1. A before cannot be taken afterwards
This is the whole argument and the other four are details.
Once the site is redesigned, the price is moved or the campaign is live, the previous state is gone. You can reconstruct fragments from your own analytics, but not the competitor's side of it, and not the measures nobody was recording. So the question is never whether a baseline is worth the effort. It is whether you would like to have one at the point where somebody asks whether the change worked.
Say it plainly when you take it, too. A first measurement is a first measurement, not a report that nothing has changed. Presenting an opening reading as though it were a comparison is one of the easier ways to make an entire measurement system untrustworthy in its first month.
2. It tells you which fights are worth picking
Knowing that a competitor is four times your size changes what you should even attempt. Knowing that most of their traffic is direct changes it again: that is a brand you cannot outbid, only outlast.
Without the comparison you end up planning against an imagined rival, usually one that is either terrifying or negligible and rarely the real one. Scale first, then the mix, then the plan. In that order, most roadmaps get shorter.
3. It separates your move from the season
Your traffic fell eleven percent. Did you break something, or did the category fall?
With a benchmark that question has an answer, because the competitor set is a control group you did not have to construct. Everybody down eleven means the month, not you. You down eleven while they hold means something you did. Same number, two opposite decisions, and no way to tell them apart from your own analytics alone.
This is the reason a benchmark is worth more the longer it runs. The first reading is a position. The twentieth is a line with somebody else's line beside it, which is the only shape in which a result can be honestly attributed.
4. It sets the order of the work
Any comparison of two brands produces a list of gaps. The useful step is sorting them by width against cost.
A gap that is wide and cheap to close is this quarter's work: a comparison page nobody wrote, a review count that never gets asked for, a market with no localised version. A gap that is wide and expensive is a strategy decision, not a task. A narrow gap is a distraction however annoying it is.
Doing this before you change anything is what stops the roadmap from being a list of whatever was discussed most recently. The reference configuration we run carries 176 structural measures, and the point of that number is not completeness. It is that the gaps get found by counting rather than by whoever spoke last.
5. It settles the argument before anyone has a stake in it
Agreeing on the measure before the work starts is a different conversation from agreeing on it after the results are in.
Beforehand, nobody owns the outcome and the discussion is about what would count as success. Afterwards, everyone owns something, and the discussion becomes a search for the metric that makes the last quarter look intentional. With any reasonably wide set of numbers, something is always up and something is always down, so an unsettled measure is an invitation to tell whichever story is needed.
Write down the measure, the target and the date, before the change. It takes ten minutes and it removes a class of argument permanently.
What a benchmark is not
It is not a scoreboard for morale, and it is not a weekly number. Benchmarked measures are lagging by nature: they move slowly, and reading them every Monday will teach you nothing except how to overreact. Read the inputs weekly and the benchmark quarterly, which is the split that keeps both of them honest.
It is also not a place for numbers nobody would act on. If a measure would not change a decision in either direction, it is taking up room that a useful one could have.
If you want the practical version, the step by step is here, and the first one against a competitor of your choice costs nothing.
Pick one competitor. See what we find.
Your first benchmark is free. One domain, measured against you, no card and no call.