From Search Console to an action plan: search analytics with Semalt

Almost every business we work with has Search Console connected. Almost none of them use it. They log in when something feels wrong, glance at the clicks graph, confirm the line has not collapsed, and close the tab. That is a waste, because Search Console is the only place on earth where you can see, with first-party data and no estimation, what people actually ask Google before they find you.

The reason it goes unused is not laziness. It is that the native interface is built for looking things up, not for deciding things. Filtering, exporting, comparing two periods and arriving at an action are four operations the default tool makes awkward. The Search Console analytics layer inside Semalt exists to fix exactly that: the same data, reorganised into views that end in a decision. It is free, with no time limit.

This article is not about the tool. It is about the six readings we run on that data every month, and what we do with each one. The tool simply makes them take minutes instead of afternoons.

First move: split brand from non-brand

This is the first cut and the one most people skip. Searches for your own name ("martinez accountants seville") and generic searches ("tax accountant seville") are two different businesses sharing one graph.

Brand searches measure recognition: how many people already know you and look you up by name. They rise when you advertise, when local press covers you, when someone recommends you. They carry very high click-through rates — 60% to 80% is normal — and convert well, but SEO did not bring them. Your reputation did.

Generic searches measure acquisition: people who do not know you and have a problem. That is the SEO job. They carry much lower click-through rates, roughly 2% to 8% depending on position, and they are the only growth genuinely attributable to search work.

The classic mistake is presenting a 35% rise in organic clicks as an SEO win when 90% of the increase came from brand searches following a local radio campaign. Separating the two flows and tracking them as independent series changes the entire conversation about results. Our rule: if non-brand traffic is not growing, SEO is not working, whatever the combined graph says.

Reading 1: the CTR gap, or revenue you have already earned and are not collecting

Sort queries by impressions and compare actual click-through rate against the expected rate for the average position you hold. Sensible orders of magnitude for non-brand search are roughly: position 1, 25-30%; position 2, 13-16%; position 3, 9-11%; positions 4 to 6, 4-7%; positions 7 to 10, 1.5-3%.

When a query with real volume performs well below its band, you have a presentation problem, not a ranking problem. The usual causes, in the order we encounter them:

The title does not contain the query. You rank on semantics, but the user does not recognise their question in your result and keeps scrolling.

The meta description is autogenerated or missing. Google rewrites it with an arbitrary snippet, sometimes your navigation menu.

A block is eating the click. A local pack, a competitor's featured snippet, an image carousel. Here the weak CTR is not your fault, and the correct decision is different: either attack that block specifically, or stop investing in the query.

Intent mismatch. You rank a commercial page for an informational query. The user sees "pricing" where they wanted "how it works" and moves on.

The appeal of this reading is economic: it requires no ranking gains at all. Rewriting forty titles and descriptions is an afternoon of work and typically produces 10% to 25% more clicks on identical traffic. It is the best effort-to-result ratio in the whole discipline, and virtually nobody does it systematically.

Reading 2: striking distance, prioritised by business value rather than volume

Queries sitting in positions 5 to 15 offer the most favourable effort-to-return ratio: relevance is already recognised, you just need a push. But prioritising them by impression volume, which is what everyone does, is an expensive mistake.

We score them on three criteria. Commercial intent: does this query precede a purchase or is it curiosity? "How much does a bathroom renovation cost in Seville" is worth ten times "history of Sevillian tilework", even if the second has triple the volume. Real distance: moving from 6 to 3 is weeks of work; moving from 14 to 3 can be a year. Concentration: if ten mid-distance queries point at the same URL, you have one job with ten returns.

That third criterion usually decides the quarterly plan. In a recent audit of a city-centre clinic, 38 queries in positions 6 to 12 pointed at four pages. Rewriting those four pages — not forty — moved 22 of those queries into the top five within two months.

Reading 3: the query-to-page mismatch

This reading uncovers the most structural problems. For every important query, look at which URL is ranking and ask whether it is the one that should be.

Three pathologies show up, covered below.

Cannibalisation. Two or more URLs alternating for the same query week by week. Google is unsure which one is authoritative and splits equity between them. The symptom in the data is an unstable average position with a sawtooth shape. The fix is almost never "optimise both harder" — it is consolidating into one and redirecting the other.

Wrong page ranking. A blog post ranks for a clearly transactional query while your service page, the one that converts, sits on page three. Usually resolved with internal links from the post to the service using descriptive anchors, plus steering the post firmly towards informational framing.

The homepage absorbing everything. Typical on small sites: the front page ranks for specific service queries because it concentrates all internal authority. It works short term and blocks growth medium term, because one page cannot be genuinely relevant to fifteen different intents.

Reading 4: countries and devices, or why your mobile CTR is lying to you

The device breakdown usually reveals a considerable gap between desktop and mobile click-through rate for identical queries. Part of that gap is natural: mobile carries more blocks above the organic results. But when the difference exceeds ten points, something else is happening.

The most common cause is title length. Mobile truncates around 50 to 55 characters, and many carefully written titles lose precisely the segment carrying the value proposition. The second is performance: if your page takes four seconds to paint on 4G, a share of clicks is lost before it registers as a session, and users bouncing back degrade your quality signals.

The country breakdown has a less obvious and very useful application in a tourist city: detecting genuine international demand. Local businesses in hospitality, accommodation, dental and cosmetic treatment frequently pick up meaningful impressions from the UK, Germany or France without ever having targeted them. That data point is what justifies — or refutes — investing in an English version. In our own case, the English edition of this site came from exactly that observation, not from a hunch.

Reading 5: pages losing air

Comparing two 28-day periods at URL level and sorting by lost clicks is a ten-minute exercise that almost nobody runs monthly. It catches three things before they become serious.

Stale content: guides with three-year-old figures that competitors have refreshed. The fix here is an honest revision, not a date change. Loss through SERP change: the query still has volume, your position has not moved, but impressions fall because a new block has taken the space. And silent deindexation: pages that vanish entirely after an accidental noindex following a deployment, a misapplied canonical, or a robots.txt block. That last case is rare and devastating, and the only way to catch it is by looking.

Reading 6: the publication cycle, properly measured

The last reading is temporal. When you publish something new, how long until it appears, how long until it settles, and where does it end up?

Having that curve measured on your own domain is worth more than any industry average, because it depends on your authority, your crawl frequency and your competition. On small domains we typically see first impressions between day 3 and day 12, and a stable position between week 6 and week 12. With that figure you stop making premature decisions — rewriting an article in week two because "it is not working" — and you can plan an editorial calendar against real timescales.

This is where having analytics and indexing in one account saves real time: submit the URL, watch GoogleBot hits arrive in the live log, then watch impressions enter on the same timeline. No reconciling exports or time zones across three tools.

Four traps in Search Console data

Before making expensive decisions, it helps to know the data has known boundaries. Ignoring them produces wrong conclusions dressed as rigour.

Average position is an average, and averages lie. A query with an average position of 8 might be at 3 for 40% of impressions and at 15 for the rest, depending on user location, history and device. In local search this spread is enormous: someone in one district and someone across the city see different listings for the same query. Before celebrating or mourning a tenth of a point, check whether the query has enough volume for the average to mean anything. Below roughly a hundred impressions a month, it does not.

Some of your queries never appear. Google anonymises infrequent searches to protect privacy. On small domains that means 30% to 50% of real clicks carry no visible query. The practical consequence: click totals in the query view will never reconcile with the overall figure, and the long tail you cannot see is usually more valuable than it looks. Do not build strategy on the assumption that you can see the whole map.

Impressions are not views. An impression is counted when your result is on the loaded page, whether or not the user scrolls far enough to see it. In positions 8 to 20 this systematically inflates impressions and depresses apparent CTR. One more reason not to compare position 4 and position 14 click rates as though they measured the same thing.

History is capped at 16 months. If you want a year-on-year comparison two years from now, you have to be exporting from today. It is the strongest argument for pushing the data into a tool that retains it beyond the native window, and something we treat as a given on any project planned beyond twelve months.

A cadence that actually holds

None of these readings is difficult. Doing them regularly is. This is the rhythm we recommend to clients and run internally:

Weekly, 15 minutes. Non-brand clicks versus last week. New pages: have they entered? Sharp drops on important URLs. Nothing else.

Monthly, 90 minutes. CTR gap across the top 50 queries. Striking-distance review. 28-day comparison by URL. Output: three to five concrete tasks, not a report.

Quarterly, half a day. Full query-to-page mismatch review, cannibalisation, country and device breakdowns, editorial calendar checked against the real maturation curve.

Every cycle must end in a task list with an owner and a date. A report nobody converts into work is wasted time in a nicer format.

Where analysis ends and execution begins

It is worth being honest about the limits. These six readings tell you what is happening and where to intervene. They do not write the content, earn the links, or restructure your internal architecture. That work still has to happen, and it is where most projects stall: the diagnosis is clear and there are no hands to execute it.

This is where the Semalt arrangement makes sense. Analytics are free because the business is managed campaigns: AutoSEO from $149 a month for a steady rhythm of optimisation and link acquisition, and FullSEO from $500 for projects needing deeper implementation. You can use the measurement side alone for months at no cost, then decide whether to outsource execution or keep it in house.

Our position is pragmatic: for a local business with one person on marketing, free analytics plus two well-directed days a month produces perfectly respectable results. For projects with more pages, more competition, or no internal resource, a managed campaign works out cheaper than hiring and training someone.

Frequently asked questions

Start with what you already have

If you have read this far, you almost certainly have months or years of accumulated data you have never examined with this structure. Nothing new needs starting. What is needed is reading what is already there.

Connect your property in the Semalt workspace, spend ninety minutes on readings 1 and 2 — the CTR gap and striking distance — and extract five tasks from them. It is the most profitable starting point we know, and verifying that costs nothing. If you would rather we ran that first analysis and handed it back as a work plan, get in touch: the initial audit is free and we deliver it within 24 hours.