54,889 founders and marketers are getting this newsletter today.
Welcome to the 842 new operators who joined us this week! 🤯

Hey Fastlane Insiders! 👋

Thanks for spending a few minutes with me this week. Q3 closes Wednesday, which means the Super Bowl of ecommerce is officially on the clock. Deloitte expects US holiday ecommerce to grow 7.5% to 8.4% this season, roughly double the pace of retail overall.

By now, most of what decides your Q4 is already in place. If your holiday inventory is coming from overseas, it's either landed or on the water. Your offers are mostly set. What's still wide open is whether you'll be able to trust your own numbers when the dust settles in January.

I recorded this week's featured podcast with Bryan Mahoney, Glossier's former CTO. I keep coming back to it, because what he warned about then is exactly what Q4 is about to do to every dashboard on this list.

That's the thread through everything this week: answers that look right and aren't. An AI report with gorgeous charts and the wrong numbers. A third of shoppers regretting a purchase an AI talked them into. Google quietly switching on checkout inside AI Mode for eligible US Shopify stores, in a way that can make your conversion rate drop while your orders climb. And the Tuesday after Cyber Monday, your affiliate dashboard is going to look like a win whether it was one or not.

Whether you're at $30K months or $3M years, the fix isn't another tool. It's writing down what normal looks like before Q4 bends every number you have.

Here's what's inside this week's edition:

🎧 The Podcast: Why your AI doesn't actually know your business, my conversation with Bryan Mahoney that matters more now than the day we recorded it.

💡 Knowledge Drops: The five numbers to lock before Q3 closes, and the affiliate partners who bring customers you didn't already have.

🔥 Tool of the Week: Plytix, a free-to-start PIM that keeps your Shopify catalog under control.

📡 Industry Pulse: Google's default checkout switch, Amazon and Shopify splitting over Meta's shopping agent, the product ChatGPT keeps skipping, Sephora's TikTok-only launch, and the AI purchase regret numbers.

Let's dive in. 👇

Your abandoned cart flows only text shoppers who opted into SMS marketing. That's about 30% of your carts. The other 70% never subscribed, so your flows never reach them.

Recover does. Trained agents text those shoppers one-to-one in real conversations. They answer sizing questions, handle shipping objections, send a code, and win the sale back. No automation, no templates, no blasts.

It runs alongside the flows you already have. Nothing to rebuild, nothing to rewrite, five minutes to install on Shopify.

You only pay for results, backed by a 10x ROI guarantee.

Most brands "doing AI" right now are doing the same thing: exporting CSVs, screenshotting dashboards, dropping it all into a chat window, and prompting until the answer looks right. Sometimes it is. Sometimes it's confidently wrong, and only someone who already knows the business catches it.

I sat down with Bryan Mahoney, co-founder and CEO of Chord and former CTO of Glossier, where he spent 12 years building the data foundation behind one of Shopify's defining growth stories. Chord now does that work for brands like Sonos, Caraway, and Ruggable. His argument is one you rarely hear from an AI company: the model was never the bottleneck.

What we got into:

Under $2M, you don't need a context layer yet. But Bryan is specific about what a $2M to $5M brand should be building right now so it's ready later. Hint: it isn't another tool.

Picture the second week of January. Revenue up 18%, conversion up, AOV up. The plan worked. Or did it? Nobody can say, because nobody wrote down what normal looked like in September.

Comparing this BFCM to last BFCM feels rigorous. It isn't. Both periods are bent in the same direction by the same forces, and two numbers wrong in the same way look beautifully consistent.

What the piece walks through:

At $30K a month, this is three numbers and about 45 minutes. Above $2M, it's all five cut by channel. Either way, it's the cheapest insurance you'll buy this quarter, and the window closes September 30.

On Cyber Monday 2025, affiliates and partners drove 21.8% of US online revenue, per Adobe. More than a fifth of the year's biggest shopping day ran through somebody else's link. So how much of your share was new customers, and how much was a toll on demand you'd already created?

Coupon and discount sites took 42.4% of US affiliate revenue in the first half of 2025, mostly by showing up in the last few minutes of a checkout that was already happening.

Three calls the piece makes:

The piece has a clear call for every stage, from under $50K a month to $5M and up, plus which partner types can realistically still go live before Black Friday.

Everything above about product data being your new sales rep runs into the same wall: where does the enriched version actually live? Managing thousands of SKUs across multiple channels out of a spreadsheet doesn’t survive contact with Q4.

Plytix gives brands, manufacturers, retailers, and distributors a powerful PIM without the enterprise price tag, complexity, or long implementation. The Magic Importer pulls your entire Shopify catalog across in one click. From there, your team can manage, enrich, and optimize product content with AI and automations, safely review changes, and sync back to all of your Shopify stores and markets whenever you're ready.

Launch products faster, expand into new markets sooner, and finally keep your Shopify catalog under control.

Your best seller is often the hardest product in your store for ChatGPT to recommend. Nile co-founder Zhao Hanbo shares a ten-minute, two-query test that shows exactly where your top product drops out, and the five things most catalogs never write down.

Google may have already switched on checkout inside AI Mode and Gemini for your Shopify store. Eligible US merchants were enrolled by default this week, and some of your tracking didn't come along for the ride. Here's what breaks, and where to look before Q4 traffic lands.

Amazon locked Meta's Muse out of its store on September 20. Within a day, Shopify opened every one of its stores to the same agent. Retail just split on whether AI shoppers get let in, and your store picked a side without you.

Sephora is sending its own shoppers to TikTok. A $265 holiday set isn't sold on Sephora's website at all. When a retailer that size gives up its own checkout for a launch, it tells you where holiday discovery is heading.

A third of shoppers regret a purchase an AI talked them into, even though 76% used AI to buy something this past year. The number that should worry brands more is how many shoppers now suspect them of gaming AI answers.

Every story this week comes back to the same trap: an answer that looks right, delivered with total confidence. An AI report with the wrong numbers. A recommendation a third of shoppers regret. A dashboard the Tuesday after Cyber Monday that says you won.

You can't stop Q4 from bending your numbers. You can make sure you know what they looked like before it did. That's all context really is: what you know about your business that a dashboard or a model doesn't.

So here's a 20-minute assignment before Wednesday. Pull July 1 to today from Shopify and your ad accounts, and write down three numbers: blended CAC, AOV beside your discount rate, and the share of revenue from returning customers. If you're a US store, open Sales channels > Agentic in your Shopify admin, note whether direct checkout is on, and put today's date beside it.

Then hit reply and tell me your returning customer revenue share. I read and reply to every email, and I'm genuinely curious how wide the range is across this list. The patterns usually turn into a future Knowledge Drop.

…and thank YOU! 🙏

More than 53,000 of you give me a few minutes every Thursday, and that still amazes me. Some of you run brands, some build the apps and agencies behind them, and some are just watching this industry change faster than it ever has. If you only keep one thing from this edition, keep this:

Every dashboard, model, and partner report is about to sound more confident than it should. Your job is to know what true looked like before Q4 started talking.

Thanks for spending part of your Thursday with me. I'm grateful to be in the fastlane with you.

Cheers!
Steve