Which answer engines do you track, and why those?
The answer engines Appear covers, why they can give you different results, and how to use the numbers for each engine.
We track ChatGPT, Claude, Gemini and Perplexity, because your buyers do not all use the same answer engine.
A business can be cited on one of them and not appear at all on another, for the same question. Older report weeks only show the engines measured at the time.
Why they do not agree with each other
Answer engines build answers in different ways, so one can name you while another does not.
| Engine type | How it answers | How quickly your changes can show up |
|---|---|---|
| Live search | Runs a web search when asked and reads the results | Fastest |
| Model knowledge | Answers from what the model absorbed in training | Slowest, on the model's schedule |
| Mixed | Decides per question whether to search | In between, and varies by question |
Most major answer engines now decide per question. A question about current pricing may trigger a search. A question about what a category means may not.
This is why one engine is a poor guide to whether the work is landing.
Why we do not treat every engine as a separate project
The main work is the same for each one: be readable, be specific, answer the question directly and be corroborated elsewhere. That helps across engines because they read the same web.
Be sceptical of anyone selling a separate optimisation package for each engine. What differs most is timing.
How to use the numbers for each engine
Open a Query to see the evidence. Queries show the answer evidence for each engine, including whether the answer is Cited.
Use Competitors for scores. Discovery is shown per engine and reporting week.
Look for a sustained pattern. If one engine moves against the others over a quarter, that is worth raising. If it moves for one week, that may be normal variation.
Do not optimise for the engine you personally use. That is a sample of one, and it skews priorities.
What these numbers do not tell you
They do not cover every answer engine your buyers might use. They also do not tell you why one engine changed its answer, although the evidence can give you a clue.