The Arithmetic of Rank Tracking

Rank tracking is where SEO budgets quietly go. Not because a rank check is expensive — it is one of the cheapest things you can buy — but because it is the only part of the workflow that runs whether or not anyone is looking.

Research is bounded by your attention. You pull data, you read it, you stop. Tracking is a standing order. You set it up once, in an optimistic mood, and it bills every day until somebody remembers to turn it off.

The multiplication

The cost of tracking is one line:

keywords × locations × devices × checks per period

Every term multiplies. That is the whole problem, because each factor looks harmless when you add it.

A hundred keywords is a modest list. Three locations is reasonable for a business with three markets. Desktop and mobile is obviously sensible; they genuinely return different results. Daily is the default in most tools, because daily sounds like diligence.

100 × 3 × 2 × 30 is eighteen thousand checks a month, from a setup where every individual decision was defensible.

Now the value side. How many of those eighteen thousand data points changed a decision last month? For most sites the honest answer is a handful. Positions do not move daily in any way you can act on; day-to-day variance is mostly noise from personalisation, location, and Google testing.

Where the multiplication is actually load-bearing

Not all of the factors are equal, which is what makes this tractable.

Frequency is almost always the cheap win. Going from daily to weekly cuts the bill by roughly eighty-five per cent. What you lose is the ability to say a change happened on Tuesday rather than that week. Unless you are actively mid-migration or watching a specific launch, that precision is not worth four times the cost — and daily data mostly tempts you to react to noise.

The exception is real: during a migration, a recovery, or the week after a large change, daily is exactly right. Turn it up deliberately, then turn it back down. The mistake is leaving it at the emergency setting permanently.

Locations are load-bearing only if you are local. If you serve one national market, one location is the honest configuration. If you are a multi-location local business, location is not overhead — it is the entire point, because your ranking genuinely differs street by street and a national average tells you nothing.

Devices are worth it less often than people assume. Mobile and desktop results do differ. But if you have never once taken a different action based on the gap between them, you are paying double for a comparison you do not use. Track mobile, since that is where the traffic is, and add desktop when you have a specific reason.

Keyword count is where the discipline belongs. This is the factor people try to optimise last and should optimise first.

The list is the decision

Most tracked keyword lists are aspirational. They contain terms someone hoped to rank for, added during a research session, and never removed.

A tracked keyword should meet one test: if it moved, you would do something. Not "you would be pleased" — you would take an action.

That test removes a lot. It removes terms you rank first for and will keep ranking first for, where movement means a rare emergency you would hear about anyway. It removes terms you rank fortieth for, where the next twenty places are not reachable this quarter and the number will not change. It removes near-duplicate variants that move together, where tracking one tells you what all six did.

What survives is usually short: the terms where you are close enough to move, on pages you are actually working on, plus a small number of genuinely important terms you want to be told about if they break.

For everything else, the data still exists — you can look up where you rank whenever you want, on demand, for the price of one check. The question tracking answers is not "where do I rank", it is "tell me when this changes", and that is only worth paying for on the terms where you would act.

Estimate before you commit

The useful habit is to price the configuration before you save it, not after the first invoice. Multiply it through yourself, or use a tool that will: PandaCrawl has a cost estimator on the rank tracker for exactly this reason, and its MCP server exposes the same estimate to an agent, so you can ask what a setup will cost before creating it.

The number is rarely alarming on its own. It is alarming compared to the number you get after applying the test above, which is often five or ten times smaller for no loss of information you were using.

A note on why this is visible at all

You can only reason about any of this if you can see what tracking costs separately from everything else. In a bundled tool you cannot. Tracked keywords come out of a plan allowance, so the cost of a bloated list is a limit you eventually bump into rather than a number you can weigh.

PandaCrawl runs on your own DataForSEO account, so rank checks appear as their own line in your month-to-date spend. That is not a feature so much as a consequence — but it is the thing that turns "I should tidy that up sometime" into a decision with a number attached to it.