An SEO report is the regular document that tells you what happened to a site's organic visibility over a period, why it happened and what will be done next. It pulls together data from Google Search Console, Google Analytics 4 and a rank tracker, but the value isn't in the data: it's in the three lines of commentary that tie the data to a decision.
Most guides on the subject list the metrics to put in. We do something else here: we show the one-page structure we use, fill it in with a client's real figures (all of them already public), and point out the three things a report has to declare in 2026 that almost nobody declares.
Report, audit and dashboard: three documents, three questions
The first misunderstanding starts on the results page itself. Search "seo report" on Google UK and most of page one is tools. On 10 October 2026 the AI Overview at the top pointed to Rank Math's SEO Analyzer, Semrush and SEOptimer, and the organic results included Seobility's website audit and Brickweb, a UK web agency that promises a website SEO audit report "in minutes". What these tools produce is an automated audit, a technical snapshot of the page. Useful, but it answers a different question.
| Document | The question it answers | How often | Who writes it |
|---|---|---|---|
| SEO audit | What's broken or could be better on the site? | One-off, or twice a year | A consultant, with the help of a crawler |
| SEO dashboard | Is something happening right now? | Always on | Nobody: it updates itself |
| SEO report | What did we do, what changed, what will we do next? | Monthly, with a quarterly review | A named person |
If you need the first, we've written a guide to the SEO audit. If you need the second, the SEO dashboard guide explains why numbers from different sources never match. This article is about the third, the only one you can't generate with a click, because the part that matters is written by someone.
The structure: one page, five blocks
A report that works fits on one page. The reason is practical: nobody reads a 24-page PDF exported automatically every month, and in the meeting it gets replaced by a sentence said out loud. Better to write that sentence in the first place.
| Block | What it contains | Length |
|---|---|---|
| 1. The summary | The most important fact of the period and the decision that follows from it | Two or three sentences |
| 2. The numbers that matter | Organic clicks, organic conversions, impressions, against the same period last year | A table of 4-6 rows |
| 3. What we did | The month's work, with dates, linked to the numbers it should move | A list of 3-8 items |
| 4. What we can't measure | This month's data limits, written down before the client asks | Two or three lines |
| 5. What we'll do next | Next month's priorities, in order | A list of 3-5 items |
Block 4 is the one you won't find in the editorial guides on Google UK's first page for "seo report" (Ahrefs' report template and SEOJuice's guide, both checked on 10 October 2026), and in our view it's what separates a credible report from one that reads as if it were written in self-defence. We come back to it below, because in 2026 that section has got longer.
Block 3, on the other hand, is the one people forget most often. Without a line saying "on 12 March we rewrote the title tags in category X", a chart that goes up doesn't explain why it goes up, and one that goes down turns into an argument.
An example filled in with real numbers
Let's fill in blocks 1 and 2 with real data. Macropix, an Italian manufacturer of LED walls and custom LED screens, is a Visilay client; the figures below are already public, in the manufacturing SEO case study and in the dashboard article. They refer to Google Italy, so the keywords are in Italian.
| Indicator | Value | Source and date |
|---|---|---|
| Position for "monitor pubblicitario" (advertising display) | from 88 (2020) to 2 (2025) | Semrush, Google Italy |
| Position for "ledwall" (LED wall, 4,400 searches a month) | 1 | Semrush, 2025 reading |
| Position for "ledwall" | 3 | Rank tracker, 21 September 2026 |
| Impressions and clicks for "ledwall" | 2,332 impressions, 28 clicks, CTR 1.20% | Search Console, window read on 22 September 2026 |
| Share of voice on the LED wall cluster | 25.55%, against 18.08% for amazon.it | Semrush, July 2026 |
| Google Ads, for comparison | 374 conversions for €16,805, €44.93 per conversion | Google Ads, July 2025-June 2026 |
The summary we'd write above this table is three sentences long: "On the LED wall cluster Macropix has the highest share of visibility in the market, ahead of Amazon. For the keyword ledwall the site has dropped from first to third place and CTR is at 1.2%. Next month we'll work on the LED wall page to win the position back and on the title tag to lift CTR."
Three sentences, four numbers and one decision. That's what the client actually reads; the rest is there if they want to look it up.
Every number has a source and a date, written next to it
Look at the table again. The same keyword, "ledwall", is in first position for Semrush in 2025 and third for the rank tracker in September 2026. Neither tool is wrong: the date, the crawler and the location setting are different. A report that just says "ledwall: position 3" without a source and a date invites the client to open another tool, find another number and wonder who's got it wrong.
The same goes for Google's own limits, which are written in the documentation and should be stated in the report rather than discovered by the client:
- Search Console makes data available 48 hours after collecting it and keeps it for 16 months. A year-on-year comparison only works if you've exported the history yourself.
- In the Performance report, data grouped by query, country, device or date is aggregated by property, while data by page is aggregated by page. Two totals side by side without a label produce a number nobody can explain.
- Searches made by very few people are anonymised and don't appear in the query breakdown, as Google explains in its deep dive into Search Console data filtering: on a site with a lot of long tail, the sum of the queries is always lower than the total.
- In Analytics 4, user-level data retention is set to 2 or 14 months and applies to explorations, not to standard reports. If someone left it at 2 months, you can't run an exploration over a quarter.
We've written a guide to SEO analytics on which source to trust when two numbers don't match.
In 2026 the report has one more line, and it's not where you'd expect
The "what we can't measure" block has grown for a specific reason: part of today's visibility produces no clicks, and part of the clicks come from channels that traditional reports don't separate.
The data on lost clicks all comes from the United States, and should be read as such. The Pew Research Center tracked the browsing of 900 US adults across 68,879 searches in March 2025: with an AI summary on the page, users clicked a traditional result in 8% of visits, against 15% without one. Links inside the summary were clicked in 1% of cases. Ahrefs, across 300,000 keywords with December 2025 data, measured a 58% lower CTR for position one when an AI Overview is present. Seer Interactive, across 3,119 queries from 42 client organisations, saw organic CTR on queries with an AI Overview fall from 1.76% in June 2024 to 0.61% in September 2025.
We haven't found an equivalent tracked-browsing study for the UK; what can be said about the effect on a single site is in our article on how AI Overviews affect SEO. For the report the point is a different one: where each kind of AI visibility ends up in the tools you use. None of the guides we read on the first page sets it out like this.
| Where you appear | Where you see it in the data | What you actually measure | What's missing |
|---|---|---|---|
| Google AI Overviews and AI Mode | Generative AI performance report in Search Console, rolled out to all sites on 31 August 2026 | Impressions only, by page, country, device and date | No query breakdown, no separate clicks |
| Clicks from AI Overviews and AI Mode | Analytics 4, Organic Search channel | Sessions and conversions | They can't be told apart from clicks on classic results |
| ChatGPT, Gemini, Copilot, Deepseek, Grok | Analytics 4, AI Assistant channel | Sessions from people who click a link in the answer | People who read the answer and don't click |
| Citations without clicks on any assistant | No Google tool | Only with third-party trackers, on a sample | Complete data doesn't exist |
Two details in the table come from the official documentation and change how the report is written. The Search Console generative AI report counts impressions only and leaves out Search Labs experiments. The AI Assistant channel in Analytics 4 collects traffic from ChatGPT, Gemini, Deepseek, Copilot and Grok and explicitly excludes AI Overviews and AI Mode, which stay inside Organic Search.
Put into one line of the report: "impressions in Google's AI answers went up; clicks from those answers can't be measured separately". Anyone who promises a client an "AI Mode traffic" column is promising a number Google doesn't publish. How to work on appearing in those answers is covered in our guide to AI SEO.
There's also a figure worth putting in the first block when traffic is flat: in the Seer study, brands cited inside the AI Overview had a 35% higher organic CTR than brands that weren't (0.70% against 0.52%). A citation doesn't replace a click, but it's a metric that moves before the click does.
The report's conversions are not the client's leads
The last line missing almost everywhere concerns the most important number. One of our own cases, anonymised: an Italian residential solar company working across six regions. Between 1 June and 26 August 2026 the ad platform recorded 618 conversions. The client's CRM counted 610, and 108 of those, 17.7%, came from areas the company doesn't serve. At a cost per lead of €50 to €80, the client estimated those useless leads at between €5,500 and €8,640.
The example comes from a Google Ads campaign, but the principle is the same for organic: a report that says "organic conversions: 40" without saying how many became real opportunities is measuring the form, not the business. When the client has a CRM, the right line is a double one: tracked conversions and qualified leads, with the difference explained. We covered how in the articles on SEO attribution and on ROI.
How often to send it
The guides agree on a range: weekly, monthly, quarterly. Our position is firmer. A one-page monthly SEO report, plus a longer quarterly review in which the priorities are questioned again. Weekly only makes sense in the first months of a project or during a migration, because SEO moves on timescales a week doesn't show; we've written about realistic timescales in how long SEO takes to deliver results.
Some UK context helps set expectations. In the UK Business Data Survey 2024, published by the Department for Science, Innovation and Technology in June 2024, 77% of UK businesses said they handled digitised data, but only 21% of those analysed it to generate new insights: 34% of small businesses, 58% of medium-sized ones and 72% of large ones. DSIT's AI adoption research, based on 3,500 interviews in 2025, puts businesses using at least one AI technology at 16%. Most small UK companies, then, don't analyse their data systematically. For many of those receiving an SEO report, the one-page version is the only format that actually gets read.
The tools to do it
The no-cost minimum is Search Console, Analytics 4 and a Google Sheet with the dates of the work done, connected in Data Studio. The name is back to the original: on 10 April 2026 the Google Cloud blog announced the end of the Looker Studio brand, and many guides still use the old name for the same product.
Above that level are the suites: Semrush has customisable report modules, SE Ranking has a Report Builder, AgencyAnalytics and Databox do white label for agencies. They solve the assembly work, not the data limits, which stay exactly the same because the source is always Google. We've compared the price lists in the article on the best SEO software.
None of these tools writes blocks 1, 3 and 4. That's why an automatically generated SEO report is still just a data export.
Four mistakes we find in the reports people send us
- The site's average position as the first number. It mixes thousands of different searches and moves for reasons you didn't cause. In the Macropix case, four keywords in the same third position had CTRs between 0% and 2.86%: position alone explains little.
- Estimated traffic confused with clicks. Again on Macropix, the rank tracker estimated 1,600 monthly visits on the tracked searches and Search Console counted 57 clicks over the available window. They're two different measures; if they're on the same page, they need different labels.
- Brand and non-brand in the same number. People who search for the company's name already know it. SEO work shows up on searches that don't contain the brand, and those need to be separated.
- No line on what went wrong. A report in which everything goes up every month stops being believed by the third month. A drop explained with its cause is worth more than three growth charts.
We've written an article on what's reasonable to expect from the people doing this work, on SEO client expectations, and a guide to the metrics worth keeping, on SEO KPIs.
In our SEO consultancy projects the monthly report is one page long and always has block 4; the numbers we judge ourselves by are on the projects page.
Frequently asked questions
It's a regular document that summarises how a site's organic visibility has performed over a period: clicks, impressions, positions and organic conversions, with a commentary on what changed and what will be done next. The data comes from Search Console, Analytics 4 and a rank tracker; the commentary is written by a person.
Usually monthly. A one-page monthly report with the main numbers and a more detailed quarterly review cover almost every project. A weekly report is only needed in the first weeks of work or during a site migration.
Five blocks: a summary of two or three sentences, a table with organic clicks, impressions and conversions compared with the previous year, a dated list of the work done, the data limits for the period and next month's priorities. Every number should come with its source and the date it was taken.
The report tracks performance over time and is repeated every month; the audit is a one-off technical check of the site that finds errors and opportunities. Tools that promise an SEO report in seconds actually produce an automated audit.
Partly. Traffic from ChatGPT, Gemini, Copilot and other assistants shows up in Analytics 4 in the AI Assistant channel. For AI Overviews and AI Mode, Search Console shows impressions only, with no query breakdown; clicks from those answers end up in organic traffic and can't be separated.
It depends on how automated the data collection is. With a dashboard connected to Search Console and Analytics 4 the numbers update themselves; what's left is writing the commentary, which in our experience takes an hour or two a month on a medium-sized site. That's the time worth protecting, because it's the only part the client reads.
Something the industry rarely says
In our experience the block clients value most is the one that feels riskiest to write: what we can't measure. Admitting that clicks from AI Mode can't be seen, that rare queries are missing, that the tracker and Search Console don't agree, looks at first like admitting weakness. The opposite happens. A client who has read those limits in your words doesn't discover them from a competitor in a sales pitch, and stops asking you why two numbers don't add up. A report that says what it can't see is the only one people trust even when the numbers go down.
