AI overview tracking: AI search summaries per query and market

Summary present

Yes or no, per query, market and run. The presence rate across runs is the headline metric.

Cited sources

Each linked source in the summary, its domain and its order. Whether you are cited is the main outcome.

Summary text

Stored for diffing and for checking how your brand or product is described.

Brand mentions

Whether your brand and competitors are named in the text, regardless of citation.

Organic results below

Positions and URLs, to see whether the cited sources are also the top-ranked pages.

Pixel offset of first organic result

How far the summary pushes the links down the page, at a fixed viewport.

Market, exit IP and viewport

Country, language, exit and device profile, so every run can be reproduced.

Run number within period

Lets you report presence and citation as rates over repeated samples, not single observations.

1. 01

   Pick queries that can trigger a summary

   Summaries appear far more often on long, question-style queries than on one- or two-word ones. Build a query set per market from your informational and comparison keywords, translated for each language.

2. 02

   Render the page, logged out, in the market

   Load each query in a headless browser from an exit in the market, with a fresh context and the market's locale. Hold one sticky session for the page load and any "show more" expansion, then discard it.

   Playwright: one fresh context per query, market in the username21 linescopy

   import { chromium } from "playwright"const [host, port] = "proxy.arcproxies.io:8000".split(":")async function check(country: string, locale: string, url: string) {  const sid = Math.random().toString(36).slice(2, 10)  const browser = await chromium.launch({    proxy: {      server: `http://${host}:${port}`,      username: `USERNAME_pool-arcfast_country-${country}_session-${sid}_sesstime-5`,      password: "PASSWORD",    },  })  const ctx = await browser.newContext({ locale, viewport: { width: 1366, height: 900 } })  await ctx.route("**/*.{png,jpg,jpeg,webp,gif,woff,woff2,mp4}", (r) => r.abort())  const page = await ctx.newPage()  await page.goto(url, { waitUntil: "networkidle" })  const html = await page.content()  await browser.close()  return html}

3. 03

   Parse the summary as a block

   Extract presence, text, cited links in order, and the position of the first organic result. Parse the organic results from the same page so citation and rank come from one observation.

4. 04

   Sample, then aggregate

   Run each query several times per period, on fresh sessions. Report presence and citation as rates, for example "cited in 4 of 6 runs", and treat changes smaller than your sampling noise as no change.

5. 05

   Diff summaries over time

   Compare the stored text and citations week on week. A new source cited, a competitor added, or a factual statement about your product changing are the events worth a person's attention.

Prices are list prices in EUR, excl. VAT. Volume discounts apply automatically at checkout — full ladder on the pricing page.