Newsletter September 2026: Agentic sales hit different
0.41% of traffic. 3.99% of orders.
Hi there,
0.41% of traffic. 3.99% of orders.
Those are one brand’s agentic numbers over 77 days this summer. Roughly 10 times the conversion their traffic share predicts. Small channel, unusual rate, and it is compounding straight into Q4.
Tamir here from 40rty. This isn’t a mass email; it’s a direct line from our product team.
This is the first BFCM where AI agents are in the buying session rather than on the sidelines. 3 updates, the full numbers, 3 releases. I’ll be in New York in two weeks; details are at the end.
Agent accessibility was just turned ON
The industry is converging on WebMCP, a way for a site to hand an agent a set of tools instead of making it click around. OpenAI, Google Chrome and Shopify are all pushing it.
Here is the part that matters to you: on Shopify it is already on. Agents can search your catalog, pull product details, change the cart, head to checkout, and answer questions from your policies and FAQs. So your product data is doing the selling now, not your theme. And your policy and FAQ content just became a support surface, meaning thin or stale content turns into a wrong answer at the worst possible moment.
We’re working with selected brands to design extended WebMCP capabilities — increased personalization, better search, intent based landing pages, quizzes and much more.
The fight is at comparison, not checkout
New research out on August 17 breaks down where shoppers actually use AI in the buying journey. About 62% is product comparison. Only 23% is checkout and 19% is post-purchase.
Everyone is watching the checkout story, agent payments, instant checkout, who owns the transaction. That is not where this is being decided right now.
It is being decided one step earlier, when an agent lines your product up against three competitors and picks one. That comparison runs on your product data. Whoever answers the shopper’s specific question best gets the recommendation, and the checkout question sorts itself out after that.
Your Q3 prompts will not survive Q4
“A healthy catalog only proves you are allowed to compete.” That is from our newest post, and it is the trap here. Your catalog can be clean and your listings can still be answering August’s questions.
Four things flip in the next eight weeks.
The shopper stops being the buyer.
- Q3
- waterproof hiking boots for wide feet
- Q4
- a gift for my dad who hikes, I think he is a size 11
The prompt is now secondhand and vague, describing someone who is not in the chat. Your listing has to answer for a person the shopper can only half describe.
The delivery date becomes a filter, not a checkout step.
- Q3
- best noise cancelling headphones under $200
- Q4
- under $200 and arriving before December 24
An agent that cannot confirm you will make the date drops you from the comparison. This one is silent and it will cost you.
Price gets checked against its own history.
- Q3
- is this worth it
- Q4
- is this actually cheaper than it was in October
Agents compare against past pricing. A discount that is not really a discount is now visible.
Occasion words replace spec words.
- Q3
- sterling silver, 18 inch chain
- Q4
- something elegant for my sister-in-law, under $150
“Elegant” is not in your catalog. That is the gap, and it is where the Q4 volume is.
Two separate checks before BFCM: is my catalog healthy, and are my listings answering this quarter’s questions.
Read: Your Catalog Has a Pulse →
You cannot get this view anywhere else
Everything above came out of our reporting, and it goes deeper than anything else on the market right now. Day by day agentic sessions, orders and gross sales. Splits per agent, so you can see ChatGPT against Gemini against Claude against Perplexity. Landing pages agents actually sent traffic to. All reconciled against your Shopify admin numbers, so nothing needs defending in a meeting.
Most brands going into BFCM cannot answer “how much did agents sell last week.” You will be able to answer it per agent, per day, per SKU.
Available on Pro → Reply and we will set yours up.
What we shipped
Bulk AI Suggestions. Fix up to 20 Product Listings in one pass. Category and category metafields, reviewed in one table, nothing written to your store until you save. Turns a two week cleanup into an afternoon.
Volume and Opportunity Score in Prompt Libraries. Every prompt now carries a search volume estimate and an Opportunity Score, with an Opportunity Map by funnel step. Work the high volume purchase-intent prompts first.
Export Suggested Changes to CSV. Current versus suggested values for every field you changed, in a second sheet. Review with the brand before anything touches the store, no product access needed.
Coming next: Shopify gives every store the same generic tools. We are building brand-specific ones on top, so an agent can ask your store about fit, compatibility, bundles or occasion and get a real answer. First ones go in this week.
The numbers from the top of this email
A US fine jewelry brand, heavy variation across stones, metals and sizes. Beautiful catalog, unreadable to agents. Attributes buried in prose, so agents recommended competitors instead.
We audited, remediated 2,055 listings in bulk, built a Prompt Library from 35 real shopper queries and published 620 Agentic Product Listings over four months.
Where it landed, measured over 77 days from June 1 to August 16:
- Catalog health went from 62 to 97 out of 100. Grade D to Grade A.
- 220 top-10 rankings.
- Agents drove 0.41% of traffic and 3.99% of orders. Roughly 10 times the conversion their traffic share predicts.
- Agentic AOV of $110 against a $96 store average.
- Monthly agentic revenue at $1,462, then $3,503, then $5,639 on August pace.
The volume is still small. The rate is not, and it compounds into Q4.
We will be in New York, September 14 to 17
I am coming in for Beanstalk. If you are there, I would like to meet. If you are not at the conference but you are in the city that week, that works too. Coffee, breakfast, a walk between sessions, whatever fits.
Reply and tell me which day works and I will come to you.
Last calls before Q4
Outside of New York, we are booking our final onboarding calls for the second half of September. After that we go heads down through peak with the brands already in flight.
If you want your catalog and prompts in shape before BFCM traffic hits, reply or find a time here.
Talk soon,
Tamir, Guy, Yoav and Elad
Founding Team @ 40rty.ai
