Two families of metrics
The whole framework rests on one simple distinction. Presence metrics measure your place in LLM answers: they react fast and steer the day-to-day work. Outcome metrics measure the real effect on revenue: leads, sign-ups, sales, with weeks of lag. The former justify the effort, the latter justify the budget. You need both.
Presence metrics: prompt tracking
Start by building a prompt library representative of your business and product goals, say e-invoicing. A prompt tracker such as Peec, Profound or Otterly runs them every day across the different models and collects three things: mentions (your brand appears in the answer), citations (your brand is quoted as a source) and position (the average rank of your mention across answers).
From these three raw signals I derive five metrics.
| Metric | How it's computed | What it tells you |
|---|---|---|
| Share of voice | Your mentions over all mentions | How you perform against others. Means nothing in business terms by itself, but its trend validates your efforts, or not. |
| Visibility | Mention rate over the number of prompts run | How well you cover the scope of your prompt set. A good working KPI. |
| Position | Your mention's rank relative to competitors | Gold, in my view: being quoted first probably means you rank higher in credibility. |
| Weighted mention | Average of the inverse rank across all prompts, on a scale of 100 (formula below the table) | Track it daily and you have a genuinely interesting business KPI. |
| Weighted share of voice | The same computation, applied to each competitor | Compares your coverage and position to the market. Richer than classic share of voice, and complementary. |
The weighted mention deserves its formula, since it is the one steering the work. For each prompt, your mention is worth the inverse of its rank: 1 if you are quoted first, 0.5 second, 0.33 third, 0 if you are absent. The average of these scores across the whole library, multiplied by 100, gives the metric. Example over 100 prompts: quoted first 30 times, second 20 times and third 10 times, that is (30 × 1 + 20 × 0.5 + 10 × 0.33) / 100 × 100 ≈ 43 out of 100.
Add two qualitative reads. Citations first: since a quoted URL influences the model's narrative, dig into which ones work, use them to justify investments or as priority markers, say the discussion threads to influence first or the video creators to contact, and spy on competitors. Sentiment next: how the models talk about you. If you regularly brief third parties who will talk about your brand, track the effect of each event, like a newly earned label, on overall sentiment or split by product.
Four complementary sources
| Source | What it gives you |
|---|---|
| Provider data | Bing and Google publish their own AI-related data. Mostly useful to justify budget allocation, or to dig into something specific. |
| Server logs | Crawler activity in your logs answers concrete questions: you shipped a technical improvement to help models understand the site, what changed on the crawl side? |
| Third-party panels | Some providers, like Similarweb, can tell whether someone visited an LLM the same day as your website. Take it with caution, but it's always a valuable angle. |
| AI Overview tracking | A good working indicator for measuring the outputs of tactics or changes made to the website: new content, new formatting. |
Outcome metrics: the real impact
Declarative data first. The question "how did you hear about us?" measures the real business impact, and its verdict often diverges from your attribution model's. Without it you miss the most common scenario: someone reads a recommendation in an LLM, types your brand into Google, clicks a paid or organic link. No attribution model sees that chain.
Analytics next. In GA4, filter the referral URL patterns specific to LLMs (chatgpt.com, perplexity.ai, copilot.microsoft.com, gemini.google.com, claude.ai): when a visitor arrives by clicking a link in an answer, you can attribute it. Complete the picture with cookie data: if a prospect visits your site, goes to an LLM, then comes back, you should see it, depending on your attribution philosophy, first, last or mixed click.
Indirect signals last. An unexplained uplift in brand and direct traffic follows the same logic as the declarative case. Run a daily analysis of your sales team's transcripts to catch prospects who mention an LLM as a source, a small automation handles it easily. And keep an eye on declarative market studies, public or paid, that look into how people search: always insightful.
| Signal | What it captures | Its limit |
|---|---|---|
| Declarative | The real channel through which prospects found you | Relies on the respondent's memory and honesty |
| Analytics (GA4) | Direct clicks from LLM answers | Blind to a recommendation followed by a brand search |
| Indirect signals | Brand traffic, sales transcripts, studies | Correlations, never proof on their own |
Bringing it all together
The last link is the simplest: gather all the data in one place and hand Claude a weekly analysis, with a short recap of how the trends are evolving, delivered as a Slack message. Measurement only matters if someone reads it every week.
In 2026, the impact of LLMs on lead generation can't be read off any ready-made dashboard. You rebuild it, presence metric by outcome metric, until both curves tell the same story.
Frequently asked questions
More analyses on measurement and generative search are available in the journal.