Client reporting for the AI search era

The scariest sentence in agency life used to be “rankings are up but leads are down.” It now has a sequel: the client who asks “what about AI?” and gets a slide of average position in response. If your reporting can’t narrate what’s happening in AI answers, you look behind even when your work is ahead, and someone else’s pitch deck will gladly explain the gap to your client.

What changed about the story

Classic SEO reporting rode a clean causal chain: rankings up → clicks up → leads up. AI search broke the middle link. Citations influence buyers without producing sessions, so organic traffic can flatten while the brand’s actual visibility grows. The report has to carry that explanation, with numbers, or the client concludes organic is dying and starts shopping the budget.

The monthly structure we use

1. TL;DR, five sentences max. Written for the person who forwards it without reading page two. Citation rate moved from X to Y, the one notable win, the one honest concern, what we’re doing next. Deltas, never levels: “61% → 78%” is a story, “78%” is a number.
2. Organic performance, the familiar GSC/GA4 view, kept, but framed against AI context: when clicks dip on queries that gained an AI Overview, say exactly that, because it converts an alarming chart into an explained one.
3. AI visibility, the citation matrix summarized: rate by engine, wins and losses called out by query name, and who took the cells you lost (clients remember competitor names long after they forget percentages).
4. Work completed, tied to outcomes. Not “published 4 articles” but “rebuilt the pricing page; Perplexity began citing it 11 days later.” Every line of work earns its place by pointing at a cell in the matrix.
5. Next month’s plan with expected impact, three to five items, each naming the queries it targets. This is the section that makes the report a contract instead of a rearview mirror.

Write it like a person

Our rule for report prose: grounded in the data, written like a colleague, no adjective without a number behind it. “Strong month for AI visibility” is filler; “ChatGPT now cites you on 4 of 5 buy-intent queries, up from 1 in March” is a report. We draft ours with Claude against the actual portal data, each report fed the prior two so the narrative stays continuous, then a human strategist edits before it ships. The drafting is automation; the judgment isn’t. Clients can tell which one wrote the sentence that matters.

The artifact that renews retainers

Quarterly, add one chart: citation share vs. the top three competitors over time, your share of all citations logged across the priority portfolio. It’s the AI-era equivalent of share of voice, it’s the chart CMOs screenshot for their CEOs, and in our experience it correlates with renewals better than any traffic graph ever did. Build it from the competitor data you’re already logging (method here), and let the trend line make the case.