How to Do SEO Reporting: Reports People Actually Read

Here’s how to do SEO reporting in one honest sentence: your report exists to answer three questions for someone who doesn’t do SEO — what happened, why it happened, and what you’re doing next — and everything that doesn’t serve one of those three answers is decoration. Most SEO reports don’t fail because the data is wrong. They fail because the three answers are drowning under forty charts nobody asked for. Great reporting is an act of selection and honest narrative, not dashboard volume.

I’ve been on both sides of this. I’ve built the 23-page monthly PDF that took a full day to assemble and that, I later learned, the client skimmed for ninety seconds before filing it in a folder called “SEO stuff.” And I’ve written the one-pager that got forwarded to a CEO with the note “this is why we keep them.” Same data. Same month. The difference was that one of them answered the three questions and the other one performed effort. Okay, let’s be honest with each other: a lot of SEO reporting is effort theater. We send big documents because big documents feel like proof of work. But the people reading them don’t want proof of work — they want to know whether the thing they’re paying for is working, and what you’re going to do about it either way. This guide is the whole system: who you’re writing for, which numbers deserve the page, how honest to be about where those numbers come from, and — the part that will do more for your career than any chart — how to report a down month without losing the room.

Who is actually going to read this report?

Before you open a single analytics tab, answer this: who reads it, and what decision do they make after? Because “the report” is not one document. It’s at least three different documents wearing the same name, and the fastest way to write a report nobody reads is to write one report for everyone.

The exec version is one page, and it leads with outcomes and decisions. What did organic search contribute to the business this period, is the trend up or down, and is there anything you need from leadership — budget, a dev sprint, a decision on a direction? Executives don’t want your average position chart. They want to know if the channel is healthy and whether anything requires their attention. If your exec summary can’t survive being read on a phone between meetings, it’s too long.

The practitioner version is the working document — the one you’d want to inherit if you took over this account tomorrow. Query-level movements, pages gained and lost, technical issues opened and closed, experiments running and what they’re showing. This version can be long, because its reader is paid to care about the detail. Just don’t send it to the exec and call that transparency. It isn’t transparency; it’s homework.

The client version sits in between, and it has one organizing principle: progress against the goals you agreed on at the start of the engagement. Not the metrics that happen to look good this month — the goals. If you never set those goals explicitly, that’s the real problem, and it’s fixable: here’s how to set SEO goals that give every future report a spine. A client report measured against agreed goals reads like accountability. A client report built from whatever went up this month reads like a sales pitch, and clients can smell the difference.

One report per audience beats one report for everyone, every single time. Yes, it’s more documents. They’re shorter documents, they’re mostly the same data at different altitudes, and each one actually gets read — which is the entire point.

What metrics belong in an SEO report?

Here’s the part nobody tells you: the metrics question is really a hierarchy question. Every number you could report sits at one of three levels, and the level determines how much page space it deserves.

Outcomes come first. Organic-attributed conversions, leads, signups, revenue — the things the business exists to produce. These lead the report because they’re the reason anyone funds SEO. And report them with attribution humility, stated plainly: organic search almost always influences more than it gets last-click credit for, because people discover you through a search, leave, and come back later through another door. Say that in the report, in one sentence, every time. “Organic-attributed” is an honest label; “organic generated exactly this much revenue” is a precision your attribution model cannot actually support.

Behavior comes second. Organic sessions, which landing pages earned them, and what visitors did after arriving — engagement, depth, the paths that lead toward conversion. Behavior metrics are the connective tissue between visibility and outcomes. They answer the question “is this traffic the right traffic?” — which matters, because traffic that bounces off a mismatched page is a content problem dressed up as a win.

Visibility comes third. Impressions, average position, and rankings for the terms that actually matter to the business. Visibility is a leading indicator, and leading indicators are genuinely useful — impressions often move before clicks do. But two caveats belong in the report, not just in your head. First, average position is an average of averages: it blends every query, every page, every location, and it can drop simply because you started ranking for more things. Say so when you show it. Second, rankings only count for terms with real intent behind them. A #1 for a phrase nobody searches is a screenshot, not a result.

Which brings us to the vanity trap, and I want to be direct about it because it’s the most common way SEO reporting goes quietly wrong: rankings up and revenue flat is not a victory lap. It’s a strategy question. It means you’re winning visibility for terms that don’t produce customers, and the honest report says exactly that — “we’re ranking better for terms that aren’t converting, so next quarter we’re re-aiming at terms that do.” That sentence stings for a second and builds trust for a year. The dishonest version — leading with the ranking chart and hoping nobody asks about the pipeline — buys you one comfortable meeting and a much worse one later.

The metrics hierarchy, on a card

Tier	Metrics	What they answer	Honest caveat to include
1. Outcomes	Organic-attributed conversions, leads, revenue	Is SEO producing business results?	Attribution undercounts organic’s influence; name the model you used
2. Behavior	Organic sessions, landing-page engagement, conversion paths	Is this the right traffic, and does the site convert it?	Analytics settings shape these numbers; keep them consistent period to period
3. Visibility	Impressions, average position, rankings for terms that matter	Is our presence in search growing?	Average position is an average of averages; third-party ranks are estimates

The hierarchy is also a layout instruction. Tier one gets the top of the page and the biggest type. Tier three gets a compact section near the bottom. If your current report inverts that — a wall of ranking charts up top, conversions in a footnote — you’ve found your first fix.

How honest is your data, really?

Every number in your report came from somewhere, and each somewhere has quirks. Naming them doesn’t weaken the report — it’s what separates a professional document from a confident-sounding one.

Search Console is your closest thing to ground truth for clicks and impressions, and it’s still truth-ish: the data is sampled, some queries are anonymized away, and the freshest days arrive late and get revised. Treat the last couple of days as provisional and compare like periods with like. If you haven’t already built the habit of pulling and interrogating this data properly, it’s worth the hour — here’s how to use Google Search Console data without tripping over its limits.

Analytics is where attribution lives, which means analytics is where your settings quietly shape the story. Change the attribution model or the conversion definition and “organic revenue” changes without a single visitor behaving differently. The fix is disclosure, not hand-wringing: name the model in the report (“last non-direct click,” or whatever yours is), keep it consistent, and flag loudly in the period when anything about the measurement changed. A footnote that says “we changed how we count; comparisons to last quarter are soft” is worth more than any chart on the page.

Third-party rank trackers deserve the bluntest label of all: there is no single true rank. Results vary by location, device, personalization, and the hour you checked; trackers sample and average their way to a number. That doesn’t make them useless — the trend line is genuinely informative. It makes them estimates, and the honest report calls them that, in writing: “positions are tracker estimates; directional, not exact.” Once you label an estimate as an estimate, nobody can retract it, because you never over-claimed it.

One rule ties all three together: never report a number you can’t explain the origin of. If someone asks “where does this figure come from?” and your answer is “the tool said so,” you don’t have a report — you have a liability with page numbers.

How do you report a down month?

This is the section that keeps your credibility, so I’m giving it the space it deserves. Every SEO has down months. Seasonality dips, algorithms shift, a migration loses a batch of pages, a tag manager update quietly breaks tracking. The down month is not the test of your work. The down-month report is the test of your character, and there is exactly one way to pass: say it plainly, first, in the top line.

“Organic sessions fell this month. Here’s what we know, what we don’t yet, and what we’re doing.” That’s the opening. Not a quietly rescaled y-axis. Not a chart soup so dense the dip needs a map to find. Not leading with the one segment that happened to grow and hoping momentum carries the meeting. The people you report to will eventually see the real number — in a dashboard, in a board deck, in next quarter’s comparison — and the only variable you control is whether they heard it from you first.

Then diagnose honestly, out loud, in roughly this order:

• Seasonality? Compare against the same period last year, not just last month. A dip that happens every year isn’t news; failing to mention it’s annual is.
• Algorithm update? Check whether the drop’s timing lines up with a confirmed or widely reported update — and hold that correlation as a hypothesis, not a verdict (more on that below).
• Lost pages? Did anything get deleted, redirected, noindexed, or eaten by a migration? Page-level diffing finds in ten minutes what topline charts hide for a quarter.
• Tracking break? Sometimes traffic didn’t fall — measurement did. A consent banner update, a tag change, a redirect dropping parameters. Rule this out before theorizing about anything else.

Next, separate what you control from what you don’t — calmly. Algorithm updates happen. They are weather, not referendum. The report should carry neither panic (“everything has changed, burn the strategy”) nor spin (“this actually proves we were right all along”). Something like: “part of this decline coincides with a confirmed update; we can’t control that, and here’s the part we can control and are acting on.” Grown-up sentences. They read like competence because they are competence.

And then state the response. A down month reported without a response plan is just bad news; a down month reported with a diagnosis and a plan is work product. What are you fixing, what are you testing, what will you know by the next report?

Here’s the asymmetry that makes all of this non-negotiable: one hidden down month costs more trust than ten plainly reported ones. When you report dips promptly and clearly, every green month you ever report becomes more believable — you’ve proven the numbers come out whether they flatter you or not. Bury one dip in chart soup and get caught, and every future report gets re-audited in the reader’s head. You’re not just reporting a month. You’re setting the exchange rate on everything you’ll ever report next.

The down-month checklist

• State the decline in the first line, in plain numbers — no euphemisms, no rescaled axes.
• Rule out a tracking break before any other theory.
• Compare year-over-year to isolate seasonality.
• Diff the page inventory: anything deleted, redirected, noindexed, or lost in a migration?
• Check timing against known algorithm updates — and label the match a hypothesis.
• Separate controllables from non-controllables, explicitly, in writing.
• End with the response: actions, owners, and what the next report will confirm.

How do you explain why things happened?

The “what happened” layer is a data pull. The “why” layer is where you earn the fee — and it runs on a habit most people adopt only after they’ve been burned: annotations beat memory. Every launch, every fix, every site change, every known external event gets logged the day it happens, in a running file next to your report template. Published the new comparison hub on the 9th. Fixed the faceted-navigation crawl trap on the 17th. Migrated the blog subdomain on the 23rd. When traffic moves three weeks later, you’re not reconstructing history from Slack scrollback — you’re reading your own log. The why layer is only as good as the diary behind it, and the diary only works if you write it as things happen, because memory is a liar with great confidence.

The second discipline of the why layer is algorithm-update honesty. When a drop (or a jump — this cuts both ways) coincides with a known update, the correlation is a hypothesis, not a proof. The honest phrasing: “the decline coincides with the update; pages matching the update’s reported focus were affected most, which supports but doesn’t prove the connection.” It’s two sentences of intellectual honesty that cost nothing in the moment and save you from the much worse position of having confidently blamed an update for what turns out to have been your own redirect bug. Explanations you have to retract are more expensive than no explanation at all — “we’re still investigating, here’s what we’ve ruled out” is a perfectly professional sentence.

What goes in the “what’s next” section?

Every report ends with the plan, and the plan has three parts: what we’re doing, what we need, and what to expect. Skip this section and your report is a rear-view mirror — accurate, maybe even interesting, and useless for steering.

What we’re doing is specific and small: the three to five actions underway, not a strategy essay. “Publishing the next six posts in the reviews cluster. Consolidating the four overlapping landing pages. Fixing the crawl budget issue on the filtered URLs.” If the quarter’s strategy is building topical authority in a cluster, say which cluster, how many pieces, and where the cluster stands against plan — that’s a strategy a reader can hold you accountable to, which is precisely what makes it credible.

What we need is the ask: the dev sprint, the subject-matter expert’s two hours, the budget decision, the approval that’s been sitting in someone’s inbox. Reports are the best ask vehicle you have, because the request arrives attached to the evidence for it.

What to expect is where expectation honesty lives, and here’s the rule I’d tattoo on every SEO contract: no traffic promises. Ever. You don’t control the algorithm, the competition, or the news cycle, so a precise forecast is a fiction with a number on it. What you can honestly offer are ranges and directions: “if the cluster performs like our earlier clusters, we’d expect visibility gains to start showing in impressions first, then clicks, over the following months.” Directional, conditional, grounded in your own site’s history rather than anyone’s industry benchmark. Under-promising isn’t timidity. It’s the only forecast style that never needs a retraction.

Can AI write your SEO reports?

Partly — and the boundary matters more than the enthusiasm. AI is genuinely good at the narration layer: paste in your verified numbers and your annotation log, and it will draft a clear, readable “what happened and why” faster than you will. For translating a practitioner report into an exec summary, it’s honestly great.

But the rules are hard rules. Paste real data — the model narrates numbers you provide; it never gets to produce one. Check the arithmetic — every percentage change and every total gets re-verified by you, because language models are fluent and innumerate in the same breath. No invented benchmarks — if a draft says “this beats the industry average CTR,” delete the sentence, because neither you nor the model knows the industry average CTR for your niche, and no credible source will hand you one that fits your exact situation. And never let AI “estimate” a number you didn’t provide. An invented figure in a client report isn’t a time-saver; it’s the retraction you’ll be writing next month. AI drafts, you verify, you sign. Whoever signs owns every digit.

How often should you do SEO reporting?

Monthly is the rhythm that works for almost everyone: long enough for data to mean something, short enough to catch problems while they’re cheap. Weekly SEO reporting mostly amplifies noise — search data wobbles week to week for reasons that resolve on their own, and narrating the wobble trains your reader to either panic or tune out.

Add a quarterly deep-dive on top: strategy against goals, what the last three months actually taught you, and what changes course next quarter. The monthly answers “is everything on track?”; the quarterly answers “is the track the right one?”

And embrace the no-news-is-ok month. Some months, the honest report is: numbers stable, work proceeding, nothing needs a decision — half a page. Padding that month into ten pages because ten pages is “what the client pays for” is reporting theater, and it wastes the exact hours that should have gone into the work the next report will describe. If you also run social reporting — same craft, different data — the one-dashboard version of that discipline is a genuine time-saver: a tool like SocialBlaze gives you publishing and analytics for every social account in one place, so the social section of your monthly doesn’t require logging into six networks. Scoped honestly: that solves social reporting, not SEO reporting — but the hour it returns is an hour the SEO section gets back.

What does a one-page SEO report look like?

Here’s the template I’d hand anyone starting out. Five sections, one page, in this order:

Section	What goes in it	Length
TL;DR	Two or three sentences: the state of organic, the single most important thing that happened, any decision needed	3 lines
Outcomes	Organic-attributed conversions/leads/revenue vs. last period and last year, with the attribution model named	1 small table
Drivers	Why the numbers moved: top gaining and declining pages/queries, tied to your annotation log	4–6 bullets
Issues	Anything wrong or at risk — stated plainly, with status. Down months lead here, never hide here	2–4 bullets
Next	What we’re doing, what we need, what to expect (ranges and directions only)	3–5 bullets

A fictional example, so you can see the register — illustration only, invented numbers:

• TL;DR: Organic demo requests grew again this month, driven by the integrations cluster. One issue: the pricing page lost visibility after the redesign — fix is live, recovery expected to show over the coming weeks. No decisions needed.
• Outcomes: 41 organic-attributed demo requests (34 last month; 28 this month last year). Model: last non-direct click, which likely undercounts organic’s assist role.
• Drivers: Integrations cluster (published over the prior two months) now drives a third of organic demo requests; two posts reached page one for their target terms. Pricing-page decline traced to heading structure lost in the redesign — annotation logged on the 12th, fix shipped on the 19th.
• Issues: Pricing-page visibility down; fix live, monitoring. Rank-tracker positions are estimates; directional only.
• Next: Six more cluster posts this month; need two hours with a product engineer for the API comparison piece; expect impressions to move before clicks on the new posts, based on how earlier clusters behaved.

Notice what that example does quietly: it names the attribution model, labels estimates as estimates, reports a problem in the same breath as the wins, ties every driver to a logged change, and forecasts in directions instead of promises. Nothing in it would ever need to be retracted. That’s the entire standard — not impressive, retraction-proof. Reports people actually read, carrying numbers nobody has to walk back: that’s how to do SEO reporting like someone who plans to still be trusted next year.

FAQ: how to do SEO reporting

What should an SEO report include?

Three things: what happened (outcomes first — organic-attributed conversions or revenue — then behavior, then visibility), why it happened (tied to a running log of changes and known external events), and what you’re doing next (actions, asks, and directional expectations). A one-page version with a TL;DR, outcomes, drivers, issues, and next steps covers almost every audience.

How often should you send SEO reports?

Monthly for the standard report, with a quarterly deep-dive that reviews strategy against goals. Weekly reporting mostly narrates noise, since search data wobbles for reasons that resolve on their own. When a month has no news, say so briefly instead of padding — a short honest report beats a long performative one.

How do you report an SEO traffic drop to a client?

State it plainly in the first line, then diagnose in order: tracking break, seasonality (compare year-over-year), lost or changed pages, and possible algorithm-update timing — labeled as a hypothesis, not proof. Separate what you control from what you don’t, and end with the response plan. Reporting a drop promptly builds more trust than ten smooth months.

Which SEO metrics matter most in a report?

Outcomes first: organic-attributed conversions, leads, or revenue, with the attribution model named. Behavior second: organic sessions and landing-page engagement. Visibility third: impressions, average position (with the caveat that it’s an average of averages), and rankings for terms with real intent. Rankings up with revenue flat is a strategy question, not a win.

Can you use AI to write SEO reports?

Yes, for narration — paste in your verified numbers and change log and let it draft the story. The hard rules: it never produces a number you didn’t provide, you re-check all arithmetic, and you delete any invented benchmark like “industry average CTR.” The person who signs the report owns every digit in it.