How to Choose Which AI Engines to Track in AmICited

Different audiences live on different AI assistants, and picking the wrong ones to monitor means optimizing for platforms your buyers never actually use.

Quick Steps

• The AI providers checkboxes sit in the Add prompts dialog, below the prompt input; all five (ChatGPT, Perplexity, Gemini, Google AI Overview, Google AI Mode) are selected by default.
• Every prompt must keep at least one provider checked, and each checked provider runs as a separate, independently scored query.
• Start with the engines your actual audience uses; B2B buyers often lean on ChatGPT and Perplexity, consumer purchases more on Google’s AI Overview.
• Tracking all five gives the fullest picture and the most reliable AI Visibility Score, but costs more run quota, so trim deliberately for large prompt sets.
• Revisit your selection periodically, since which engines matter can shift as usage patterns change.

What are AI engines, and why does tracking the right ones matter?

An “AI engine” (also called an AI provider or an answer engine) is any AI system that takes a user’s question and generates a direct, synthesized answer instead of, or alongside, a list of blue links. ChatGPT, Perplexity, Google Gemini, Google AI Overview, and Google AI Mode are the five most widely used engines today, and each one draws on a different mix of training data, live web retrieval, and ranking logic to decide which brands, products, and sources it mentions. That’s why the same question asked on two different engines can produce two very different sets of cited brands, a phenomenon researchers call citation divergence, and it’s the reason single-engine tracking gives an incomplete picture of your AI search visibility.

This matters because these engines are increasingly where buying research starts. A shopper asking ChatGPT to recommend project management tools, or asking Perplexity to compare two SaaS vendors, is getting an answer synthesized from whatever sources that engine’s model considers authoritative, and if your brand isn’t among them, you’re invisible at the exact moment a prospect is forming their shortlist. Unlike traditional search engine rankings, where one algorithm (Google’s) dominates measurement, AI answers are fragmented across at least five systems with materially different retrieval methods: ChatGPT leans on a mix of training data and browsing, Perplexity is built around real-time web search with heavy citation of sources, Gemini draws on Google’s knowledge graph and search index, and Google’s AI Overview and AI Mode sit inside classic Google Search results but generate their own synthesized answers above the organic listings.

Because each engine can cite a different set of competitors for the same query, share of voice (the percentage of AI answers in which your brand appears relative to competitors) can swing dramatically depending on which engines you’re measuring. A brand that dominates Perplexity citations might barely register in Gemini’s answers, or vice versa. That’s precisely the gap the AI providers selector in AmICited is built to close: rather than assuming one engine represents your whole AI visibility, you choose exactly which platforms get queried for each set of prompts, so your dashard reflects where your actual audience is asking, not just where it’s easiest to measure.

Where to find it in AmICited

The AI providers row appears in the Add prompts dialog (+ New prompt), below the prompt input and above the Country/Tag/Schedule row. It’s available whichever input method you use (paste, CSV, or generate from URL), so the same provider logic applies no matter how you build out your prompt library.

The engines you can pick

Each provider is a checkbox you toggle on or off, and the choice you make here determines which engines actually run your prompt and get scored:

• ChatGPT: OpenAI’s assistant, the largest single AI user base and often the first engine brands want visibility into.
• Perplexity: an answer engine built around live web retrieval, known for showing its source list alongside every answer, which makes it one of the more transparent engines to audit for citations.
• Gemini: Google’s own conversational model, which draws on Google’s search index and knowledge graph in ways that can differ meaningfully from what ranks in classic organic search.
• Google AI Overview: the AI-generated summary that now appears above traditional results for many Google searches, drawing from a blend of top-ranking pages.
• Google AI Mode: Google’s more conversational, multi-turn search experience, which behaves closer to a chat interface than a traditional search results page.

The helper text sums up the rule: “Pick which AI platforms to query. All are selected by default; a prompt must keep at least one.”

What it measures, and how it affects your results

Every provider you leave checked becomes a separate run for that prompt. AmICited queries each selected engine independently, records whether and how your brand is cited, and rolls those results up into your prompt’s per-engine breakdown. This is the mechanism behind the platform’s core AI rank tracker : it isn’t tracking one abstract “AI ranking,” it’s tracking your citation position separately on each engine you’ve chosen, because a brand can rank first in one engine’s answer and not appear at all in another’s.

That per-engine granularity is also what makes your overall AI Visibility Score meaningful rather than misleading. If you track only ChatGPT, your score reflects only ChatGPT: a strong result there can mask the fact that you’re essentially invisible in Google AI Overview or Google AI Mode , two placements that sit directly inside the search results your buyers already use every day. Selecting a broader, deliberately-chosen set of engines up front means the score you look at each week is actually representative of your total AI footprint, not just the slice that was easiest to check.

How to choose which engines to track

1. Start with where your audience actually asks. If your buyers lean on ChatGPT and Perplexity for research, prioritize those two over engines your audience rarely touches. Every industry skews differently: B2B SaaS buyers often research heavily in ChatGPT and Perplexity, while consumer purchases are more likely to surface through Google’s AI Overview.
2. Keep the majors on for a baseline. Tracking all five gives the most complete picture of your share of voice and lets the per-engine breakdowns on your dashboard reveal gaps you wouldn’t otherwise see. It’s common to be strong on one engine and nearly invisible on another.
3. Trim to save runs if you’re tracking a very large prompt set and only care about specific engines. This is a practical tradeoff: broader coverage costs more of your run quota, so a large prompt library might justify narrowing coverage on lower-priority prompts while keeping full coverage on your highest-value ones.
4. Revisit the selection periodically. Engine usage shifts: a platform that was niche a year ago can become a meaningful referral source, and it’s worth re-checking your provider selection against current traffic and referral data rather than setting it once and forgetting it.

This selection isn’t limited to individual prompts, either. If you’re managing visibility across multiple brands or client accounts, a common setup for agencies , being deliberate about provider selection per client keeps run usage predictable while still giving each account accurate, engine-specific data instead of a generic average.

After you’ve chosen

Once prompts are live, use the All AI Models filter on the dashboard and the per-platform breakdowns to compare how you perform on each engine. This is usually where the most actionable insight in AmICited shows up: rather than one flat “you’re doing fine” number, you can see that you’re winning citations reliably in Perplexity, showing up inconsistently in ChatGPT, and missing entirely from Google’s AI Overview for a set of high-intent queries. That pattern tells you exactly where to focus content and technical work next, rather than optimizing blindly across every engine at once.

If you’re still building out the prompt list these providers will run against, it’s worth reading how to find which prompts trigger AI to mention your brand : the provider selection only pays off if the underlying prompts reflect real buyer questions. And if you want the fuller picture of why single-engine measurement consistently understates a brand’s actual reach, the deeper breakdown of how to track your brand across AI platforms walks through the multi-platform approach in more depth, including how to weight engines by the traffic and citations they’re actually driving back to your site.