The short version
- Search Console cannot answer "am I gaining or losing against competitors". A rank tracker can.
- A local business needs 60-150 queries across four groups, not 1,200.
- Without geographic granularity (3-6 observation points) local data is a fiction.
- Track exactly three competitors: the leader, the peer and the riser.
- Treat the local pack as a separate ranking — it is won on entirely different factors.
One question comes up in every review meeting and is almost never answered properly: "are we gaining or losing ground against the competition?" Search Console cannot answer it, because it only sees your data. Analytics cannot either. Answering it requires observing the results page from the outside, systematically, with the same methodology every week. That is what a serious rank tracker does, and it is the job of the Google SERP module in Semalt.
But a badly configured rank tracker is an anxiety machine: graphs going up and down, screenshots in the client WhatsApp group, decisions taken on two-position movements that mean nothing at all. This article is about configuring it so it produces information rather than noise, with the particular emphasis a local market demands.
Why position tracking still matters
Every so often somebody announces that rankings no longer matter and only conversions count. That is half a truth wrapped in analytical laziness. Conversions are the end result and depend on a dozen factors besides SEO: seasonality, pricing, paid competition, landing page quality, reputation. When they fall, you do not know why. Position is an intermediate metric, far closer to the cause, which is exactly why it diagnoses well.
Three things only an external tracker will tell you:
What competitors are doing. If your clicks drop 20%, you need to know whether the whole market dropped, whether one specific competitor overtook you, or whether Google changed the SERP layout. Three different problems with three different solutions.
What the SERP actually looks like. How many ads sit above, whether a local pack appears, whether there is a featured snippet and whose it is, how many organic results survive above the fold. Position 3 on a page with four ads and a local pack is worth less than position 5 on a clean one.
What happens where you do not appear. Search Console records no impressions for queries you never surface on. Opportunities you have not touched yet are invisible in your own data by definition.
Choosing queries: fewer and better
The most common configuration error is tracking too much. We have inherited projects with 1,200 monitored keywords where nobody looked at anything, because the dashboard was unreadable. For a local business, 60 to 150 well-chosen queries cover the real market.
We organise the selection into four groups, and this matters more than the total count, because each group is read differently.
| Group | Queries | Read frequency | What it tells you |
|---|---|---|---|
| Commercial head | 10 - 20 | Monthly | Underlying competitive position; slow movement |
| Commercial tail | 30 - 60 | Fortnightly | Return on recent work; where money is made |
| Geo-modified | 15 - 40 | Monthly | Whether local SEO works outside the centre |
| Watchlist | 10 - 20 | Quarterly | Early warning on emerging competitors |
Group 1 — Commercial head (10-20 queries). The ones describing your main service with buying intent: "employment lawyer seville", "full flat renovation seville". High competition, high value, slow movement. Reviewed monthly. Watching them weekly adds nothing.
Group 2 — Commercial tail (30-60). Specific variants with less volume and more intent: "unfair dismissal lawyer seville", "small bathroom renovation seville price". This is where money is genuinely made and where movement is fast and attributable to specific actions.
Group 3 — Geo-modified (15-40). Your main query crossed with districts and surrounding towns. This group is the one that reveals whether your local SEO genuinely works or only works in the city centre.
Group 4 — Watchlist (10-20). High-volume informational queries where you do not compete directly but where competitors are visibly building authority. They serve as an early warning: when someone starts systematically occupying this layer, they will appear in the commercial layer within six months.
Geographic granularity: the detail that decides everything in local search
This is where most local projects go wrong, and it deserves a pause.
Configure tracking at country level and your data becomes a statistical fiction. Configure it at city level and you improve enormously but still lose crucial information, because Google does not treat a city as a point: it adjusts locally-intended results according to the user's actual position, and the difference between searching from the historic centre and searching from an outer district can be several places and an entirely different local pack.
Tracking at country or even city level without observation points produces a respectable average position that can hide the fact that half the city cannot find you.
The practical consequence: a business with a single city-centre location can sit at position 2 for half the city and outside the top 10 for the other half, while showing a respectable average that hides the fact that half its market cannot find it. If you have multiple locations, or serve the whole metropolitan area, tracking positions without geographic granularity is simply useless.
Our standard configuration uses between three and six observation points: the centre, one north, one east, one on the western riverbank, and where the service area justifies it, one in the outer commuter belt. That reveals the real coverage map, and the decisions change: instead of "we need better rankings", the plan becomes "we need to build presence in the east, where we do not exist".
Competitors: pick three, not fifteen
The competitor comparison is the most useful function in the module and the worst used. The temptation is to add everyone appearing on the SERP. The result is a chart with fifteen lines where nothing is distinguishable.
We track exactly three, chosen on deliberately different criteria:
The leader. Whoever dominates today. Not to copy, but to measure the distance and understand what content structure sustains their position.
The peer. Someone your size, your age, your budget. This is the honest comparison: if they rise and you do not, the gap is execution, not resources. It is the most uncomfortable and most instructive line on the chart.
The riser. The one who did not exist eighteen months ago and now appears in more queries every month. They are the ones who will take your market in two years, and the only competitor you still have time to react to.
From that comparison comes the one aggregate metric we find genuinely useful: share of visibility across your query set. Not an average position, which as we have seen lies, but what percentage of the total available presence you occupy against them. It is a number you can take into a board meeting without technical explanation, and it answers the opening question directly.
Reading movements without losing your mind
General rule: no individual movement means anything. Patterns mean something.
| What you see | What it means | What to do |
|---|---|---|
| 1-3 places on one query | Statistical noise | Nothing. Do not even put it in the report |
| Drops across many queries on one URL | Page-level problem: content overtaken, technical fault, lost internal links | Investigate that specific URL |
| Drops across the whole domain | Algorithm update or global incident | Check whether the whole SERP moved; wait two weeks |
| Competitor rising for months | They are doing something structural | Analyse what they publish and where links come from |
| Positions lost but clicks steady (or reverse) | SERP layout change | Review blocks: local pack, AI, ads |
The local pack: a SERP inside the SERP
On local queries with commercial intent, the three-listing Google Business Profile block occupies the dominant visual position and absorbs a very substantial share of clicks. Tracking your organic position without tracking your presence in that block means measuring half the match.
Organic ranking
Won with content, authority and technical health. Slow, stable movement.
Local pack
Won with proximity, primary category, data consistency and recent reviews.
Consequence
An excellent website can be shut out of the pack. Confusing the two costs a quarter of misdirected work.
The first thing to understand is that they are two independent rankings with different factors. Organic is won with content, authority and technical health. The local pack is won with proximity to the user, data consistency, a correctly chosen primary category, review volume and freshness, and activity on the listing. An excellent website can be shut out of the pack, and an impeccable listing can appear in the pack with a mediocre website behind it. Confusing the two leads to three months of content investment to solve a problem that a primary category change would have fixed.
The second point is the geographic volatility mentioned earlier, amplified. The pack recalculates on proximity far more aggressively than organic: appearing when the user is a kilometre away and vanishing at three is perfectly normal. For a business with a physical location, this defines the real catchment area — a business fact rather than a marketing one. It tells you how far your natural radius reaches and where you would need advertising or a second location to compete.
The third is competitive. In the pack you compete against a set of rivals that often does not overlap with your organic competitors: businesses with a poor website but two hundred recent reviews. When tracking shows the three pack listings always belong to the same names while you sit fourth in the expanded list, the quarter's work is clear, and it does not run through the blog.
Practical recommendation: flag which queries in your set trigger a local pack, and treat those as a separate category with their own actions. Mixing pack and non-pack queries in one table produces averages that mean nothing and workplans that attack the wrong thing.
From data to action: four typical decisions
A tracker that changes no decisions is an expense. These are the four decisions we most often take from this data.
Redistribute geographic effort. When observation points show uneven coverage, the priority stops being new content and becomes local presence: listings, reviews, citations and area pages with genuine content, not templates with the district name swapped.
Abandon a query. Unpopular and frequently correct. If you have spent two years at position 8 on a query where the top seven are national portals with incomparable budgets, that effort pays better elsewhere. SERP data — who occupies the positions and with what kind of page — is what lets you take that decision with an argument rather than out of fatigue.
Attack a featured snippet. When tracking shows a query carries a featured snippet held by someone you outrank organically, a quick win is available: restructure the relevant block of your page to answer the question in forty or fifty words, in extractable format.
Get ahead of the riser. When you spot the competitor climbing through the informational layer, the answer is not to imitate them. It is to occupy the adjacent commercial layer before they arrive, which is where they are not yet.
Frequency, reporting and honesty
Daily tracking makes sense for monitoring deployments or catching penalties, and no sense at all for steering a project. We work with a weekly read at group level, never individual queries, and a full monthly review.
On reporting, an opinion that does not always land well: a ranking report with two hundred rows in red and green is not transparency, it is a place to hide. A client cannot evaluate two hundred rows, so they evaluate the dominant colour — and the dominant colour can be engineered by choosing which queries to include. An honest report carries five numbers: share of visibility, movement by group, geographic coverage, material changes, and the three actions for the month, with a detailed annex for anyone who wants to check.
One rule we apply without exception: the tracked query set is fixed at the start of the quarter and not changed midway. Adding easy keywords mid-quarter to improve how the report looks is the most widespread way of lying with true data in this industry.
How it fits with the rest of the measurement
Position tracking answers "where am I against others". Search Console analytics answers "what do people do when they see me". AI visibility answers "am I named when there is no SERP at all". Three distinct questions, none of which substitutes for the others.
The practical benefit of holding them in one Semalt account is not the saved subscriptions, real though that saving is for a small business. It is that all three share a period and a domain, which lets you ask the question that actually matters. If you gain three places across the commercial tail and clicks do not move, you have a CTR or SERP-layout problem, not a ranking problem. Without the three layers side by side, that conclusion takes weeks.
And if you decide to execute through a managed campaign — AutoSEO from $149 a month, or FullSEO for more demanding projects — tracking stops being a surveillance panel and becomes verification of delivered work, which is how it should have worked all along.
Minimum viable configuration
60 queries across four groups, three geographic points and exactly three competitors. Four weeks untouched. In week five you get your first useful read.
Sign in to Semalt See Google SERPFrequently asked questions
How many queries should be tracked?
Between 60 and 150 for a local business. Beyond that the dashboard stops being readable and nobody makes decisions with it.
Is daily tracking worth it?
Only for watching a deployment or catching a penalty. To steer a project: weekly by group, monthly in full.
Why three competitors and not all of them?
Because fifteen lines on a chart are unreadable. Leader, peer and riser cover the three questions that actually matter.
Can the local pack be tracked like organic?
It can be tracked, but not mixed in. They are rankings with different factors: flag which queries trigger a pack and handle them separately to avoid meaningless averages.
A starting configuration you can build today
To begin without over-engineering: open your workspace, add the domain, and configure 60 queries spread across the four groups described, with three geographic points and exactly three competitors. Then leave it alone for four weeks — no daily checking. In week five you will have your first useful read, and it will be more informative than any audit built on a single snapshot.
If you would rather the setup were done by someone who has run the exercise a hundred times — query selection and geographic points are the part that determines whether the rest is worth anything — let's talk. Getting it right first time saves a quarter of useless data.