A product signal is evidence about what users actually did, as distinct from a view about what they will do. The useful question in a planning meeting is not whether a claim sounds right, but what kind of signal is behind it - and most claims turn out to have none.
The evidence ladder, weakest to strongest
Not all evidence is equal, and knowing the order is most of the skill.
- Someone's confident opinion
- Zero signal, and the most persuasive thing in the room. This is the default input to most roadmaps.
- One customer asked for it
- Weak but real. Worth writing down, not worth building on its own - the loudest customer is not the average one.
- Several customers asked unprompted
- Genuinely useful, especially if they described the problem rather than requesting a solution.
- Support and error volume
- Strong, because it counts things that happened rather than things people said. A bug with 400 occurrences is a fact.
- Behaviour in the product
- Strongest available without a test - drop-off points, features nobody opens, paths people take instead.
- A controlled test
- Strongest, and usually out of reach at small traffic. See A/B testing.
Why the loudest voice wins by default
Because checking costs a tangent nobody in the meeting wants to take. A specific-sounding claim with nothing behind it beats a careful one with a source, simply because looking it up would derail the agenda. The fix is not more discipline in the meeting - it is having the number already on the page when the claim gets made.
Date every number
A figure without a date is not a signal, it is folklore. "Customers keep hitting this" was true in March. Whether it is true now is a different question, and one nobody can answer if the number arrived without a timestamp. Any evidence worth quoting is worth quoting with the window it covers.
Frequently asked questions
How much evidence is enough to build something?
It depends on what the mistake costs. A two-day change needs far less than a two-month one. The useful habit is matching the evidence you gather to the size of the bet, rather than applying one bar to everything.
Is qualitative feedback worth anything?
Yes, and often more than a weak quantitative signal. Five recorded conversations will tell you why people drop off; a dashboard only tells you that they do.
How Spectr helps
Prioritise with numbers instead of whoever sounds most certain
Real error and feedback numbers from PostHog and Sentry, quoted with a date in your specs
Related terms
Spend less of the week on documents
Spectr drafts the spec, the requirement document and the stories from your meetings. You review and approve.
Free plan available. 30-day free trial on Starter.