Writing an SEO Report a Client Will Actually Read

Published 2026-08-31 · 7 min read · by the isaeo team

Why reports go unread

The typical SEO report is organised around what the tool exports rather than what the reader needs to decide. It opens with rankings and traffic charts, continues through crawl statistics, and buries the two sentences the client actually wanted on page 30.

The client's questions are always the same three: is it working, what did you do, what's next. A report that answers those in the first half-page gets read. Everything else is supporting evidence that exists to be checked, not consumed.

The structure that works

1. What changed. Scores now versus last period, with direction and magnitude. Two or three sentences. If nothing changed, say so — that's information too.
2. What it means. Translate. "AEO score rose from 41 to 58" means nothing; "your service pages are now structured so AI assistants can quote them, where before they weren't" means something.
3. What we did. Specific, past tense, verifiable. Named pages and named fixes.
4. What's next. Three items maximum, in priority order, each with an owner. If something needs the client's developer, say that explicitly here rather than in a footnote.
5. Appendix. Per-page tables, full issue lists, the crawl detail. Nobody reads it; its presence is what makes the summary credible.

How to present scores

A single number with no context is a vanity metric. Three rules make scores useful:

• Always show the delta. 72 tells the client nothing. 72, up from 64, tells them you're working.
• Split by discipline. One score hides which problem you're solving. See the five disciplines — technical at 90 and content at 40 is an actionable picture.
• Never present a score you can't explain. If the client asks why it's 72 and the honest answer is "the tool said so", don't include it.

Explain the scale once, in the report, every time. Clients don't remember whether 70 is good, and asking them to is a small insult.

What to leave out

1. Vanity metrics. Impressions with no click context, keyword counts, "pages crawled" as an achievement.
2. Every low-severity issue. A list of 400 missing alt attributes makes the real problems invisible. Summarise the count, appendix the detail.
3. Jargon you'd have to define. If "canonicalisation" needs a footnote, write "duplicate page addresses" instead.
4. Competitor speculation. Unless you have data, "competitors are likely investing in content" is filler.
5. Anything you can't act on. A finding with no recommendation is an anxiety generator.

Cadence

Monthly suits most retainers: long enough for changes to register, short enough to stay relevant. Weekly reporting on SEO is mostly noise — the signal doesn't move that fast, and you'll spend the retainer writing reports.

Whatever you pick, keep the format identical every month. The value of a recurring report compounds only if it's comparable to the last one, which is why the what-changed section is the most valuable page in it. Automating the data collection is what makes this profitable — see the monthly workflow and white-label delivery.