Your analytics is hiding some of your visitors. They're AI agents — and they're shaping what people buy.
Corrected 24 September 2026. The first version of this post gave a share of agent traffic and a breakdown of agent behaviour from "the stores I've sampled". Some of those patterns, like an agent leaving for a competitor's page and coming back, can't be seen from a store's own data, and we can't stand behind the figures, so they're gone. It also described Tactical features (a Slack bridge, an agent welcome banner, paid tiers covering steps 3–5) that we've since parked. The argument and the playbook are unchanged.
ChatGPT visited a product page yesterday. Someone had asked it for a leather laptop bag under $200; it browsed a few stores and recommended one. Maybe yours, maybe not. The store never finds out which.
This is happening on DTC storefronts every day, and the merchant usually has no idea. Their analytics dashboard shows visitors and a conversion rate. What it doesn't show: some of those visitors weren't people clicking around. They were agents (ChatGPT-User, PerplexityBot, ClaudeBot, Google's crawlers) doing the research a person used to do, then influencing what that person bought.
I've spent months building tooling to see this traffic. Some of what I've found is what you'd expect. Some of it isn't. And the gap between merchants who can see it and merchants who can't is going to widen.
Here's what I've learned, what I think it means, and what to do about it if you run a $1–10M DTC store.
What the traffic looks like
The first thing you notice once you can see it is that agent visits don't look like human ones. Agents tend to go straight to a product page rather than browsing a category, because the person already asked about a specific product somewhere else. Their sessions are often short in time and deep in pages. Most never touch the cart; a few do.
What you can't see from your own store is just as important: the agent's prompt, the other stores it looked at, the prices it compared. Your data shows what happened on your site, and only that.
Standard analytics (GA4, Shopify's reports, anything keyed to bounce rate and session duration) either filters this traffic out as "bot noise" or, when the agent runs a real browser, counts it as a person. Either way it isn't legible on the dashboard the merchant actually looks at.
Why this is happening now
Three things lined up.
1. Agent traffic became real, fast. OpenAI publishes the user agents its crawlers use (OpenAI bots reference), and so do Perplexity (Perplexity crawlers), Anthropic, Google, Common Crawl and ByteDance. What changed in 2025–2026 is that these crawlers stopped being curiosities and started showing up on ordinary storefronts. Cloudflare's Radar dashboards let you watch AI crawler activity yourself.
2. End-user behaviour caught up. People who asked ChatGPT to debug their code last year now ask it where to buy a sound machine. The shift from "search → click → buy" to "ask → buy" is partial (humans still browse, still click ads), but the agent is increasingly in the loop on the consideration step. By the time the human shows up on your storefront, the agent has already shaped what they're looking for.
3. The infrastructure to act on it is barely there. Most analytics platforms still treat agent traffic as a filtering problem. Most merchants don't know what their best-performing product for AI assistants even is. There's a window where the operators who pay attention will compound an advantage on the operators who don't, and then it closes, partly because tooling catches up, partly because agent traffic gets too big to ignore.
That's why this is the year to start measuring. Not 2027.
The bet I'm making
I'm building Tactical on a deliberate two-step bet.
Step one (now): intelligence. Show merchants what agents are doing on their store. Which agents, what they look at, what they buy, where they leave. The wedge is informational: you can't act on traffic you can't see. The moat is time-decaying: as standards like ACP and AP2 mature, signed-agent traffic will make detection easier and the measurement business will compress. That's fine. Measurement isn't the endgame.
Step two: response. The durable product is helping merchants respond to agent traffic, not just see it. Not "your store has a chatbot." More like deciding, per agent, what it's allowed to do on your store, and acting when an agent does something that matters. The intelligence wedge is the on-ramp.
Why DTC mid-market specifically, why $1–10M GMV stores and not the long tail or the enterprise? Because that's the band where:
- Agent traffic is already worth measuring, not a rounding error
- Operators have budget to act on it ($99–$500/mo SaaS works without procurement)
- Decisions are fast enough to validate ideas in weeks, not quarters
- Trust in template-driven autonomy is realistic; the long tail is too price-sensitive, the enterprise too slow
What I'm deliberately not building:
- Marketplace integrations (Amazon, eBay, Walmart): different product, post-purchase signal only
- Protocol-native runtimes for ACP/AP2: standards still firming
- Enterprise: a different distribution motion that doesn't fit a solo founder's pace
Holding scope is the hardest part. Most merchants ask for the next thing five minutes after seeing the first thing. The discipline is shipping the wedge fully before chasing the next layer.
The agent-ready DTC playbook
If you run a DTC store and you've read this far, here are the five things to actually do. Each is DIY-able in about half a day. Tactical helps with the first two.
1. Stop filtering agents as bot noise
The first move is just turning on visibility. In Google Analytics, "bot traffic" is a single excluded bucket: useful for keeping scrapers out of your reports, useless for telling "ChatGPT visited my product page on behalf of a real customer" apart from "a scraper ripping my catalog."
What to do:
- Look at your raw server logs (or your CDN logs: Cloudflare, Fastly, Vercel) for the past 30 days. Filter to user agents matching
ChatGPT-User,GPTBot,PerplexityBot,ClaudeBot,Google-Extended,Amazonbot,Bytespider. Count the requests. - For each agent, count how many distinct product URLs it touched.
You're looking for a baseline. The exact number isn't the point; establishing that this traffic exists on your store is.
2. Classify agents apart from each other
Not every agent is the same. ChatGPT-User (fetching a page because a ChatGPT user asked a question) is fundamentally different from GPTBot (crawling for training data). Both come from OpenAI; only one is a sales signal.
What to do:
- Tag agent visits by kind.
ChatGPT-User,Perplexity-Userand similar user-triggered fetchers are shoppers' proxies. GPTBot,Google-Extended,CCBotare training crawlers. They matter for how AI knows your catalog, less for this week's sales.- Apply the tag when you collect the data. Most off-the-shelf analytics won't do this for you. You'll need a small classifier (a user-agent list plus behavioural heuristics) or something purpose-built; Tactical does this, and so does a CDN with bot management if you write the rules yourself.
The output you want: a per-store breakdown of the agents acting for a shopper versus the crawlers indexing or training.
3. Make your structured data agent-friendly
Agents read your storefront the way a screen reader does: they need the facts about a product, not the marketing copy. The single highest-leverage thing you can do for agent visibility is make sure your product pages emit clean, complete Product schema in JSON-LD.
What to do:
- Open one of your top product pages. View source. Look for
<script type="application/ld+json">. Validate it at validator.schema.org and Google's Rich Results Test. - Make sure every product has
name,description,image,offers(withprice,priceCurrency,availability),brand,sku, and ideallyaggregateRatingandreviewif you have them. - For variant-heavy products: emit an offer per variant or a
hasVariantarray, not a flat "starts at $X" string. - On Shopify, the standard themes get most of this right but often skip
aggregateRating. Worth checking. - On WooCommerce, check what your theme and SEO plugin output; many stores need an SEO plugin that emits full Product schema, or to write it themselves.
This is the work that compounds. The agents reward stores that make their job easy.
4. Get a human in the loop for high-intent agent sessions
When an agent on your store is doing the slow, deliberate version of a buying decision (viewing a product, reading specs, checking stock and price, leaving, coming back), you want to know while it's happening, not in next week's report.
What to do:
- Define what a high-intent agent session means for your store. A working baseline: an agent visitor, three or more product pages, a cart event.
- When that pattern fires, get an alert where you'll see it. Tactical sends an email; any analytics tool with webhook outputs and a short script can post to Slack.
- Decide your response in advance. Most merchants haven't thought about what they'd do with a high-intent agent session, so the alert becomes noise. Maybe it's an inventory check, maybe a pricing review, maybe nothing, but make it deliberate.
5. Welcome the agents you want
This is the most contrarian recommendation. Most operators think about agents as a threat: scrapers, bot traffic, lost revenue. That misses something. A meaningful share of agent traffic to your store is trying to recommend you. It just needs you to make that easy.
What to do:
- Give agents something machine-friendly: a clean product feed, spec sheets, or an
llms.txtat your root. There's no formal standard forllms.txtyet, so treat it as cheap and optional. - For your top 20 products, write a one-paragraph plain-text spec that strips the marketing copy, and make it easy to find from the product page.
- Decide which agents you don't want, too. Training crawlers and shopping agents are different visitors, and you can treat them differently.
What's next
The five-step playbook above is roughly where the intelligence wedge ends: visibility, classification, a basic response, and a friendlier surface for agents that want to recommend you. Doing all five takes a serious DTC operator a couple of weeks.
If you're a DTC operator in the $1–10M band who's curious what your agent traffic looks like, or you've done the playbook and want help with what comes next, I'm working with a handful of stores at a time. The fastest path is the Enquire form on the Tactical billing page, or email [email protected]. I read everything personally.
If the playbook is enough for you, Tactical's free tier gives you 100 agent sessions a week to do steps 1 and 2 against your actual store. No sales motion: sign up, paste a snippet, and your first data arrives within about 15 minutes.
The window is open right now. The operators who started measuring agent traffic this year will have a year of pattern data on every operator who didn't. That's the bet I'm taking.
Sourabh, building Tactical solo. Found this useful, or disagree with the bet? Email [email protected]. I read everything.