A research programme does not go stale evenly. Some measures move every year. Some have not moved in four. Yet most tracking programmes re-ask everything, every wave, on the same calendar.
We recently worked through four years of repeat waves in one long-running programme, measure by measure, and netted every change against the sampling noise measured inside each wave. A surprising share of what looks like wave-to-wave movement turns out to be noise rather than change. Once that is taken out, the measures sort themselves into a ladder.
Every year
What people say drives the purchase decision was the fastest-moving measure we could demonstrate, several times faster than brand preference. Attitudinal batteries and satisfaction belong here too: satisfaction tracks the product that is actually shipping, so any line or supplier change resets it.
Every 18 months
Brand perception, functional and emotional, and the price-value layer. Fast enough to matter, slow enough that annual fielding is mostly re-confirming what you already know. Willingness to pay tracks the price environment rather than the calendar, so treat 18 months as the outer bound.
Every two years
Purchase intent against a fixed concept set (a new concept set is new data, not a refresh), repurchase intent, and channel and retailer preference, where a distribution change is the real trigger.
Every three years
The stable core: brand preference, aided awareness at rest, consideration, usage frequency, feature importance. This is where budgets are most often wasted, re-asking questions whose answers have not moved since the last time they were asked. One caution: price sensitivity looks stable right up until the category reprices, and then it is wrong overnight. Three years is its outer bound, not its plan.
Every five years, deliberately
The segmentation, the category and competitive frame, and the demographic frame. These are not refreshed because they decay. They are rebuilt because the market changed shape, and rebuilding them breaks comparability with everything that came before.
No clock at all
Ad and concept response is measured per campaign or not at all. And any question whose wording, scale or option list changed is not a refresh question, it is a comparability question: the trend breaks outright.
Two rules sit over the whole ladder.
The calendar is the floor, not the rule. A repricing, a competitor launch, a distribution change or a category shock resets the clock for anything it touches, regardless of when it was last measured. The largest movements we found were events, not drift, and no schedule would have caught them.
Advertising changes the ladder while it runs. Awareness and ad recall are among the most stable measures at rest and among the least stable during a campaign, so they move to a quarterly read while media is live.
What to do with it
Concentrate the fielding budget at the top of the ladder, where fresh human data earns its cost. Stop re-asking the stable core every wave. Budget for triggers, not just for waves, with a reserve for unscheduled fielding. And treat the frame as sacred: anything in the five-year band should change deliberately and be documented, because changing it silently invalidates every comparison that crosses it.
If you run a tracker and would like to know which of your questions are due and which you could stop paying for, City Research can map your own programme the same way.