include a one-page GEO summary in all monthly digital marketing reports as standard, even for clients not on a dedicated GEO retainer.

How do I track AI referral traffic in Google Analytics?

GA4 has no AI channel by default, so AI referrals land in ordinary referral traffic. Build a segment filtered on session source containing chatgpt.com, chat.openai.com, perplexity.ai, claude.ai, gemini.google.com, copilot.microsoft.com and grok.com. Expect small numbers, and expect them to understate the real impact, because a user who reads an answer without clicking generates no referral at all.

What is a good GEO score to report?

Context matters more than the absolute number. Across 1,000,000+ sites scored by SearchScore, only 217 reach AI-Ready, a GEO score of 80 or above. That is 0.022%, roughly 1 in 4,600, and the highest score recorded anywhere is 86.6. So a client at 55 is ahead of most of their market, and framing the score against that distribution is far more useful to them than framing it against 100.

Should GEO reporting replace SEO reporting?

No. They measure different channels and the two rarely move together. A page can rank first on Google and never be cited by an AI engine on the same topic. Report both, keep the metrics separate, and resist blending them into a single visibility figure, which obscures more than it reveals.

What if the client has no citations at all?

That is the normal starting point and it is reportable. Show which competitors were named instead, and for which questions. A zero with a named competitor set is an actionable finding. A zero with no context looks like a failure, which is the same data told badly.

Check your AI visibility

Enter your URL at SearchScore for a free AI visibility Score. See how ChatGPT, Perplexity and Google AI see your site - and exactly what to fix.