Market Signals and Business Decision Making

The most common failure I see in market intelligence programs is not missing data — it's unassigned data. Hundreds of signals are monitored, beautifully rendered in dashboards, and owned by no one. A signal without a decision attached is decoration. This post is the taxonomy I use to sort decoration from instrumentation.

Three tiers of signal relevance

1. Decision-critical — the signal feeds a specific, dated decision with a named owner. Example: sustained competitor undercut ahead of your annual pricing review. These get alerts, thresholds and SLAs.
2. Context-shaping — the signal changes how leadership reads the next quarter but isn't tied to a single decision. Example: category-wide demand language shift. These get briefs and trend annotations.
3. Archive — the signal is worth recording but rarely worth interrupting anyone for. Example: a minor competitor's A/B test on a landing page. These get stored, indexed and left alone.

The classification is done *with* the owning team, not for them — which is also how you discover that half your current dashboard belongs to tier three.

Thresholds are negotiations, not calculations

Statistical significance and business significance are different quantities. A 2% price movement is noise for a commodity category and a fire alarm for a margin-thin one. We set thresholds in workshops where the analyst brings the distribution and the decision owner brings the stakes; the number that emerges is a commitment — "if this fires, I will act" — not a z-score.

Escalation paths make monitoring real

When a decision-critical signal fires, three questions must have pre-agreed answers: who is told, in what channel, with what decision deadline. Without this wiring, even perfect monitoring degrades into "interesting, anyway back to work." We write the escalation paths into the program documentation and rehearse them with dry-run alerts. It feels ceremonial. It works.

The review loop

Every quarter, audit the tiers: which alerts led to actions, which actions moved numbers, which signals should be promoted or demoted. Programs that run this loop get sharper annually; programs that don't accumulate tiers of zombie dashboards. The loop is the product.

The value of a signal is the quality of the decision it improved. Everything else is telemetry.

— Maya Chen

If you're building or rescuing a program, our market intelligence practice is structured around exactly this decision-first design — and the competitive intelligence overview covers the monitoring layer that feeds it.