Our newsletterRanking methodology: S to D tiers
Every card here carries a tier badge, and most carry one or two plain warnings underneath. This page explains what decides them.
What becomes a card
We constantly search for the latest marketing intelligence: independent research, vendor reports, and insightful practitioner articles — all become sources for Stacy’s Brain. We record who published each piece, when the data was collected, who was studied, how, and whether the publisher has something to sell on the back of the finding.
We consciously exclude pieces reporting someone else’s study, and arrange cards so they carry either a trend (what should I know) or an idea (what should I do).
Each card carries one finding from one source. It shows, word for word, the sentence or chart label the number came from, plus a note explaining what the number actually measures.
Evidence as the central criteria
This is the largest single input. We sort every finding into one of eight kinds, from weakest to strongest:
- Stated intention — people were asked what they plan to do.
- Expert judgment — a practitioner or analyst gave an opinion.
- Modelled or projected — the number came out of a model or a forecast.
- Self-reported behaviour — people described what they did.
- Observed behaviour — someone measured what happened.
- Quasi-experimental — groups were compared without random assignment.
- Experimental — groups were assigned at random.
- Meta-analysis — many studies were pooled.
A survey about intentions and a randomised experiment can report the same percentage and mean very different things. The tier reflects that.
What else moves the ranking
Whether the sample is disclosed.
A finding that names who was studied and how many of them there were outranks one that does not.
Whether the method is disclosed.
If the source never explains how a number was calculated, the figure is not safe to quote on its own, and the card is rated accordingly.
How old the data is.
Most findings hold their value for a few years. Anything about AI search and AI citations does not: the platforms change in months, so we hold that topic to a one-year line and everything else to a longer one. Where the source never says when the data was collected, we fall back to its publication date and tell you we did.
Who paid for it.
We mark every source with the publisher's commercial interest. A vendor measuring its own customers is a normal and useful thing, and it sits at moderate. The mark goes to high only when the finding itself argues for the category the publisher sells — when the research and the sales pitch point the same way.
The warnings on cards
Alongside the tier, a card can carry one or more standing warnings. There are about fifteen of them.
Each is attached only when that specific weakness is present in that specific finding, and each is written for you rather than for us:
- People were asked what they plan to do. Nobody has checked what they actually did.
- The data shows two things moving together. Which one drives the other was never tested.
- The publisher has a commercial stake in this finding. Nobody neutral produced it.
- This was measured in one market. Other markets work differently.
- This is an average. Real results vary a lot between segments.
- A card with three warnings is not a bad card. It is a card you now know how to use.
What the rating does not tell you
The tier rates the evidence, not the idea.
A weak study can still be pointing at something real. A low tier means the proof is thin –nothing more.
It also rates a single finding, not the whole source. A careful report can still contain one shaky number, and a promotional one can still include a solid measurement. We rate the finding in front of you.
It is not advice. A high tier means the finding is well evidenced where it was measured. Whether it transfers to your market, your category, or your budget is yours to judge, and the warnings exist to help you make that call.
When we get it wrong
Building evidence for Stacy’s Brain is a multi-stage, semi-manual process that can’t be fully automated without compromising quality. Each insight goes through several rounds of human review.
Even so, mistakes can happen — and we’re working to minimize them. If you notice an incorrect card, please let us know.