Doily tracks 9 AI engines because they give different answers, cite different sources, and are used by different customers. It checks paid plans daily because answers change from run to run. It labels DeepSeek results approximate because of how Doily runs that engine. This post explains each decision so you can judge the numbers for yourself.
The 9 engines
In order, they are ChatGPT, Google AI Overviews, Google AI Mode, Gemini, Perplexity, Claude, Copilot, Grok, and DeepSeek. Doily collects the answers through official APIs and licensed data providers. For each engine and prompt it records whether your brand was mentioned, whether a page of yours was cited, which competitors appeared, and which sources were cited.
We list the engines as separate rows because they behave as separate systems. A report that says “visible in AI” without saying where is not much use to a business trying to decide what to fix.
Why Google appears twice
Google AI Overviews and Google AI Mode are 2 different features. Google’s own documentation says they “may use different models and techniques, so the set of responses and links they show will vary.” Outside data agrees. Ahrefs compared 540,000 pairs of AI Mode and AI Overview responses to the same queries, using U.S. data from September 2025, and found they cited the same URLs only 13.7% of the time, while the meaning of the answers was 86% similar.
If Doily folded them into one score, you could be cited in one and absent in the other and never know. Keeping them apart means you see which feature is the problem.
Why not just a few engines
Not every business needs all 9. Which ones matter depends on where your customers ask. We do not claim to know that for you. We would rather you could check, so the full list is there and you choose how much of it to run.
That is also why the plans cover engines differently:
| Plan | What it checks |
|---|---|
| Free | 3 prompts weekly on ChatGPT and Google AI Overviews |
| Starter | 30 prompts daily across 5 engines, or 75 prompts on ChatGPT and Google AI Overviews |
| Team | 40 prompts daily on 5 engines, and all 9 engines weekly |
Credits meter the checks. A check on ChatGPT, Google AI Overviews, Google AI Mode, Gemini, Copilot, or DeepSeek uses 1 credit. A check on Perplexity, Grok, or Claude uses 2. Caps are hard, and there are no surprise overages.
Why DeepSeek is labeled approximate
DeepSeek runs in Doily as open weights on a U.S. cloud, grounded with Google results. That is a different setup from asking DeepSeek’s own service. The engine is real, but the way it is run means its answers may not match what a person gets in DeepSeek’s own app.
So Doily marks DeepSeek results as approximate wherever they appear. Treat them as a direction, not a precise reading. If DeepSeek shows you in most answers, that is a good sign. If it shows a small change from last week, do not build a plan on it. The label is there so a number never looks more certain than it is.
Other engines do not carry the label, but the same caution on single runs applies to all of them, as the next section explains.
Why daily checks
SparkToro and Gumshoe.ai had volunteers run the same brand-recommendation prompts many times across ChatGPT, Claude, and Google’s AI features. They reported less than a 1 in 100 chance that the same list of brands came back twice for ChatGPT or Google’s AI. Ahrefs found that Google AI Overviews changed in 70% of consecutive checks in its sample. Ahrefs sells AI visibility tools, and the SparkToro study was run with Gumshoe.ai, which sells AI tracking. Both describe their methods.
What they imply for tracking is plain. One check is one draw. To see how visible you are, you need many draws over time, and the more often you check the faster the sample grows. That is why paid plans check prompts daily. Free checks weekly, which is slower to build a picture but enough to see whether you appear at all.
Per-engine cadence is adjustable, and the product prints the cadence next to every engine’s numbers. A figure checked daily and a figure checked weekly should not be read the same way, so the label stays attached.
What you see for each engine
For every engine and prompt, Doily reports mentions, citations, share of voice versus competitors, sentiment, the sources cited, and the other brands named in the answer. You can set the country or city a prompt is checked for, because local answers can differ from national ones.
The cited sources are often the most useful column. They show which pages the engines are leaning on for your topic, including pages that are not yours. If the same directory or review site is cited in every engine, that is a clue to where your effort should go next.
Rankings are tracked separately
AI answers and Google rankings are different measurements, and Doily keeps them apart. Rankings are checked daily on Google, on desktop and mobile, by country, state, or city. For each keyword Doily notes the SERP features on the page, including whether an AI Overview shows. That pairing lets you see a keyword where you rank well and an AI Overview sits above you, which is a different situation from a keyword with no Overview at all.
What this means when you read the numbers
- Look at rates, not single answers. Appearing in 6 of 10 checks says more than appearing today.
- Read each engine separately. A blended score can hide a gap of 40 points between engines.
- Read DeepSeek as approximate. Use it for direction.
- Look at the cadence label. Compare like with like.
- Use the cited sources. The pages engines credit tell you what to improve.
What we do not claim
Doily does not claim to know why an engine chose a source, and no outside tool does either. Google says no third-party tool has access to its internal ranking or AI systems. What Doily does is record what the engines say, as often as your plan allows, and show it per engine, per prompt, and by country or city, so you can see what changes and where.
For setup, our guides on what AI visibility is and tracking AI prompts walk through choosing engines and prompts. The guide on reading share of voice explains the comparison against competitors.
