Confirmation across three sources: search × ads × Amazon

Any data series, taken alone, knows how to lie. Searches also rise on a scandal, not only on a buying wave. Ads also multiply when a single big player tests aggressively, not only when the market moves. And Amazon has seasonality, promotions and ranking artefacts.

That's why the mature question isn't "is this ingredient rising?" — it's "on how many independent sources is it rising at the same time?"

What you'll be able to do after this chapter

understand what each of the three sources measures and what it can't see alone;
use source agreement as a gate: what moves on towards the formula and what stays under observation;
avoid the concentration trap — the "growth" produced by a single brand.

The principle

The three sources look at the same reality from different angles:

Searches (weekly series per market) measure consumer curiosity — the earliest signal, but also the noisiest;
Ads (the launch rhythm) measure advertiser conviction — other people's money bet on the theme;
Amazon (review velocity) is the proxy closest to real purchases — the latest of the three, but also the hardest to move without sales behind it.

Each can rise alone for reasons that don't help you. When two or three rise together, the alternative explanations fall one by one — and that's exactly what RavenBI's trend engine rewards: the number of confirming sources (source count) amplifies the score, and disagreement tempers it. An ingredient "exploding" on a single source and one confirmed on three are not the same kind of information, even if the percentages look alike.

The workflow in RavenBI

On the ingredient page from chapter 4, don't stop at the search curve: put it next to the ad pressure (already on the page) and next to the theme's Amazon signals (chapter 7 gives you the full lens).
In Explore, watch for the discoveries that explicitly mention multi-source confirmation — the radar lifts them above single-source ones precisely because they're rarer and more solid.
For your candidate, check the disagreements manually: searches rising while ads have been silent for months? Ads exploding on flat searches? Every disagreement has an explanation — and the explanation is often worth more than the score.
Check concentration: open the theme's recent ads and see how many different brands stand behind them. Growth carried by a single brand isn't a market trend — it's one competitor's strategy.
Only what passes this gate moves on to chapter 6, where the ingredient becomes a formula.

How to read the signals

3 sources in agreement — the gold signal; rare, and precisely for that reason valuable. And the more visible it is, the shorter the window tends to be.
2 sources in agreement, the third flat — solid; the third source lags naturally (Amazon reacts last, searches first).
1 source — a hypothesis. It goes on the Watchlist and waits; you don't build a formula on it.
Sources in active disagreement (one rising, one falling) — not an average, but a message: someone is wrong, and your job is to find out who, before you bet.

The typical mistakes

treating confirmation as "bureaucracy" when the first source looks spectacular — that's exactly when a day's pause is cheapest;
ignoring the natural lag between sources and rejecting a good signal because Amazon "hasn't confirmed yet";
missing concentration — one brand's growth read as a market move;
redoing by hand what the radar does continuously — manual verification is for understanding; monitoring is the machine's job.

Exercise

For the candidate left from chapter 4, write three lines — searches, ads, Amazon — and note for each: the direction, a hint of magnitude, and what could explain the growth other than real demand. If on two out of three you found no plausible alternative explanation, the candidate has passed the gate.

A confirmed ingredient is a raw material, not a product. Next comes the question separating launches that can speak from those condemned to vague marketing: what formula do you build around it — and what are you legally allowed to promise about it?