A good SEO report answers three questions. What changed, why it changed, and what you should do about it next. What usually arrives instead is a stack of charts pulled out of a tool, and the sections that take up the most pages are reliably the ones that change the least.
Get this wrong and you pay a retainer for a document you cannot act on. You approve another quarter because the line is green, or you replace a provider who was doing the right work because the line is red. Both mistakes come from the same habit, which is reading the document as a verdict on the provider rather than as evidence about your site.
What an SEO report is, and the three documents people mean by it
Three unrelated documents share the name, which is why two people can argue about one without noticing they are discussing different things.
The audit. A one-off crawl that lists what is technically wrong with your site. SEOmator’s free audit tool, read in September 2026, describes running 172 on-page checks across 16 weighted categories. It is a snapshot of defects and it says nothing about whether anyone is buying from you.
The ranking report. Positions for a list of keywords, tracked over time. It answers where you sit, not what it earned you.
The monthly performance document. What happened to traffic and conversions, why, and what is planned next. Whatagraph’s client template, read in September 2026, organizes exactly this into four tabs, performance summary, organic visibility, organic traffic and organic conversions.
The first is diagnosis, the second is a symptom tracker, the third is a progress record. A provider who sends you the first every month is describing the same defects repeatedly. One who sends only the second is showing you a number that moves on its own. If you are buying diagnosis, that is what an SEO audit is for, and it should be bought once rather than repeated.
Why whoever writes your report decides what goes in it
Redefine Web sells audits and written reports. So when this article tells you to demand something, you can demand it of us, and the standards below are the ones we expect to be held to.
The reason the document grows is not mysterious, and the clearest statement of it comes from the people selling the software that produces it. Swydo’s guide for agencies, read in September 2026, says the report is the only regular proof of work most clients get, so it carries more weight in a renewal conversation than the work itself. The same guide names retention, upsell and faster approvals as what good reporting delivers to the agency.
None of that is dishonest. It does explain the shape of what you receive. A document whose job is to survive a renewal conversation grows sections that look like effort, and a document whose job is to help you decide stays short and says uncomfortable things. Ahrefs puts it plainly in its own guide, read in September 2026, describing reports as a paper trail that documents progress, without which it is hard to prove the work you have put in.
So read every section by asking who it serves. That single question sorts most of what follows.
The nine sections a monthly search report usually contains
The industry has converged. Ahrefs and Semrush, both read in September 2026, publish near-identical section lists, and SE Ranking’s 2026 client guide adds the same spine with local and AI visibility bolted on.
| Section | Where the data comes from | The question it answers |
|---|---|---|
| Executive summary | Written by a person | What happened and what now |
| Organic traffic | Analytics | Did more people arrive |
| Conversions from organic | Analytics, with goals configured | Did arrivals turn into business |
| Keyword rankings | A rank tracker, on a chosen list | Where you sit for tracked terms |
| Technical health | A crawler | What is broken on the site |
| Content performance | Analytics and Search Console | Which pages earn attention |
| Backlinks | A link index | Who is vouching for you |
| Competitors | A third-party estimate | How others appear to be doing |
| Next actions | Written by a person | What gets done and by whom |
Two rows in that table are written by a human being and the other seven are exported. That ratio is the whole story. The exported rows can be produced in minutes for any site by anyone, and the written rows take the judgment you are actually paying for. A document heavy on the seven and thin on the two is cheap to make and expensive to buy.
One section appears on none of the three lists, and its absence is worth noticing. Nothing in the standard spine records what the provider actually did that month, in hours or in shipped changes. You get outcomes, and you get recommendations, with the work itself left as an inference between them. Ask for it as a line of its own, because a month with no shipped changes sitting beside a rising traffic line is a month you were carried by the season.
Which sections change a decision, and which fill pages
Not one of the guides above ranks its own sections or marks any as optional. So here is the ranking, judged on one test. If the section came back with the opposite result, would you do anything differently this month. For the two sections a person writes rather than exports, ask instead what changes if they are missing, because a summary has no opposite.
| Section | Verdict | Why |
|---|---|---|
| Next actions | Changes a decision | It is the only part that commits anyone to anything |
| Conversions from organic | Changes a decision | Traffic that never converts is a cost, not a result |
| Executive summary | Changes a decision | If it cannot be written honestly, the month had no story |
| Content performance | Changes a decision | Names the specific pages worth more work |
| Organic traffic | Context only | Moves on seasonality and brand spend you did not control |
| Technical health | Context after month one | The defects were found in the audit, this is a repeat |
| Keyword rankings | Context only | Depends entirely on who chose the tracked list |
| Backlinks | Rarely | Counts move without anyone doing anything |
| Competitors | Almost never | Third-party estimates of other people’s private numbers |
Four sections carry the decision and five are context. That is not an argument for deleting the five. Context is how you avoid overreacting to one bad month. It is an argument about proportion, and about where the pages went. If the four decision sections occupy two pages and the five context sections occupy eighteen, the document was built to be received rather than used.
A context section becomes a decision section occasionally, and knowing when is the difference between reading well and dismissing things. Backlinks matter in the month you lose a large number at once, which usually means a site that linked to you was redesigned or shut down. Competitors matter in the month a serious rival enters your category. Technical health matters in the month after a redesign, when a template change can quietly remove something that was working. Organic traffic matters when it moves hard and the season does not explain it, because an unexplained fall is usually a broken page rather than a market. Rankings matter when a term you genuinely sell on drops off the first page, which is an event rather than a drifting average. Outside those moments they are weather, and weather is for glancing at.
What the score on a free audit actually tells you
Almost every free tool opens with a number out of 100, and it is the most prominent thing on the page and the least useful. SEOmator’s free audit tool, read in September 2026, grades on a 0 to 100 scale with letter bands, A from 90, B from 80, C from 70, D from 50 and F below that.
On the same page SEOmator states it audited more than 100,000 websites across its network between January and July 2026, with a median score of 71 out of 100. Read those two facts together and the grading scale stops meaning what it appears to mean. A 71 is not a C in the sense of a school report. It is the middle of the pack, which is where half of all sites sit by definition.
The deeper problem is that every score is a weighted average chosen by the vendor, and the weights differ between tools. The same site scores differently on two of them without a line of code changing. That is why running your URL through several of the free SEO tools produces a spread rather than an answer.
Use the score for one thing only. A sudden drop on the same tool, measured the same way, is worth investigating. The absolute number is not a grade on your business.
How to read a keyword ranking report without fooling yourself
A ranking report shows positions for a list of keywords. It does not show the keywords that were left off the list, and the list was chosen by the person being judged on it.
That is the first thing to check, and checking it takes a minute. Ask for the tracked set as a file. Count how many terms carry your brand name, because brand terms rank first for everyone and inflate any average. Then count how many describe what you sell in the words a stranger would use. A set that is mostly brand and mostly long phrases nobody searches is a set built to produce green arrows.
Second, positions are personal now. Results vary by location, device and history, so a single number for a keyword is an average over a lot of different screens. Ask which location and device the tracking is set to, and whether that matches where your customers are.
Third, a position is not a visit. Moving from 14 to 9 can produce almost nothing, and moving from 4 to 3 can produce a lot, which is why position changes belong next to clicks from Search Console rather than on their own page. How positions translate into visits is covered in SEO ranking, explained.
What a technical audit report owes you beyond a list of errors
A crawler will always find something. The question is whether what it found matters, and a raw export never answers it.
Look at what the checks actually are. SEO Site Checkup’s own guide to reading its output, read in September 2026, walks through tests including a favicon test, a deprecated HTML tags test and a media print test alongside genuinely consequential ones like indexability and page speed. Those three can fail on a site that ranks perfectly well, and they can pass on a site nobody can find.
The same guide gives a useful concrete standard, noting that search engines truncate titles at roughly 70 characters and descriptions at roughly 160. That is the kind of finding you can act on in an afternoon.
So what a technical document owes you is not completeness, it is triage. Every finding should carry three things. What breaks for a visitor or a crawler if it stays, how many pages it touches, and who fixes it. A finding with none of the three is a line item that exists to make the file longer. Writing findings in a form somebody will actually act on is covered in our guide to how to conduct a website audit.
SEO reporting tools, and the part they cannot do for you
The software is genuinely good now. It connects to analytics and Search Console, pulls rank data, crawls the site, applies your logo and sends the file on a schedule. None of that is the hard part.
What no generator produces is the sentence that says this month was worse than last month and here is why. Every guide read for this article agrees on the section list and none of them ranks it, because ranking sections requires knowing your business and a template cannot. The two human-written rows from the table above are precisely the rows that no automation fills.
This has a practical consequence when you are comparing providers. A beautiful document proves the provider bought good software. It proves nothing about the work, because the same file can be generated for your competitor in the time it takes to paste a URL. What the platforms can and cannot do is set out in our comparison of website ranking software.
There is one more thing worth knowing about the software. A large part of this market is sold white label, meaning the platform is built so that the agency’s branding replaces the vendor’s entirely. That is ordinary and nobody is being deceived, but it does mean the polish in front of you was bought rather than made. If the design is the most impressive thing about what you receive, you have learned something about a subscription rather than about the work.
Judge the parts a tool did not write. That is a short read, and it is the read that tells you something.
How often the document should land, and why monthly is the default
SE Ranking’s 2026 client guide, read in September 2026, describes monthly as standard for agency retainers, weekly for high-velocity campaigns and quarterly for strategic reviews. That matches what the work actually looks like.
Monthly is the default for a reason that has nothing to do with convenience. Search changes are slow to register and noisy in the short term. A week of data will show you movement that reverses, and reacting to it costs more than ignoring it. A month is roughly the shortest window in which a change in the work can show up as a change in the outcome.
Live dashboards are a different product and a useful one, but they answer a different question. A dashboard tells you the current number. The monthly document tells you what somebody decided to do about it. If a provider offers you a live link instead of a written summary, you have been given the raw material and charged for the analysis.
Quarterly is right for strategy and wrong for accountability, because three months is long enough to lose a quarter before anyone raises a hand.
One cadence question is worth settling before you sign, which is what month one looks like. The first document of an engagement should not resemble the others. It should carry a baseline, meaning the numbers exactly as they stood before anyone touched anything, captured and agreed by both sides. Without that baseline every later comparison runs against a moving start, and any argument about whether the work helped becomes unresolvable. Ask for it in writing, dated, before the work begins.
The questions to ask about any report you are sent
Ask these in writing, before you renew. The answers separate a provider doing the work from one producing a document about it.
| Ask this | A good answer contains | The reply that should worry you |
|---|---|---|
| Which three findings would you act on first | Three named pages or fixes, ranked, with reasons | Everything in the file is important |
| Who chose the tracked keyword list, and when was it last changed | A named person, a date, and the reasoning | It was set up at the start |
| Which of these numbers would change if you did nothing next month | An honest list, usually most of them | All of them reflect our work |
| What did not work this month | A specific thing, and what replaces it | Everything is on track |
| Show me conversions, not sessions | Goals configured, with the definition stated | We report traffic, conversions are your side |
| Who writes the summary, and how long does it take | A named person and an honest number of hours | It is generated automatically |
The third question is the one that does the most work. A provider willing to tell you which of their own numbers are weather rather than results is a provider who will also tell you when something is wrong.
There is one more question, and it is for you rather than for them. What decision were you hoping this document would help you make. If you cannot answer that in a sentence, nobody can build something that answers it either, and you will keep receiving thorough files that leave you exactly where you started. Most engagements that drift do so because nobody ever wrote down what the reporting was for, and both sides then default to volume as a proxy for effort.
Red flags in a report, including the ones we have to answer
Sessions with no conversion line. If the document never reaches money or leads, nobody has connected the work to the business, and a rising session count can sit happily beside a falling number of customers.
The tracked keyword list grew after a bad month. Adding easier terms lifts an average without anything improving. Compare the list against last quarter’s.
A flag we have to answer, and you should hold us to it. The written summary is the part we charge for and the part that takes real hours, so ask us how many hours go into it and who writes it. A provider that cannot name the person is selling you an export with a covering note.
A second flag we have to answer. Any firm that both does the work and grades it is marking its own homework, and that includes us. Keep one measurement we do not control. A rank check you run yourself, or the lead count your own team keeps, is enough to catch a document that has drifted from reality.
A score presented as a result. A number out of 100 rising is not an outcome. It is a vendor’s weighted average of its own checks, and it belongs nowhere near the summary. The same applies to any website checker output pasted in without triage.
What to do with the document once you have it
Read it in the order it was not written in. Start at the next actions, then the summary, then conversions. If those three hold together, the rest is supporting material and you can skim it. If they contradict each other, you have found the month’s real story and the charts will not help.
Then do one thing with it. Pick the single highest-ranked action, put a name and a date against it, and check next month whether it happened. Most of what goes wrong in a search engagement is not bad analysis, it is good analysis that nobody actioned, arriving again the following month slightly reworded.
Acting on it does not mean approving it. Approval is what most documents are built to collect, and approval commits nobody to anything. Acting on it means one named person, one date, and one line added to whatever list your team already works from. If a finding cannot survive being turned into a task in your own system, it was never specific enough to be a finding in the first place.
If the document keeps telling you the same things, the problem has moved. Either the fixes are not being made, which is a scope conversation, or they are being made and nothing is changing, which is a strategy conversation. Those are different meetings with different people. Which provider is right for which is covered in how to pick SEO services, and if the site itself is the blocker a full website audit answers that faster than another month of charts.
What we would do first
Take whatever you were sent last month and run this on it, which takes about fifteen minutes. Cover every chart with your hand and read only the sentences. Swydo recommends this test in its own agency guide, read in September 2026, and it is the fastest honest thing in this whole category. If the sentences tell you what happened, why, and what is next, the document works and the charts were supporting evidence. If covering the charts leaves almost nothing to read, you have been buying exports.
Then count two things. How many pages carry the four decision sections, and how many carry the other five. And find the next actions list, take the top item, and ask whether it appeared on last month’s list too. A recommendation that has survived three documents unactioned is not a recommendation anymore, it is furniture.
Do that before you change provider. The document and the work are different things, and a good team can produce a bad file. It is worth one direct conversation to find out which you have, because switching costs you three months of continuity to fix a formatting problem.
If you would rather see what a ranked, written set of findings looks like before you judge anyone else’s, get a free website audit and hold us to everything above.
