In partnership with
Stop Paying for 10 Tools. One AI Does It All.
Most e-commerce sellers are running their store across 6 to 10 separate tools — and spending more time managing software than growing their business. StoreClaw replaces your entire stack with one autonomous AI engine that monitors competitors, optimizes listings, automates marketing, and tracks real profit across Shopify, Amazon, and beyond.
It doesn't wait for you to ask. It runs 24/7 in the background, so you wake up to a full dashboard instead of a list of things you forgot to check.
Connect your store, and StoreClaw gets to work — no prompts, no complex setup, no six-app stack.
Free to start. No credit card required.
First, a correction
Last week I told you ChatGPT's shopping answers had got thinner, and I showed you a product card that had lost its retailer panel between 20 and 30 August. My analysis missed a crucial part. Logged-in free users do not see the same difference.
What we are actually seeing is the logged-out free experience diverging from the logged-in free one, which makes sense when you think about where OpenAI is heading with ads. It wants more people signed in, because a signed-in user is one it can serve personalised ads against, so thinning out what you get without an account is a perfectly logical way to push people through the door.
I did not check the logged-in version before publishing, and for that I apologise.
Here is the difference.
Logged in on a free account, and much the same as it was before.
Logged out on a free account. This is the view I showed you last week.
Thinking of starting your own newsletter?
I write this one on beehiiv, and I have got more out of poking around their tools than from any amount of advice about how to grow a list. If starting your own has been sat on your to-do list for a while, this is a reasonable week to stop thinking about it.
Free 14-day trial, then 20% off for your first three months.
Affiliate link.
The attribution question has a boring answer
Three different client conversations this week ended in the same place, which usually means the question has reached the boardroom. Some version of: the visibility work is going well, now show me it is doing something commercial.
One of them put it precisely. They want a credible framework linking AI visibility to a key conversion by the end of the year, and they were explicit that it does not have to be direct attribution. Evidence of a commercial outcome is enough.
That is a far easier brief than the one the industry keeps trying to answer, and most people are still reaching for the harder one.
What customers said when asked how they found the company, January 2025 to July 2026.
So here is the boring answer. Put "where did you hear about us?" in your onboarding flow, make the answer attach to the customer record rather than sitting in an anonymised survey and report it every month alongside everything else you report.
Most brands already have that field. The trouble is that most of them have it configured so the answer cannot be joined back to a person, which turns it into a satisfaction survey rather than a source of attribution, and that is the bit worth fixing. It is a smaller job than any model you were thinking of building instead.
We logged a new client this week with the source recorded as ChatGPT. One line in a CRM, and the question was answered for that customer.
None of this gives you a clean attribution model, and I don’t think that is possible with our current tech. What it gives you is a number you can point your effort at, and once you have that, you can start putting budget behind the work rather than defending it.
If you have the resource and the team for it, you can go a long way past a dropdown. Fund some actual research, and ask customers about the conversation that brought them to you. I once saw a marketer offering $50 Amazon vouchers in exchange for a customer's ChatGPT history, which sounds extreme right up until you remember that OpenAI gives us none of this, and that history is the only real record of how that decision got made.
It helps to think about how the journey actually runs. Someone does most of their thinking inside the platform, asks their questions, narrows the list down and only leaves once the need has been met and they are ready to buy. Then they arrive on your website and it lands in your analytics as direct, or as a lift in brand search through organic or paid, with nothing anywhere saying ChatGPT sent them.
That is why a 30-second question at the point of sale is worth more than it looks, and why your sales team should be asking it on every call.
What to do today. Find out whether your how-did-you-hear-about-us answers can be tied back to a customer record. If they cannot, that is your fix, and it is probably an afternoon's work for whoever owns the form.
How I know this
The chart is self-reported attribution from a brand I have worked with for several years, anonymised and shared with permission. Customers are asked how they heard about the company when they sign up, and the two lines are two of the answers plotted over time. Internet search is people saying Google. AI search is people saying ChatGPT, Perplexity or Claude.
It is not offered as a model of how to do attribution, it is just what happened once the question was actually being asked and recorded properly.
One limitation worth being straight about. A good share of what lands in the Google bucket will in fact have been Google AI Mode or AI Overviews, because people describe all of it as Google. So if anything the AI line here is understated, and the real gap between the two is narrower than it looks.
That is it for this week, until next time.
Have a good one.
Andy
🗣️ This week's stories
What caught my eye in AI search this week.
The EU has decided ChatGPT is a search engine. It has been designated a Very Large Online Search Engine under the Digital Services Act, because the rules have no category for generative AI and it answers you by searching the web. Obligations start in four months. Bloomberg
OpenAI's ad business is at $1bn annualised, against the $2.4bn it forecast for this year. If you have been planning around ChatGPT ads becoming a serious channel in 2026, plan around that number instead. CNBC
Creator briefs are starting to look like a map of AI search queries. Handy if you are trying to explain to a client why their influencer budget and their AI visibility work are really the same budget. Digiday
Investors are piling into agents that shop for you. Newcomer mapped more than 35 startups doing it, and the detail worth keeping is Phia, accused of claiming referrals it had not earned, because nobody has settled who gets paid when an agent is the thing doing the clicking. Newcomer
GPT-6 Astra started rolling out on Wednesday, with OpenAI hinting fairly heavily at AGI. Every model change this year has moved citation behaviour within days, so check which model you are being served before you react to next week's numbers. The Verge