What a signal is, and is not
A competitor posts five videos about a new service in a week. That is an observation: it has a source, a date and a count. "They are pivoting into our market" is an interpretation. It might be right, but it is a second step, and treating the two as one is how teams end up reacting to noise.
Write the observation down first, in plain terms, with its source and date. Only then write what you think it means, and how sure you are.
Five questions before you act
- Where did it come from? A post, an ad library record, a review, a search result. Each says something different.
- When? A spike from last week and a pattern over three months are different things.
- How many observations? One post is an experiment as often as it is a strategy.
- Compared to what? Their own usual activity, and the rest of the category. A jump that is normal for everyone is not news about one competitor.
- What would make it false? If you cannot name anything, you are holding a belief, not reading a signal.
Common traps
| What you see | What it does not prove |
|---|---|
| An ad in a public ad library | How much they spend, or whether it works. A library shows that an ad ran. |
| A post with unusual engagement | Sales. Engagement measures attention on that post. |
| A rise in search interest | Market share. It shows what people looked for, not what they bought. |
| Showing up in an AI answer for a query | A ranking everywhere. It holds for that query, region and date. |
| A run of angry reviews | A trend, until you know how many reviews they usually get, and when. |
From signal to decision
The useful output of a signal is rarely a big move. It is a small, cheap test with a date on it.
- Write the hypothesis in one sentence: "Their referral offer is pulling smaller customers away from us."
- Name what would confirm it and what would refute it.
- Choose the smallest test that would tell you: a question in your next five sales calls, one message variant, one week of a matching offer to a small segment.
- Give it an owner and a review date.
Observation: Brightline Studio (brightline.example) published four posts about a referral discount between 2 and 9 October; the category's other accounts published none.
Interpretation, held loosely: they are testing referrals as a cheaper way to win small customers.
Test: ask the next five lost prospects how they heard about the provider they chose. Review on 23 October.
