Keyword Research in AI: A 6-Step Process for Finding Topics and Prompts Worth Targeting

Traditional keyword research starts with a query: a specific string of words with a monthly search volume you can look up. You pick the ones with enough volume and the right intent, and you write for them.

AI search does not work like that. When someone asks ChatGPT or Google's AI Mode a question, the system does not simply look up that exact wording. 

Google's own description of AI Mode says it breaks a question into subtopics and runs multiple searches at the same time before it writes an answer. So the thing you are really competing for is not one keyword. It is a topic, and the specific prompts that live inside that topic.

That changes what "keyword research" means. This guide walks through a six-step process, from finding a trending topic to ending up with a short, ranked list of prompts that are actually worth writing for. Each step gives you a specific number to act on, so you are not guessing.

Keyword Research in AI

Start by choosing a category and finding which topics inside it are gaining attention in AI answers. Check each topic's score together with its month-over-month trend, because the score alone can point you at the wrong topic. 

Pull the real prompts people ask about that topic, load them into a prompt tracker, and look at where competitors are already mentioned or cited, and you are not. Then focus on the prompts with a growing gap and revisit the list on a regular schedule.

Why Topics and Prompts Replace Keywords

A keyword is a fixed phrase that can be counted. A prompt is a full question or request, written however the person likes, and an AI system turns it into several related searches behind the scenes. Two people can word the same need very differently and still get answers built from the same pool of sources.

That is why the unit of research moves up a level. You research the topic (for example, "footwear advice"), then the representative prompts inside it (for example, how to convert between shoe sizes), and then you track those prompts over time. 

Search volume still has a place in your wider SEO work, but it counts a different behavior from AI answer attention, and the two can disagree. A topic can attract real attention in AI answers around prompts that show little or no measurable search-engine query volume.

The process below uses Similarweb's AI Search Intelligence tools, since that is what the source article walks through. The logic, which is to choose a topic, read its trend, collect real prompts, track them, and look for competitor gaps, applies with any tool that gives you similar data.

How to Run Keyword Research in AI

Step 1: Pick The Topic or Category to Focus on

Open the AI Research tool and start in a Trending Topics report, filtered to your category. The example in the source uses "Sporting Goods" inside Retail and Consumer Products.

Filter by category before you look at any score. The score in the next step is worked out relative to the top topic within whatever scope you select. If you leave the report on all industries, whichever topic is dominating AI conversation in general that month will bury the opportunities that matter in your category.

Step 2: Check the Metrics That Show Whether a Topic is Worth Pursuing

Every topic in the report has an AI Interest Score, an index from 0 to 100 showing how much attention the topic gets in AI-generated answers compared with the single most-discussed topic in that category during the chosen period. 

A topic scoring 100 drove the most AI answer volume. A topic scoring 50 drove about half as much.

It is a relative measure, not an absolute one, so never read it on its own. Pair it with the month-over-month (MoM) change and the shape of the trend line.

The source's sporting-goods example shows why. Look at three topics:

By score alone, Apparel Trends and Home Footwear look best. Read next to the trend, the picture flips. The two highest-scoring topics are cooling off slightly, while Footwear Advice, with a lower score, is one of the fastest-growing topics in the category. 

That is where new content has the best chance of being noticed while the topic is still forming. The score by itself would have pointed at the wrong topic.

Step 3: Pull the Representative Prompts For the Topic

Each topic can be expanded into a detail panel with a written summary of what people are actually asking and a short list of representative prompts. For Footwear Advice, those included questions about converting between shoe sizes, telling regular from wide-width models, and whether new shoes should feel tight at first.

These are not keywords. They are the real phrasing AI systems are handling, and they are the raw material for the next step.

Step 4: Add Those Prompts to a Prompt Tracker

Take the representative prompts and load them into a prompt tracker, so they are monitored for visibility, sentiment, and citations continuously rather than checked once.

There is a plain reason to do this. A topic score is a snapshot. A tracked prompt is a time series. You need the second one to know whether anything you publish is working.

Step 5: Check Competitor Gaps Across the Same Prompt List

This is the step that turns a topic score into a decision about where to invest. Once your prompts are running, a prompt-tracking view typically gives you four counts.

Start with the citation gap. Those are prompts where AI systems are already pointing to a competitor as the source, and you have no presence at all.

Then filter the mention gap view in two ways at once: "latest mention status: not mentioned" together with a competitor brand filter. With both sets of data, you can see where a specific competitor is ahead and decide what to focus on.

The source works through an example using adidas against competitors in sporting goods. Four lessons come out of it.

1. The gap can sit in a different place than you expect: The three highest-visibility rows in the citation-gap list were all service questions, such as returns and exchanges. 

In the unfiltered prompt list, the highest-visibility prompts were product questions, and adidas already had a recent mention on both. 

So the gap was not in product recommendations, where adidas already appears. It was in the service and policy layer.

2. Direction matters more than the number: Within the gap list, the period-over-period change moves both ways. The exchange-policy prompt was up 57.14% month over month, a growing gap. But the prompt about returning shoes worn briefly indoors had the same visibility score as the return-policy prompt and was heading down. 

A prompt about shoes for both gym lifting and treadmill runs was down more sharply, despite positive sentiment. Two prompts can show the same visibility and opposite trajectories. When two gap prompts are close, rank them by the change column, not the visibility column.

3. The best opportunity is not always the biggest one: The prompt "can you help me pick a size if I'm ordering athletic shoes online" had only 43% visibility, lower than the three service prompts above it. 

But it was rising 42.86% month over month and had positive sentiment. Moderate visibility, steep growth, and favorable sentiment together are a stronger signal than a larger gap that is shrinking.

4. Zero visibility is not the same as losing: Several prompts in the filtered view, such as cushioning without squishiness, breaking in new shoes without blisters, caring for suede, and replacing worn shoes, showed no measurable presence for any brand, no change over the period, and no sentiment recorded, because there was not yet an answer pattern to analyze. Nike still appeared as the top brand on every one of them, with others trailing. 

That does not mean a competitor is beating adidas on those prompts. It means visibility has not consolidated around anyone yet, adidas included. 

Treat them as open-field opportunities, not as a fight to take ground from a rival, and prioritize them differently from the return and sizing prompts, where there is an active, measurable gap against a specific competitor.

Step 6: Set a Review Cadence

Both an AI Interest Score and a citation gap count keep moving. Set a weekly or monthly review of the same topic and prompt list, and track the direction of change, not only the current position, for both month-over-month change and citation gap count over time.

A prompt whose citation gap shrinks after you have published against it is the nearest thing this workflow has to a success metric.

Running the Process Across Several Categories or Brands

Running the six steps on one topic gives you a one-off report. Running them as a standing process across every category you track is what compounds. Once the method works on one topic, scale it in two ways.

First, widen the topic net before you widen the prompt list. Pull the top 5 to 10 topics in each category you cover, sorted by month-over-month change rather than raw score. 

Carry a topic into Step 3 only once it clears a minimum trend threshold you set for your category, meaning a rising MoM change, not just a high starting score. That keeps the number of prompts entering your tracker in proportion to real momentum, not to how many categories you happen to be watching.

Second, if you track more than one brand or competitor set, such as a multi-brand portfolio or a category with a different set of competitors in each region, run the citation-gap filter from Step 5 separately for each brand and competitor pairing. 

A gap against one competitor set can look quite different from the gap against another, even on an identical prompt list, because different competitors win different prompts. 

Share the underlying prompt tracker across brands where topics overlap, but keep the competitor filter and the resulting priority list separate for each one.

Prioritizing When you are Tracking Several Topics

Once several topics are running, you will have more prompts with a citation gap than any content team can handle in one cycle. Rank across topics, not just within one, using three cuts layered together. First, MoM change: is the topic still gaining attention? 

Second, citation gap size: how much unclaimed ground is there? Third, whether closing the gap matters for your goals this quarter, which depends on whether the prompt is mainly a citation play with little click behavior or still carries real clicks.

A topic with a smaller citation gap but a steep upward trend will usually beat a topic with a bigger but flat gap. The smaller number is still growing, and closing it pays off in the metric you are actually trying to move.

Frequently Asked Questions

What is the Difference Between an AI Interest Score and Keyword Search Volume?

Search volume counts actual queries typed into a search engine. An AI Interest Score measures how much of the AI-generated answer volume in a category a topic takes up, relative to the top topic in that category and period. Both are demand signals, but they measure different behavior and often diverge on the same subject.

How Many Representative Prompts Should I Track Per Topic?

There is no fixed number. A sensible start is the full representative-prompt list a topic surfaces, which is typically three to a handful. Then expand based on which prompts show a citation gap against competitors. That keeps the tracked list focused on prompts with evidence behind them rather than every possible phrasing.

Is a High Citation Gap Always a Good Reason to Write Content?

No. Check how the prompt behaves. If it is a pure citation play with very little click behavior, prioritize being the cited source over traffic-focused formatting. If it still carries real click volume, treat it like any other high-intent opportunity.

How Often Should I re-run This Process?

Review the same tracked prompt list weekly or monthly rather than treating any single pull as final. Both AI Interest Score and citation gap counts shift as competitors publish and AI systems re-crawl.

What if a Topic Has a High AI Interest Score but no Prompts Worth Targeting?

Move on to the next topic in the category. Do not force content around a topic where the underlying prompts do not map to anything your brand can credibly answer, or where competitors have no measurable presence to catch up to.

Do I Need a Separate Prompt Tracker for Every Brand I Manage?

Not necessarily. Share the topic and prompt research where categories overlap, but keep the competitor citation-gap filter separate for each brand, because the competitive set and the resulting gap list are rarely the same.

Related Searches:

• AI Platform Personas: Why ChatGPT, Gemini, and Claude Attract Completely Different Users
• UGC for AEO: Why AI Cites Reviews and Reddit Over Brand Content, and How to Use It
• AEO for SaaS: How to Optimize Product Pages for AI Search Visibility