About the Lytms Market Score
A single 0–100 reading of how strongly the market steers a buyer toward a product, in a category, today.
Nobody can buy a position on this index.
That is a statement about how the score is built, not a promise about how we behave. The two signals a vendor could actually move with money — how many reviews a product has collected, and how often it gets mentioned in communities — are not collected here at all. Not gathered, not stored, not weighed. A signal we do not hold cannot quietly start counting later.
This matters because that is the layer most software rankings are built on, and it is the layer with a price. Reviews can be incentivised, campaigned for, and bought outright. Mentions can be seeded. Any index that lets those signals decide a position is, in effect, selling one — whether or not it means to.
The rest follows from the same rule. There is no paid placement on any Lytms surface. Claiming a listing lets a company correct facts about itself; it never changes a score. Customers, prospects and companies who have never heard of us are measured identically, by the same code, on the same day.
What decides a position
Four signals, chosen because a vendor cannot buy any of them directly. Each is a measurement of what the market does, not of what a company says — and one of them is simply whether that is changing.
- AI answers
- What AI engines answer when buyers ask what to use for a job.
- Search results
- Where the product stands in the search results buyers actually see for the category’s buying queries — including who owns the answer box.
- Buyer demand
- How many buyers search for the product by name.
- Movement
- Whether the product is gaining or losing ground against the others in its market, measured between dated captures rather than claimed.
What we deliberately do not measure
Review counts and community mentions decide nothing here, and we no longer collect them at all. Both are purchasable — reviews can be solicited, mentions can be seeded — and an index built on them ranks whoever spent the most on looking recommended. Carrying them as unweighted context was the earlier answer; removing them outright is the honest one, because a signal we hold cannot quietly start counting later.
What we publish, and what we hold back
Published: which signals we measured for a product, which we could not, the plain fact behind each one, and the date we measured it. Where an AI engine named a product, we quote what it said, name the engine, and date the quote — never a paraphrase, and never our summary of it.
Held back: how the signals combine, and exactly what we ask. Not to be mysterious — because a published recipe is a specification for gaming it, and a market score that can be optimised against stops measuring the market. The evidence is the check on us: every number on a board travels with the plain fact it came from, and a claim you cannot verify against your own experience of the market is a claim you should discount.
What the score does not claim
It is not a judgment of product quality. We do not test the software, and we do not predict your results with any tool. A high score means the market is steering buyers toward a product right now — which is a different question from whether it is right for you. It reads the market; you make the call.
It is also not independent of company size. A larger company is cited by AI engines more often, ranks better on the queries buyers use, and is searched for by name more — partly because it is larger. Three of the four signals carry some of that, so a high score reflects the attention a product commands today, and being established is part of what produces attention.
Two things bound that, and one of them is not working yet. We never measure size directly: no employee counts, no revenue estimates, no domain-authority scores — the components that turn “big” into “good” outright are not in the recipe at all.And the largest single weight sits on change rather than level, because who is gaining ground is the one question incumbency cannot answer. That signal needs two comparable dated captures of the same product, and it is not reading on these boards yet — every product page says so where its line would be.
Scope and coverage
Scores are measured for the US region, in English. A product whose signals we have not measured enough of is listed under “Tracked, not yet scored” rather than given a number — an unscored product is a gap in our measurement, not a judgment about the product. A category does not publish at all until enough of its products are scored: a board too thin to represent its market is a sample, and we would rather show nothing.
Corrections & disputes
See a factual error? Write to us. A human reads every dispute, and corrections ship with the next update. Correcting a fact can change what a board says about a company; it does not change what the company paid, because the answer is always nothing.
Who compiles this, and what has changed
The Lytms Index is compiled and maintained by Yash Agrawal, founder, Lytms AI. The boards are computed; the method, and every correction to it, are his. What has changed →
The changelog carries every change to what the Index publishes, dated — including the corrections, which are most of it. A ranking you cannot check is a ranking you should discount, and publishing what we got wrong, beside what we say now, is how this one is checkable.