Introducing the Foglift Tracker: Close the Loop from AI Citation to Revenue

The gap in citation-only tools

Every AI search visibility platform on the market answers the same question: when a user asks an engine about your category, does the engine name your brand? That is a real signal. It is the entire category that Foglift, Profound, Peec, Otterly, and Athena exist to measure. But it stops at the answer. The user reads the answer; they click through, or they don't; the engine logs nothing back to you either way. From your side of the connection, the loop is open.

The Foglift Tracker closes that loop. It is a one-line install on your own web server that logs two things: every visit from an AI crawler like GPTBot or ClaudeBot, and every human visit referred from a ChatGPT, Claude, Perplexity, Gemini, or Microsoft Copilot answer. The crawler half is the leading indicator; the referral half is the revenue-relevant signal. Both flow into the same dashboard, through one npm package, with no cookies and no fingerprinting.

Two signals, one install

On a request-by-request basis, the tracker runs two quick checks. If the User-Agent header matches a known AI crawler, the visit is reported as a crawler hit with the matched bot name. If the User-Agent looks human but the Referer header points at one of the canonical AI search hostnames (chatgpt.com, claude.ai, perplexity.ai, gemini.google.com, copilot.microsoft.com, notebooklm.google.com), the visit is reported as a referral. Everything else, which is the vast majority of your traffic, is never reported at all.

The two signals tell different stories. A crawler hit tells you an engine fetched a specific page. It is the warmup move; on its own it doesn't pay off, but it's the leading indicator that a citation may follow. A referral click tells you the citation has already landed and a human acted on it. That is the revenue-relevant event: someone read a ChatGPT or Perplexity answer that mentioned you, decided to follow up, and is now on your site, three feet from the signup button.

The centerpiece: per-visit landing-page attribution

Every referral click carries the exact page the visitor opened. That sounds incidental; it is in fact the bridge from did AI cite me to which AI answer cited me, and for what query.

Consider a concrete example. Imagine your B2B SaaS product is a project management tool. Over a week your tracker logs the following referral visits:

• A visitor from ChatGPT lands on /pricing. Reasonable inference: ChatGPT cited that pricing page in an answer to a pricing-shaped prompt (“how much does Acme cost,” “cheapest project management tool for a team of five”).
• A visitor from Perplexity lands on /integrations/asana. Reasonable inference: Perplexity cited that integration page in an answer to an Asana-comparison prompt or an Asana-migration prompt.
• A visitor from Claude lands on the homepage. Reasonable inference: Claude cited your brand at a generic recommendation level rather than recommending a specific feature.
• A visitor from Gemini lands on /blog/agile-sprint-planning. Reasonable inference: Gemini cited that blog post in an answer to a methodology question, not a product comparison.

None of those engines told you what the prompt was. They never will. But the landing page is a leaky proxy for the prompt class, because the engine is choosing a page that semantically matches what the user asked. If you cross-reference the landing page with the content that lives on it, you can reconstruct the rough prompt shape that earned the citation. Repeat the exercise across a week of referral traffic and patterns emerge: which engines cite you for pricing prompts vs. feature prompts vs. comparison prompts; which engines like your blog posts and which prefer your product pages; which integrations get cited by which engines.

That is the reverse-engineering channel that no citation-only tool gives you. Citation tools tell you the engine's output. The tracker tells you the engine's output paired with the user's intent, indirectly but reliably enough to act on.

Install

The package ships first-class adapters for Next.js, Express, plain Node.js HTTP servers, and Cloudflare Workers. Install once, set FOGLIFT_API_KEY in your environment, and reports start flowing within seconds of the next matching request.

npm install @foglift/tracker

The Next.js install is a four-line middleware:

// middleware.ts
import { trackAITraffic } from "@foglift/tracker/nextjs";

export const middleware = trackAITraffic({
  apiKey: process.env.FOGLIFT_API_KEY!,
});

export const config = {
  matcher: ["/((?!_next/static|_next/image|favicon.ico).*)"],
};

Express:

import express from "express";
import { trackAITraffic } from "@foglift/tracker/express";

const app = express();
app.use(trackAITraffic({ apiKey: process.env.FOGLIFT_API_KEY! }));

Plain Node.js HTTP servers and Cloudflare Workers each have their own export with the same shape. Full reference at the Foglift Tracker product page.

Generate an API key at foglift.io/dashboard/settings. The tracker pipeline is free on the Growth plan and above; the free product tier includes scans but not the persistent tracker.

What changes when you have this data

Three things, in increasing order of leverage.

First, the crawler feed tells you when an engine has noticed new content. A blog post ships on Monday. Tuesday morning, GPTBot crawls it. By Wednesday, ClaudeBot follows. That sequence is the earliest possible signal that the content is now eligible for citation. It also tells you the reverse: a page that no AI crawler has touched in 90 days is almost certainly invisible to the engines and worth investigating.

Second, the referral feed tells you which content is actually getting cited. A page that earns three Perplexity referrals per week is being cited weekly. A page that earns zero referrals over a month is either not being cited or is being cited in answers that don't drive click-through. Either way it's information you can act on with a content rewrite, a schema upgrade, or a structural change to the page.

Third, the landing-page-per-visit field tells you the prompt shape behind each citation. If three of your last ten ChatGPT referrals landed on /pricing, the engine is citing you for pricing prompts. That is a different optimization problem than being cited for feature-comparison prompts. Foglift's actions engine reads the tracker feed and routes recommendations accordingly: pricing-cited pages get one set of nudges (schema for Offer, microcopy emphasizing transparency, FAQs answering price-shape questions); feature-cited pages get a different set.

Why this ships now

The AI search visibility category is maturing. The first wave of platforms answered the question “does the engine cite my brand?” That question matters, but on its own it leaves the operator one full step short of revenue. The next wave has to answer “what did the citation cause?”. Foglift is shipping the instrumentation for that wave first, because the underlying data (referrer headers, user-agent strings) is already in every HTTP request on your server; the only thing missing was a clean way to classify and log it. The tracker is that clean way.

It is also the half of the loop that Profound, Peec, Otterly, and the broader citation-only field structurally can't do, because they observe the engine externally. They watch the answer. The tracker watches your server. Both halves are necessary; together they form the complete optimize / crawl / cite / click loop that Foglift now measures end to end.

Get started

Install the package, set your API key, and watch your AI referral feed start populating. The dashboard at foglift.io/dashboard/crawlers renders both feeds side by side, with referrals on the primary tab and crawlers on the secondary. The product page at foglift.io/tracker covers the full adapter list, the FAQ, and the install snippets in one place.

If you already run on the Growth plan, the tracker is included. If you're on the free tier and want the closed-loop view, the pricing page has the upgrade. Citation-only is the past five years of AI search tooling. The closed loop is the next five.

Fundamentals: Learn about GEO (Generative Engine Optimization) and AEO (Answer Engine Optimization) (the two frameworks for optimizing your content for AI search engines).