The Monthly AI Citation Audit: 5 Steps to Check Whether ChatGPT and Perplexity Actually Recommend Your Brand

Your brand is either being recommended by ChatGPT, Perplexity, Claude, and Google AI Overviews right now — or it is not. The difference between guessing and knowing is a structured monthly audit that takes about two hours and gives you the only reliable measure of whether your marketing is actually working in AI search.

I have been running citation audits for months. Here is the exact process.

Why AI Citations Require a Monthly Audit

AI citations are volatile. Scrunch's analysis of 3.5 million citation events puts the median half-life of an AI citation between four and five weeks, with ChatGPT fastest at roughly 3.4 weeks and Perplexity slowest at roughly 5.8. Half the sources ChatGPT or Perplexity cites today will not be cited a month from now. Across identical prompts re-run on the same engine, roughly 1 in 9 of the top-10 cited sources changes each time.

Knecht Strategies puts it bluntly: 40-60% of your AI citations will turn over every month. A single snapshot tells you almost nothing. You need a baseline, a trend, and a diagnosis of what moved.

Without a monthly audit, you are operating blind in the channel where buyers increasingly start their research.

Step 1: Build Your Prompt Library

The foundation of any citation audit is a locked set of 20-30 buyer-intent prompts that represent how your customers actually ask ChatGPT or Perplexity for recommendations. These are not keyword lists — they are the natural-language questions your buyers type when evaluating vendors.

Start with three categories:

• Category queries: "What are the best [your category] tools?" or "Compare [your category] vendors"
• Problem queries: "How do I solve [problem your product addresses]?"
• Brand queries: "Tell me about [your brand]" and "Is [your brand] good for [use case]?"

The GEO Brand Citation Index tracks 36 brands across 3 verticals using 54 standardized queries — that scale is a useful reference for how deep your prompt library should go.

Lock the library. Run the same prompts every month. New prompts expand it — they never replace existing ones, because the trend on each prompt is the signal.

Step 2: Run Each Prompt Across Four AI Engines

Run every prompt across ChatGPT, Perplexity, Claude, and Google AI Overviews. Record two metrics per prompt per engine:

• Visibility: Was your brand mentioned anywhere in the response?
• Citation: Was your brand's URL linked as a source?

The distinction matters. Visibility without citation means the engine has heard of you but does not trust your content enough to link to it. Citation without visibility — rare but real — means your content is used as background knowledge without attribution.

The GEO Brand Citation Index found measurable citation-rate differences between engines for the same brand. A brand that is cited consistently by Perplexity may be invisible in Google AI Overviews. Engine-level tracking is not optional.

Step 3: Calculate Your Citation Rate and Benchmark It

Your citation rate is straightforward: (Total Citations / Total Checks) × 100. Calculate it per engine and in aggregate.

Benchmarks: most brands starting citation tracking see rates between 5% and 15% across a blended engine set. Above 20% on any single engine is strong for most categories. Below 5% means the engine has not yet indexed your content as a reliable source.

The more useful number is the month-over-month delta. Changes of 3 or more percentage points signal something you published — or did not — that shifted your standing. Flat rates are not neutral. They mean competitors have not outpaced you yet.

Step 4: Diagnose What Sources Drive Your Citations

This is the step most existing guides skip, and it is the one that makes the entire audit actionable.

When you see a citation change, trace it to the source type that caused it. An analysis of 23,387 AI citations by Omniscient Digital found that 68-85% of citations for branded queries come from third-party sources — not the brand's own website. Muck Rack's research confirms the pattern: earned media drives roughly 84% of AI citations.

Classify your citations by source type:

• Owned content: Your blog, documentation, product pages
• Earned media: Press coverage, analyst mentions, third-party reviews
• Third-party comparison: G2, Capterra, industry roundup lists
• Community: Reddit mentions, forum discussions, Hacker News threads

When citations drop, check whether a key third-party source was updated, removed, or outranked by a competitor. When citations increase, identify which new placement triggered it.

Here is the data point that reframes everything: Ahrefs, across 75,000 brands, found brand web mentions correlate with AI Overview brand visibility at r=0.664, against r=0.218 for backlinks and r=0.18 for Domain Authority — roughly 3x stronger for mentions. That is a correlation, not a disclosed selection rule. The sources that matter for AI citation are fundamentally different from traditional SEO signals.

Step 5: Build Your 90-Day Action Plan

A citation audit without a response plan is just a report. After each monthly audit, document three things:

1. Source gaps: Where competitors are cited and you are not. Pull the actual URLs the engines cited and identify what type of content earned the citation — I wrote about how to run a competitor citation audit recently.
2. Decay risks: Which of your existing citations are supported by content older than 6 months that has not been updated.
3. Leverage opportunities: Where you have visibility (brand mention) but not citation (linked source). This is the highest-ROI fix because the engine already knows you exist.

Prioritize by speed-to-impact: updating existing cited content shows results fastest because the engine already trusts the source URL. New earned media placements typically take 4-6 weeks to influence citation rates.

How to Report AI Citation Results to Your Board

AI citation data is new to most executive teams. The report format matters as much as the data.

Lead with a single number: your blended citation rate across all engines and prompts. Show the trend — three-month minimum. Show the competitive frame: your citation rate vs. named competitors on the same prompt set.

Do not present raw prompt-by-prompt data to executives. They need the aggregate trend and the strategic implication: "We are cited in X% of AI-generated recommendations for our category, up from Y% three months ago. The primary driver was [source type] — specifically [named placement]."

Connect citations to pipeline when you can. If you track referral traffic from AI engines — and you should, I wrote about how to set up the GA4 AI assistant channel recently — layer the traffic data onto citation movement to show which citations drive actual visits.

Metronyx AI publishes a useful citation report template if you need a starting structure. Their framework tracks 12 KPIs across 6 report sections and takes about 2 hours to compile monthly.

Common Mistakes That Waste Your Audit

Three patterns I see teams repeat:

Running prompts inconsistently. If you change prompt wording between months, your delta is meaningless. Lock the library. Track changes separately.

Ignoring engine-specific citation density. Muck Rack's May 2026 analysis measured this directly: ChatGPT cites sources in 96% of responses but averages only five citations; Gemini cites in 82% averaging eight; Claude cites in just 55% but averages thirteen when it does. A "citation rate" that blends these without accounting for how often and how densely each engine cites will mask engine-specific problems.

Confusing Domain Authority with citation likelihood. Traditional SEO authority metrics do not predict AI citation. Brand mentions are roughly 3x more predictive than backlinks. If your audit diagnosis defaults to "we need more backlinks," you are solving the wrong problem.

FAQ

How long does a monthly AI citation audit take?

Plan for 2-4 hours initially to build your prompt library and run the first baseline. After the library is locked, each monthly audit takes about 2 hours — roughly 30-60 minutes running prompts, and another hour on source-type diagnosis and action planning. Tools like Indexly and Presence AI can automate the prompt-running step.

What is a good AI citation rate?

Most brands see rates between 5% and 15% across blended AI engines. Above 20% on any single engine is strong. The more important metric is your month-over-month trend and your rate relative to direct competitors on identical prompts. A 10% rate that is climbing is better than a 15% rate that is flat.

How often do AI citations change?

Frequently. Scrunch's analysis of 3.5 million citation events puts the median half-life of an AI citation between four and five weeks. Knecht Strategies reports 40-60% monthly turnover in citation sets. This volatility is why monthly auditing is a minimum cadence.

Should I audit AI citations separately from SEO rankings?

Yes. AI citation and traditional search ranking are driven by different signals. Ahrefs, across 75,000 brands, found brand web mentions correlate with AI Overview brand visibility at r=0.664, against r=0.218 for backlinks and r=0.18 for Domain Authority — roughly 3x stronger for mentions, and a correlation rather than a disclosed selection rule. Run your citation audit as a separate operating discipline with its own prompt library, metrics, and action plan.

What tools can automate AI citation tracking?

Several platforms now offer citation tracking dashboards: Indexly tracks share of citation across ChatGPT, Perplexity, and Gemini. Presence AI monitors citation trends over time. Omnia focuses on Perplexity tracking specifically. The tools handle prompt running and recording — the source-type diagnosis and action planning still require a human operator.