Most SEO reports are a screenshot of a traffic graph and a paragraph explaining why the line went up. The client nods, nobody changes anything, and the invoice goes out. That is not reporting. That is a mood board.
A report is supposed to change a decision. Which page do we fix next? Which title do we rewrite? Which 404 do we redirect? Which page should we delete? If the report does not answer one of those, it is decoration.
I have been on both sides of this. I have sent the 20-page PDF, and I have received it. The 20-page PDF is usually a way to avoid saying "I do not know yet".
The three reports that actually change something
1. Pages with impressions and no clicks
This is the cheapest traffic on any site. Google already shows the page for a query. People see it and do not click. The fix is usually the title and meta description, not a new article. Pull the last 90 days from Search Console, sort by impressions, filter out the pages with a good CTR, and look at the ones with 1,000 impressions and zero clicks.
This site had 820,323 impressions and 169 clicks in the 90 days to 9 November 2026. That is a CTR of 0.02%. Most of those impressions were on page six or worse, so the title was not the problem. The ranking was. That is a different report. The point is that the impression number alone told me nothing until I looked at position and clicks.
2. 404s with clicks
A 404 with no clicks is housekeeping. A 404 with clicks is a leak. Googlebot hitting a dead URL is one thing. A person landing on it is another. Pull the 404 list from Search Console, join it to the pages that had clicks in the last 90 days, and write the redirect map for the ones that matter.
On one of my own sites I found 667 Googlebot 404 hits across 396 URLs. Most of them were old image sizes. A handful had real clicks. The redirect list took an afternoon and removed a month of noise from the crawl logs.
3. Page weight and LCP on the pages that convert
A site can have a fast homepage and a slow money page. The homepage gets the test; the money page gets the traffic. Pull PageSpeed field data for the five pages that actually convert, not the five you like. If the LCP is over 4 seconds on a phone, the copy is not the first problem.
On one site the image cleanup removed 6.57 GB of duplicate files and 8,327 orphan copies. Across sixteen sites the average page went from 3.5 MB to 2.9 MB. The worst LCP went from 9.8 seconds to 5.5. That is not a ranking guarantee. It is a page that is less annoying to load.
Where Semrush fits
Semrush is good at the things Google will not tell you: competitor keywords, paid ads, backlink gaps, and the shape of a market. It is not good at telling you what happened on your own site. Search Console is better for that, and it is free.
The useful split is: Search Console for first-party truth, Semrush for the competitive picture, PageSpeed for field performance, and a spreadsheet for the decisions. A reporting tool that tries to replace all four is usually a dashboard with a login and a renewal email.
Other tools in the category are fine. Ahrefs is strong on links. Moz is strong on a few metrics and weaker on others. SE Ranking is cheaper. The right one depends on which question you ask most. Do not buy a suite for a question a free export can answer.
The report template I use
- Cover. What we looked at, what we changed, what we did not touch, the date, and who did the work.
- Before and after. Page weight, requests, image bytes, LCP, 404 count, inode count, plugin count. Every row dated and read from a file, not from memory.
- Change log. Page by page, what changed, with the old version saved. If a change is reverted, the log says so.
- Open items. The things I recommend and did not do, with the reason: your call, my call, needs a licence, needs a paid feed.
- The honest paragraph. What this does not promise. No rankings, no traffic, no revenue.
That last section is the one that gets me repeat work. A buyer who has been burned before can tell the difference between a report that admits the limits and a report that is selling the next invoice.
What I would not put in a report
A "health score" that goes from 62 to 78 with no explanation. A keyword ranking chart for a term the client does not care about. A backlink count from a tool that has not crawled the site. A traffic graph with no annotation. A recommendation to "improve content quality" with no page, no example, and no next step.
If a number cannot change a decision, cut it. The report gets shorter, the meeting gets shorter, and the work gets clearer.
If you want the audit that produces this kind of report, start with the free scan. If you want me to run the whole thing, the Fix Sprint is $1,500 for one site up to 50 pages. You get the before and after numbers, the change log, and the open items. No ranking guarantee.