Ask ten agencies what goes into their monthly SEO report and you will get ten different answers, but nine of them will contain the same fatal ingredient: a table of keyword positions. It is the default, it is easy to produce, and it is close to useless as a management document. The client does not run a business that sells rankings.
This article is about the reporting and analytics layer of the Semalt platform — what it measures, how the tracking actually works under the hood, and how to assemble it into a monthly report that survives contact with a finance director. The tooling matters, but the framing matters more, so we cover both.
Key takeaways
- A keyword position table is the default SEO report and close to useless as a management document.
- Report segmented visibility, not average position — visibility moves in proportion to consequence, average position does not.
- Tagging keywords by intent at project setup is the highest-value habit in the entire platform.
- Include what did not work. A report with no failures reads as marketing and devalues everything else in it.
Why most SEO reports fail their audience
A report fails when the reader cannot answer "so what should we do?" after reading it. Position tables fail that test in three specific ways.
First, they average away the signal. Twenty keywords up two places and five keywords down twenty places produces a positive-looking average and a collapsing business. Second, they have no value weighting: position one for a query nobody searches sits in the same table as position eleven for the query that drives half your revenue. Third, they have no causality — a table of numbers tells you what happened but never why, so the meeting fills with speculation.
The fix is not a prettier table. It is a different reporting object: a small number of segmented, value-weighted metrics, each paired with what changed and what happens next.
Framing used in the reporting section below, not vendor benchmarks.
How rank tracking actually works, and why the number moves
Before trusting any tracker, it helps to know what the number is. A tracked position is a measurement of one SERP, from one location, on one device type, at one moment, usually depersonalised. Real users see something else — personalised, localised to their exact position, influenced by their history.
Semalt lets you pin the parameters that matter: search engine, country, city-level location for local queries, device, and language. That configuration is the difference between a number you can act on and a number that jitters for no reason. A Budapest dental clinic tracked at country level will show noise, because the SERP for "fogorvos" in the 13th district is not the SERP in Debrecen. Track it at city level and the volatility mostly disappears.
Set this once, at project creation. Location, device split and language are not settings you change casually — changing them mid-campaign breaks historical comparability. Decide them deliberately before the first data point, and document the choice in the project so the next person to open it knows why the numbers look the way they do.
The second thing worth understanding is SERP feature awareness. A "position 3" that sits below a map pack, a video carousel and four ads is not position 3 in any sense the user experiences. The platform records which features are present on tracked queries, which turns an inexplicable click-through collapse into an obvious one: your ranking did not move, but Google added an AI overview above it.
Set tracking parameters once, at project creation. Location, device split and language are not settings you change casually — changing them mid-campaign breaks historical comparability. A Budapest clinic tracked at country level will show noise, because the SERP in the 13th district is not the SERP in Debrecen.
Visibility: the one aggregate metric worth reporting
If you report a single number, report visibility rather than average position. Visibility weights each tracked keyword by its search volume and the estimated click-through rate of its position, producing an index that approximates "share of available attention in this keyword set".
Its virtue is that it moves in proportion to consequence. Going from position 9 to 6 barely changes visibility; going from 4 to 2 changes it a lot, because that is where the clicks are. Average position treats those moves as equal — three places either way — which is precisely why average position is a misleading headline metric.
Visibility also survives keyword set changes badly, which is a feature rather than a bug: it forces the discipline of freezing your tracked set for the duration of a campaign. Adding easy keywords mid-quarter to make the chart look better is a temptation the metric makes visible.
Segmentation: where the real story lives
The single highest-value habit in the entire platform is tagging keywords into groups at setup. Not because grouping is sophisticated, but because ungrouped data hides the story.
The segmentation that works for most commercial sites:
- Branded vs non-branded. If branded terms carry the visibility curve, your SEO is not working — your marketing is. Separating them is the single most honest thing you can do in a client report.
- Transactional vs informational. These behave differently, respond to different work, and convert at wildly different rates. Blending them produces a meaningless average.
- By page type or service line. Category pages, service pages, blog. This tells you which part of the site is earning and which is decorative.
- By location, for multi-branch businesses. A clinic with three Budapest locations has three different competitive situations.
Once segmented, the monthly conversation changes character. Instead of "we went up a bit", it becomes "transactional visibility on service pages rose 14% while informational visibility declined 6%, because we deprioritised blog maintenance to ship the service page rework — which is what we agreed". That is a report a client can act on.
Connecting positions to traffic, and traffic to money
Rankings are a leading indicator. Traffic is the intermediate one. Revenue is the only one anyone outside the marketing team cares about, and the reporting chain has to reach it or the work stays vulnerable at every budget review.
With Search Console connected, the platform aligns impressions, clicks and average position against its own tracked data on one timeline. That alignment is where diagnosis becomes possible. Three patterns come up constantly, and each means something specific:
There is a full walkthrough in The New Semalt: One Platform That Runs Your Entire SEO Workflow.
| Pattern | Likely meaning | Response |
|---|---|---|
| Impressions up, clicks flat | Ranking for more queries, but titles or snippets are not winning the click; or SERP features are absorbing it | Rewrite titles and meta descriptions; check for AI overviews and feature blocks |
| Clicks up, conversions flat | Attracting the wrong intent, or the landing page does not deliver on the query | Review intent match; the ranking page may be the wrong page |
| Positions up, impressions down | Search demand itself is falling, or queries are being answered in the SERP | Seasonality check first; then evaluate whether the topic still deserves investment |
The third row has become far more common. When impressions decline while rankings hold, the query is often being answered directly on the results page. That is not a failure of your optimisation, but it is a strategic signal: the value of that query has structurally decreased, and the effort belongs elsewhere.
The live dashboard versus the monthly PDF
Giving stakeholders read-only dashboard access changes the relationship. The monthly report stops being the moment when information is revealed and becomes the moment when it is interpreted, which is a much better use of an hour of everyone's time.
It also removes the worst dynamic in agency reporting: the client who suspects nothing is happening between reports. When they can see the crawl running, the positions updating and the ticket log moving, that suspicion never forms. We have found that clients with dashboard access ask for fewer status calls, not more.
The caveat is that a live dashboard needs a guide. Hand someone an unexplained visibility chart and they will interpret every weekly wobble as a crisis. Spend fifteen minutes at handover explaining what normal volatility looks like, which metric is the headline, and which charts to ignore day to day.
Building the monthly report
Here is the structure we use, in order, and the reason for each part.
- One headline number with its direction
Segmented visibility for the commercial keyword set. Not average position, not total keywords tracked. One number, compared to last month and to the campaign start.
- What we did
Three to five items, each phrased as a completed change rather than an activity. "Rewrote and republished six service pages" beats "content optimisation work".
- What moved, and why we think it moved
Here the aligned timeline pays for itself: technical changes, publication dates and ranking movements on one chart, so causality is shown rather than asserted.
- What did not work
Include it. A report with no failures reads as marketing and trains the client to discount everything else in it.
- What happens next, and what we need from you
Usually the second half matters more, because the blocker is nearly always a development ticket in someone else's backlog.
Everything else — full keyword tables, complete crawl output, backlink lists — goes in an appendix or, better, stays in the dashboard where the interested reader can go find it.
Reporting pitfalls that quietly destroy trust
Changing the keyword set to improve the chart. Clients notice eventually, and when they do, every previous report becomes suspect.
Reporting rankings for queries with no volume. Position one for a phrase nobody types is not an achievement, and presenting it as one signals either naivety or evasion.
Ignoring seasonality. A Budapest hotel's traffic falls in November for reasons that have nothing to do with SEO. Always compare year-over-year alongside month-over-month, or you will take credit for spring and blame for autumn.
Attributing everything to your own work. Competitors change, algorithms update, demand shifts. Reports that claim total causality are wrong often enough to be noticed.
Hiding the definition. If you report a composite metric like visibility, state how it is calculated somewhere in the document. Unexplained proprietary scores are the fastest way to make a numerate client distrust the whole report.
Belongs in the report
- Segmented visibility, branded separated from non-branded
- Completed changes, phrased as outcomes
- Technical events on the same timeline as rankings
- What failed, and what you need from the client
Belongs in the appendix, or nowhere
- Full keyword position tables
- Rankings for queries with no search volume
- Raw crawl exports and backlink dumps
- Unexplained proprietary scores
Cadence: what to look at, and how often
Weekly, look at the dashboard delta and nothing else — five minutes, checking for anything that moved sharply enough to investigate. Sharp movements are almost always technical, and technical problems are cheapest to fix immediately.
See also: The Semalt Website Audit.
Monthly, do the full reporting cycle described above, plus a competitor visibility check. If your curve is flat but the whole competitive set declined, you gained relative share, and that is worth saying out loud.
Quarterly, revisit the keyword set itself. Not to improve the chart, but because businesses change: new services, discontinued products, shifted target markets. A tracked set that is never reviewed slowly stops describing the business it belongs to.
Reporting for local and multilingual businesses
Local reporting has a structural problem that generic dashboards handle badly: the query that matters is not the one being tracked. A restaurant in the 7th district does not compete for "restaurant Budapest" — it competes for map pack placement on a hundred variations of "restaurant near me" issued from within two kilometres. Classic rank tracking cannot see most of that.
What you can do is proxy it well. Track a representative set at city and district-level location, record map pack presence as a separate signal, and pair it with the metrics that actually describe local performance: profile views, direction requests and calls from the business profile. Reported together, those tell a coherent story that a restaurant owner recognises. A position number alone does not.
Multilingual sites have the mirror problem: one report averaging three language markets describes none of them. A Budapest clinic serving Hungarian, German and English patients is running three different campaigns against three different competitive sets, with completely different demand curves. Segment the tracked set by language and report the three curves separately, or the strong market will conceal the weak one for as long as the retainer lasts.
The practical setup takes minutes: one keyword group per language, each pinned to its own location and language parameters, and one visibility chart per group in the report. It looks like more work than a single number, and it is — but it is the difference between a report that describes a business and one that describes an average.
Verdict
The analytics layer in Semalt does the things this kind of tooling has to do — configurable tracking, volume-weighted visibility, Search Console integration, segmentation, shareable live dashboards. What makes it worth building a process around is that it sits on the same data as the crawl and the competitor set, so technical changes, publications and position movements are readable on one timeline instead of three.
We cover this in detail in Competitor and Backlink Analysis with Semalt.
That single property is what turns reporting from a monthly chore into a diagnostic tool. When you can see that positions on category pages fell in the same week a deployment changed the canonical logic, you are not writing a report. You are solving a problem, and then reporting that you solved it.
If your current reporting is a keyword table someone assembles by hand on the last Thursday of the month, the fastest improvement available is structural rather than analytical. Set up a project, segment the keyword set properly, and let the first month of data tell you which part of your funnel is actually moving.
Frequently asked questions
Why report visibility instead of average position?
Visibility weights each keyword by search volume and by the estimated click-through rate of its position, so it moves in proportion to consequence. Moving from position 9 to 6 barely changes visibility; moving from 4 to 2 changes it a lot, because that is where the clicks are. Average position treats both as three places and is therefore misleading as a headline metric.
Can I add keywords to the tracked set mid-campaign?
You can, but visibility handles it badly by design, which is a useful discipline. Adding easy keywords in the middle of a quarter to improve the chart is something clients notice eventually, and when they do, every previous report becomes suspect. Review the set quarterly for genuine business reasons — new services, discontinued products — not to improve the curve.
Impressions are rising but clicks are flat. What does that mean?
Usually one of two things: you rank for more queries but your titles and meta descriptions are not winning the click, or SERP features and AI overviews are absorbing it. Check which queries carry answer surfaces before rewriting anything — if the demand is being satisfied in the results page, better titles will not recover it and the effort belongs elsewhere.
Should clients get direct dashboard access?
Generally yes, with a fifteen-minute handover explaining what normal volatility looks like and which metric is the headline. Read-only access turns the monthly report into an interpretation meeting rather than a reveal, and clients who can see work happening between reports ask for fewer status calls, not more.
Open your Semalt dashboard
Audits, rank tracking, competitor data and reporting in one place. Sign in and you will be looking at real numbers for your own domain within minutes.
Sign in to SemaltOr browse the service overview at semalt.com.