How this site works

marketaiguide.com is a pure AEO litmus test, published by DevOpser LLC. Each list is what an AI assistant says when a buyer asks it for the best products in a category. We publish the answer, the rubric behind it, and how it changes week to week. You can reach a person at hello@devopser.io.

What AEO means here

Answer engine optimisation is the work of being named, and described well, when someone asks an AI assistant instead of a search engine. This site measures the outcome of that work directly: does the model know your product, and how highly does it rate it?

How a list is made

• A rubric, written by the model. For each category, Claude is asked what a buyer should judge products on. It returns four to six weighted criteria. That rubric is frozen and published at the top of the list, so every week is scored the same way.
• A question, with nothing else. The model is asked to name the products it would most recommend and to score each one from 1 to 10 on every criterion, with a reason for every score. It has no web access, and nothing from this site, no vendor list and no submissions go into the prompt. The exact prompt is printed on every list.
• Arithmetic, not judgement. The rank is the weighted average of the model's scores. We do not reorder, add, remove or edit entries. The one check we run is that each named product has a real website, so an invented product never gets a page.
• Every week, again. The same question is asked weekly and every run is kept. Each list shows the movement since last week, a trend line, and who dropped off.

What it measures, and what it does not

It measures how an AI model already sees a brand: whether the model knows it, which category it files it under, and what it believes about it. That is what a buyer hears when they ask an assistant.

It does not measure which product is best, and a model can be wrong or out of date. Treat every score as evidence of perception, not as a product review. Check anything that matters with the vendor.

Week-to-week movement mixes real change with noise. The prompt runs at temperature 0, but models are not perfectly repeatable, and a new model version can reshuffle a list overnight. The trend across several weeks is the signal; a single week is not.

Which model

Claude, through Amazon Bedrock. Each list names the exact model that ranked it. One model is one view of the market; other assistants may answer differently.

The conflict of interest, up front

DevOpser LLC also makes DevOpser. If the model names DevOpser on a list, it is marked "our tool" and gets a disclosure, and it is ranked by the same prompt and rubric as everything else. If the model does not name it, it is not there.

We do not describe this site as independent, because it is not.

How the site makes money

Today it does not. Nobody pays to appear here and nobody can pay for a better position, because we do not set positions. Every outbound link to a vendor is a plain link.

Improving your score

We cannot change your rank, and nobody who claims to can either. What you can change is what the model learns about you. How to improve your AEO explains what moves it.

Corrections

If the model states something wrong about your product, the fix is at the source the model learned it from, and the guidance page explains how. If the site itself gets something wrong, tell us at hello@devopser.io.