How to Measure Whether Your AI PR Campaign Actually Worked

Your PR agency says they got you cited in ChatGPT. Maybe they did. But unless you can measure citation rate across engines, track whether those citations actually describe your product correctly, and connect the activity to pipeline, you are paying for screenshots, not results. Here is the measurement framework I use to evaluate AI PR campaigns.

Why Traditional PR Metrics Fail for AI Campaigns

Impressions, AVE, and media mentions were designed for a world where humans read articles. AI PR campaigns operate differently. A placement in a tier-one outlet now has two pathways to value: the direct human audience and the AI-mediated pathway where AI assistants absorb the placement into their training data and retrieval systems. Presenc AI's measurement framework calls this the "dual-pathway" model — and most PR agencies only report on the first path.

The gap is real. Fullintel's 2026 PR measurement research documented what many of us already suspected: traditional impressions-based reporting misses the compounding value that placements create inside ChatGPT, Perplexity, Claude, and Google AI Overviews. If your agency is still sending you clip books and reach estimates, they are measuring the wrong thing.

The Five Metrics That Actually Matter

I have narrowed this down to five numbers that tell you whether an AI PR campaign produced results:

1. Citation rate across AI engines. What percentage of relevant queries result in your brand being cited? This is the core metric. Run a fixed set of commercial-intent queries across ChatGPT, Perplexity, Claude, and Google AI Overviews on a regular cadence. Crackle PR publishes their methodology openly: 25 commercial-intent queries across 4 AI systems, 3 runs each, producing 300 scored answers per quarter. They track first-mention rate, mean rank, and total mention rate. You can replicate this approach for your own brand.

2. Citation accuracy. Being mentioned is not enough. If ChatGPT describes your product incorrectly or attributes a competitor's feature to you, that citation is doing damage. Check that the AI engines describe your actual offering, position, and differentiation correctly.

3. Source-page retrieval volume. Monitor which of your pages AI bots actually retrieve. Server logs from ChatGPT-User, PerplexityBot, ClaudeBot, and Applebot show you which content the engines are pulling for answers. If your PR campaign created new source pages, you should see retrieval activity on those specific URLs.

4. Pre/post citation shift. Compare citation rates before and after the campaign. Spinsucks identifies four PR metrics that connect to AI visibility, but the one that matters most for campaign evaluation is whether your citation rate moved. A campaign that generated placements but did not shift citation rates did not work.

5. Pipeline attribution. This is where it gets hard and where most agencies stop. PR placements feed AI training data with a lag — Presenc AI documents a four-to-twelve week carryover where a high-tier placement gradually appears in AI answers. You need to connect the chain: placement published, AI citation detected, branded search or direct visit attributed, pipeline generated.

How to Set Up a Quarterly AI PR Measurement Cycle

Here is the operational sequence I recommend to any CMO evaluating an AI PR campaign:

Week 1 (baseline). Before the campaign launches, run your commercial-intent query set across all four engines. Record citation rate, accuracy, and which sources the engines cite. This is your baseline.

During campaign. Track placements as usual, but add a column: which placements are on pages that AI engines can crawl and retrieve? A placement behind a paywall or in a PDF that AI bots cannot index has zero AI visibility value.

Week 8-12 (post-campaign). Rerun the same query set. Compare citation rates. Check whether new placements appear as cited sources. Review server logs for bot retrieval on campaign-created URLs.

Quarter close. Calculate the delta. Did citation rate increase? Did accuracy hold or improve? Can you trace any pipeline to the AI-mediated pathway?

What a Measurement Dashboard Looks Like

Metric	Baseline (Pre-Campaign)	Post-Campaign	Delta
Citation rate (% of queries citing brand)	Measure across 25+ queries	Rerun same set	+/- percentage points
First-mention rate	% of answers where brand appears first	Same methodology	Movement direction
Citation accuracy	% of citations correctly describing offering	Manual review	Accuracy trend
AI bot page retrievals	Server log count by bot type	Same period comparison	Volume change
Pipeline from AI-attributed traffic	CRM-tagged conversions	Same attribution window	Revenue delta

This is not a vanity dashboard. Every row connects to a decision: increase investment, change the placement strategy, fix accuracy problems, or fire your agency.

The AI-Mediated Pathway Most Agencies Ignore

Presenc AI's research frames this clearly: PR now operates through two distinct pathways. The direct pathway is what agencies have always measured — someone reads the article, clicks through, converts. The AI-mediated pathway is what happens when that article enters the training data or retrieval index of an AI engine and starts generating citations weeks or months later.

Most agencies ignore the second pathway entirely because it is harder to measure and the lag makes it difficult to attribute to a specific campaign. But the AI-mediated pathway is where the compounding value lives. A single well-placed feature in a tier-one outlet can generate citations across ChatGPT, Perplexity, and Claude for months after publication.

The measurement approach: use AI visibility tools to track citation rates over time, and correlate citation appearance with the publication dates of specific placements. The signal is there if you look for it.

Red Flags in an Agency's AI PR Reporting

If your PR agency claims AI visibility results, here is what to watch for:

• Screenshots without methodology. A screenshot of ChatGPT mentioning your brand proves one query at one moment. Without a fixed query set, multiple runs, and time-series data, it proves nothing about campaign impact.
• No baseline comparison. If the agency cannot show you what citation rates looked like before they started, they cannot prove they moved the number.
• Mixing up impressions and citations. Media impressions (how many people might have seen the article) are not AI citations (how often the AI engine cites that article when answering queries). These are different metrics with different measurement methods.
• No accuracy audit. Being cited incorrectly is worse than not being cited. Any serious AI PR measurement includes a check on whether the citations accurately represent your brand.
• Ignoring engine differences. ChatGPT, Perplexity, Claude, and Google AI Overviews each retrieve and cite sources differently. Meltwater's guide to measuring PR performance across LLMs documents why per-engine measurement matters. An agency reporting a single blended "AI visibility" number is hiding the details.

How to Start This Week

You do not need expensive tools to begin. Here is the minimum viable measurement approach:

1. Write down 15-25 queries your buyers actually search when evaluating your category. These should be the commercial-intent queries where you want to appear.
2. Run each query in ChatGPT, Perplexity, Claude, and Google AI Overviews. Note whether your brand is cited, where it appears, and whether the description is accurate.
3. Check your server logs for ChatGPT-User, PerplexityBot, ClaudeBot, and Applebot traffic. Identify which pages they are pulling.
4. Record everything in a spreadsheet. This is your baseline.
5. Repeat monthly. The trend is the signal.

If your PR agency cannot produce this data, or cannot explain how their work moves these numbers, you have the wrong agency for this era.

FAQ

How often should I measure AI PR campaign results?

Run your full query set quarterly at minimum, with monthly spot checks on your top 10 commercial-intent queries. Citation rates can shift within weeks of a major placement, but the compounding effect of AI-mediated PR typically takes 8-12 weeks to stabilize. Monthly tracking catches fast movers; quarterly tracking reveals the campaign's real impact.

What is the difference between AI citation rate and share of voice?

Citation rate measures how often a specific AI engine cites your brand when answering a defined set of queries — it is a precision metric tied to specific questions. Share of voice measures your brand's overall presence across a broader set of AI conversations relative to competitors. Both matter, but citation rate is the better metric for evaluating a specific PR campaign because you can tie it directly to the queries the campaign targeted.

Can I measure AI PR results without specialized tools?

Yes. Manual query testing across ChatGPT, Perplexity, Claude, and Google AI Overviews gives you accurate citation rate data. Server log analysis for AI bot traffic requires basic log access but no specialized tooling. The manual approach does not scale to hundreds of queries, but 15-25 commercial-intent queries measured monthly is enough to evaluate whether a campaign moved the needle.

How do I know if my PR agency is actually delivering AI visibility results?

Ask for three things: a fixed query set they test against, baseline measurements from before the campaign, and time-series data showing citation rate changes correlated with specific placements. If they can produce all three with per-engine breakdowns, they are measuring seriously. If they send screenshots or aggregate "AI mentions" without methodology, they are guessing.