Competitive keyword research with AI is no longer a task that requires a four-figure annual subscription to an SEO platform. At Choco Media, we run the same competitive keyword gap analysis that used to take us hours and an Ahrefs seat — now with a combination of ChatGPT, Perplexity, and Google Search Console. The results are not identical to what you get from a dedicated tool, but they are close enough to drive real content decisions, and they cost nothing beyond what you are already paying for AI access.
This post is for marketing teams and small agencies who want to understand what keywords their competitors are ranking for, where the gaps are in their own content, and which topics are worth pursuing next — without committing to €300 per month in software. If you already have Ahrefs or Semrush and are happy with them, this is not an argument to cancel your subscription. But if you are looking for a lean competitive keyword workflow that produces actionable output, here is what we actually use.
We will walk through each step: how to identify the right competitors, how to use AI to surface keyword themes and gaps, how to cross-reference with free data, and how to turn the findings into a content brief your team can act on. The competitive keyword research ai workflow described here takes about two hours the first time and around 45 minutes once you have run it a few times.
Step 1 — Define the competitive set correctly
The first mistake most teams make is using the wrong competitors. Your business competitors and your content competitors are often different. A brand selling project management software might compete with Monday.com in sales conversations but with Notion’s blog in search — because Notion publishes extensively on productivity topics that overlap with their audience.
For keyword research purposes, we want content competitors: sites that consistently appear in search results for the queries your target audience is asking. There are two ways to find them.
- Manual search: Take five to ten of your core topics and search them in Google. Note which sites appear repeatedly across multiple queries. These are your content competitors.
- AI-assisted discovery: Ask ChatGPT or Perplexity: “Which websites publish the most authoritative content about [your topic area] for [your target audience]?” Cross-reference the suggestions with your manual searches. AI is not perfectly reliable here, but it surfaces sites you may have missed.
Aim for three to five content competitors. More than that and the analysis becomes unwieldy. We typically pick two direct product competitors and two or three content-only competitors whose audience overlaps with ours.
A note on AI hallucination risk at this stage
When you ask an AI model to name competitors, it can invent sites that do not exist or misrepresent the actual keyword focus of a domain. Always verify the suggested domains by visiting them and checking whether they publish content your audience would actually read. This is not optional — we have had ChatGPT confidently name sites that were either defunct or completely off-topic.
Step 2 — Pull surface-level keyword data from Perplexity
Perplexity’s search-grounded responses are useful here because it can reason about what a site covers without you having to crawl it manually. The prompt structure we use is:
“Analyse the blog and resource content on [competitor domain]. What are the main keyword themes they appear to target? List the topics by cluster, including example titles or question formats they address.”
Run this for each of your three to five competitors. You will get a thematic map rather than exact keyword data — Perplexity is not a keyword database — but it gives you a fast read on what territory each competitor has staked out. Copy the output into a shared document or spreadsheet, organised by competitor.
Then run a second prompt: “What keyword topics relevant to [your topic area] do none of these competitors appear to cover well?” Feed in the competitor list. This is the AI gap analysis step — it is imprecise, but it consistently surfaces two or three angles we had not considered, and it takes three minutes.
Step 3 — Use Google Search Console for your own baseline
Before you can identify gaps, you need to know where you already have traction. Google Search Console is free and shows you which queries are returning impressions for your site — including queries where you are ranking on pages two or three, which are often the easiest wins.
- Open Search Console and go to the Performance report.
- Filter by page to find your top-performing pages, then drill into the queries driving impressions for each.
- Export the data as a CSV — we typically pull three months of data to smooth out weekly variation.
- In a second export, filter for queries where you have impressions but position is greater than 10 (i.e., you are not on page one). These are your “position 11 to 30” opportunities.
Once you have your own data, you can contrast it with the thematic map you built from Perplexity. Anything your competitors cover that your Search Console data does not reflect is a candidate for the gap list.
Combining Search Console data with AI analysis
Paste your top 20 to 30 Search Console queries into ChatGPT and ask: “Based on these queries, what related keyword clusters am I likely missing? What questions would someone searching these terms also want answered?” This is a prompting pattern we return to repeatedly — using your own data as the context for AI gap identification. The model reasons about semantic relationships rather than pulling from a keyword database, but for content strategy purposes, it is often enough.
Step 4 — Run a content gap audit with ChatGPT
Now you have: a thematic map of what each competitor covers, your own Search Console baseline, and an initial list of potential gaps. The next step is to systematise this into a content gap audit.
We use a simple three-column format in a spreadsheet:
- Topic / keyword theme — a short description of the topic or question cluster
- Competitor coverage — which of your competitors appear to cover this (from the Perplexity analysis)
- Your current coverage — whether you have published content on this, and how it appears to rank (from Search Console)
Paste your combined findings into ChatGPT and ask it to help you populate this table. The prompt: “Here are the keyword themes my competitors cover, and here is what my Search Console data shows I already rank for. Help me create a content gap table identifying where competitors have coverage that I don’t.” Provide the raw data as context. The model will organise it faster than doing it manually and will often surface overlaps and groupings you missed.
This is one of the areas where our SEO work has shifted most noticeably in the past year — the analytical layer that used to require an Ahrefs export can now be approximated with AI-assisted synthesis of simpler inputs.
Step 5 — Validate gap topics with Google’s own free tools
AI analysis tells you what might be worth pursuing. Validation confirms it. Before you commit to writing a piece on a gap topic, run two quick checks.
- Google Autocomplete: Type the topic or question into Google and note what the autocomplete and “People Also Ask” sections return. If the autocomplete returns relevant variations, there is search demand. If it returns nothing, the query may be too niche or phrased incorrectly.
- Google Trends: Free, often underused. Run your gap topic through Google Trends to see whether interest is growing, flat, or declining. A declining topic is not necessarily a bad target, but it changes the priority calculation.
- Search Console again: After identifying gap topics, search them in Google and note whether any of your existing pages appear — even on page three or four. If they do, updating and expanding the existing page is almost always faster than publishing a new one.
When to trust AI analysis and when to verify harder
AI-assisted keyword analysis is best at identifying thematic gaps and organising ideas. It is not reliable for specific search volume numbers. When a model tells you a topic gets “high search volume” or “thousands of monthly searches,” treat that as a hypothesis, not a fact. Verify with Google Trends or a single free search in a tool like Ubersuggest’s basic tier before making production decisions.
Step 6 — Prioritise the gap list
A gap list without prioritisation is just a long to-do. We score each gap topic on three dimensions, each rated one to three:
- Relevance: How closely does this topic connect to our core services and audience? A one is tangential; a three is squarely in our wheelhouse.
- Validation signal: How strong is the evidence of actual search demand? A one means autocomplete returns nothing; a three means multiple autocomplete suggestions, a “People Also Ask” box, and a positive Google Trends trend.
- Content effort: Scored inversely — a three means we can cover this well in a single focused post; a one means the topic would need a pillar-scale treatment to compete.
Multiply the three scores and sort descending. The top items on the list are your next content priorities. This is deliberately simple — we have tried more sophisticated scoring models and they add friction without meaningfully improving the output.
Step 7 — Brief the content from the gap analysis
A gap analysis only creates value when it produces published content. The final step is turning each prioritised gap topic into a content brief your team can execute.
We use a lightweight brief format: target keyword, audience intent (what are they trying to figure out or do), competitor angle (what is the existing answer in the market), our angle (how we cover it differently or better), internal link targets, and a rough word count. This typically takes 15 to 20 minutes per topic with AI assistance — you paste the gap topic and what you learned about competitor coverage, and ask the model to draft a brief structure.
If you want a deeper look at the structural patterns that make AI-assisted content perform in both search and AI answers, our post on AI SEO in 2026 covers the ranking signals that matter across both surfaces.
What this workflow does not replace
We run this workflow for our own content and for clients who are early-stage or working with limited budgets. It does not replace a full keyword database for sites with significant existing authority where incremental gains from precise long-tail targeting matter. For those situations, a tool like Ahrefs or Semrush genuinely earns its cost. But for the majority of small and mid-size businesses doing content marketing — including most of the clients we work with — this workflow surfaces enough actionable insight to fill a content calendar for the next quarter.
How to keep this workflow running without it becoming a project
The risk with any research workflow is that it happens once and then never again. Competitive keyword research is most valuable when it is a recurring habit, not a one-time audit.
- Schedule a quarterly gap review — two hours every three months is enough to update the competitor map, check Search Console for new impressions, and refresh the priority list.
- When you publish a new post, run a quick Perplexity check: “What related questions does this topic typically lead people to ask next?” Add any gaps this reveals directly to the brief queue rather than a separate backlog.
- Set a Google Alert for your top two competitors. When they publish something new, note the topic. If it is in territory you have not covered, add it to the next quarterly review.
The tools change, but the habit of regularly asking “what are our competitors covering that we are not?” is what keeps content strategy from going stale. AI makes running that habit faster. It does not make the habit optional.
A realistic picture of what AI can and cannot do here
We want to be direct about the limits of this approach. AI-assisted competitive keyword research with free tools gives you a thematic map, not an exact traffic picture. You will not know whether a competitor’s post gets 200 or 20,000 monthly visits. You will not get precise search volume for long-tail queries. And AI analysis introduces hallucination risk that requires verification at every step.
What you will get is a fast, low-cost view of the content landscape that is accurate enough to make good content prioritisation decisions — which is what most teams actually need. In our experience, the gap between “AI-assisted with free tools” and “full Ahrefs account” in terms of content quality and strategic direction is smaller than the price difference suggests.
If you are curious about how we structure the full content and SEO workflow for clients — from gap analysis through to publishing and measurement — our AI content creation service covers the end-to-end approach. And if you want to explore whether a structured approach makes sense for your situation, we are straightforward about fit: start here.