Call them at nine on a Sunday
A rank tracker tells you where you are. It cannot tell you whether the thing is worth being first at.
The most useful number I know of in a local competitive analysis is not in any tool, costs nothing, and you collect it with a telephone.
Call your competitors at nine o’clock on a Sunday night. Count how many of them answer.
That’s it. That is the measurement. If you sell anything with an emergency in it — water where water should not be, no heat in February, a door that won’t lock — then some meaningful share of the money in your market is moving at hours when a tool’s dashboard is a flat line and a human being is standing in a hallway deciding who to try next. Competitor after-hours answer rate. Nobody sells it to you because nobody can; it isn’t in a database anywhere. It is, however, entirely knowable by anyone willing to spend forty minutes and look slightly odd.
I bring it up because of what it exposes about the numbers we do pay for.
What we nearly built instead
The job on the table is a set of skills that work out where a local business should spend its attention. The obvious build — and I want to be honest that it was the first shape drawn — is a rank tracker. Feed it your keywords, check your position every day, draw the line, watch it go up.
Reasonable. It’s what the category sells. And it’s the wrong unit, for a reason that is faintly embarrassing once you say it out loud:
A rank tracker tells you where you are. It cannot tell you whether the thing is worth being first at.
You can be position one on a query nobody asks. You can be position eleven on a query where positions one through ten are a directory, a directory, a forum thread, a manufacturer’s PDF and four more directories — which is to say position eleven on a query nobody has actually answered yet. Those are wildly different situations and the rank chart draws them identically.
The unit is an opportunity
So the thing being scored is not a ranking. It’s an opportunity: a cluster of related questions, asked in a particular place, put through three questions in order.
- Is anyone asking? Volume, and how it moves through the year.
- Are the current answers weak?
- Can we actually win it?
Question two is where the whole design lives, and it is the one the industry answers with a number between zero and a hundred.
We are not using that number. “Weak” is a composition, not a score. Go and look at page one and write down what kind of thing each result is — a local operator, a directory, a forum, a manufacturer’s documentation, thin syndicated filler someone published in forty towns at once. If page one for a question in your town contains no local operators, then nobody local has answered it. That is what underserved means here, and you can see it with your eyes in about nine seconds per query.
Observed, not scored.
The difference matters because a difficulty score is a summary of how hard it would be for some site to rank, and you are not some site. You are the only result on that page with a van, a licence and a phone number in that county. A page of directories is a hard page for another directory and an open door for you. One number cannot hold both of those, so it holds neither.
The ceiling nobody tells you about
The third question — can we win it — has a version most owners find out about the expensive way.
Map visibility falls off with distance from your verified address. Not gently, and not at a boundary you chose. So for any town on your list there is a real ceiling on how much of that town’s map-pack business you can capture, and it is set by geography you cannot argue with.
Which means some of the pages you were about to build should never be built. If a town sits far enough out that you will not surface in the map results there, an area page for it is a page you pay to write, pay to maintain, and which competes for attention with the towns where you could have won. The analysis is supposed to tell you which ones those are before the invoice, not after.
That is the growth hack, if it deserves the name: most of the value is in the work you decline.
What’s actually built
Nothing. I want that on the record, because this is a design and designs read like results if you let them.
There are no collectors, no scripts and no data. What exists is a shape: two grains of record, both keyed to the site’s existing slugs and both stamped with the run that produced them. An observation grain — one row per query per time it was looked at, append-only, never edited. And an opportunity grain — one row per service, area and segment, per run, carrying the verdict and the evidence that produced it.
Snapshots, deliberately. Not a daily series. Opportunity moves on the scale of seasons and competitors’ decisions; a chart of it sampled every morning is a mood ring with extra steps.
The target is a matrix of twelve services across eight towns — ninety-six cells — and the reason the scoring can key to those at all is a decision someone made much earlier. The instance’s configuration file opens with this:
This is the seam. Every layout and component in the site reads from here and nothing hardcodes a business detail, so standing up a new instance is editing this file plus the design tokens.
Twelve services and eight areas, typed in one file, shaped identically to the database tables that will replace them later. The analysis has something to attach to because somebody wrote the business down as data instead of as pages. That is unglamorous, it was done weeks ago for an unrelated reason, and it is the only thing making any of this possible.
Two questions are still open and are not mine to close: where the query data comes from, and which repository owns the output. Both are the kind of choice that quietly commits you to a vendor’s shape if you make it casually, so they’re staying open.
If you run a business, this is the thing
You are being sold scores. Difficulty, authority, visibility, health — a number between zero and a hundred, arrived at by a method you cannot inspect, about a market the vendor has never been to.
Some of them are fine. None of them can see the things that decide whether you win: whether the people currently ranking are businesses like yours or aggregators wearing your town’s name, whether a competitor picks up the phone at nine on a Sunday, whether the map will even show you in the town you just commissioned a page for.
Every one of those is observable. By you. This week, with a browser and a phone, for nothing.
So before you buy the dashboard, go and look at page one of the five questions your best customers actually ask, and write down what kind of thing each result is. Then start dialling.
Observed, not scored. It’s cheaper, and it’s yours.