Justin Aronstein

The most expensive BIN in a subscription brand's chargeback-alert program barely showed up in their actual chargebacks. Nobody noticed until they broke the alert report down by BIN for the first time. One BIN was eating 26% of their entire alert spend. Not 26% of revenue. 26% of the alerts they were paying for, all on a single issuer's cards. So we looked at what those alerts were actually preventing. Roughly 1,000 alerts fired on that one BIN over the window. Behind them: 28 chargebacks. A thousand alerts. Twenty-eight chargebacks. They were paying per alert, five figures a month on that BIN alone, to deflect a number of disputes they could have absorbed for a fraction of the cost. This isn't an alert provider failing. The alerts were real. The issuer behind that BIN just over-triggers. It flags suspected-fraud cases the cardholder never disputed, and every flag bills you. The program was sold as protection. On that BIN, it was a subscription to noise. The fix wasn't ripping out alerts. It was laying alert count next to chargeback count, per BIN, and finding the BINs where the ratio had inverted. On the worst one, the cheaper move was to stop paying for alerts entirely and just eat the handful of chargebacks that came through. Most operators audit their alert program by total spend and total deflection. Both numbers looked fine here. The leak only showed up one BIN at a time. Pull your alerts by BIN against chargebacks by BIN before your next renewal. The BIN eating a quarter of your alert spend is almost never the BIN behind a quarter of your disputes. #Chargebacks #Payments #Subscriptions

4

"But Kyle, the spreadsheet said our conversion 'has to go up.' Shouldn't this be something owned by Product Marketing?" I’ve seen this play out more times than I can count. And I've been responsible for guiding people in the wrong direction more than once. Here's what usually happens: Leadership builds a model, the model spits out a number, and suddenly, “Stage 1 to Stage 2 needs to improve by X%.” So we hand that number to a team (PMM) and tell them to go make it happen. Completely the wrong way to go about it. The first mistake is assigning a number to someone who can’t actually control it. Product marketing gets this all the time. Close rate. Late-stage conversion. Revenue that depends on pricing, sales execution, product gaps, competitive pressure, and a dozen other variables. PMM can influence outcomes. They don’t own the entire system. When we pretend they do, they end up getting blamed for things they didn’t cause and doing random work to prove they’re “helping.” Which leads to the second issue. Everyone runs in different directions. One person rewrites messaging. Another rebuilds the deck. Someone else launches a campaign. It all feels productive, but none of it is coordinated. EOQ shows up, and someone asks what actually worked, and the honest answer is usually “we’re not sure.” Then comes the third problem. Product marketing gets pulled apart. Sales wants enablement. Product wants launches. CS wants customer stories. Everyone has a reasonable ask. So PMM gradually becomes the highest-paid request taker in the company. So how do you fix it? Here's what has worked for me in the past. Especially after The fix isn’t complicated, but it does require you to ACTUALLY DO IT. Pick a North Star. One metric everyone agrees matters most right now. Then be explicit about what you’re trying to change, why you believe it’s the right lever, and what specific projects you’re going to run to move it. Conversion rate might be it, and that's okay! One more thing that matters more than people realize: don’t report the metric every week. If you have a long sales cycle, weekly conversion updates are a waste of time. Report the work weekly. What shipped, what you learned, what you’re changing next. Look at the metric quarterly. Models are useful. They tell you where you want to go. They don’t tell you how to get there. That part is still on us. Ya know, the humans. ;)

80

14 Comments

The "Tech vs. CPG" Perspective. Why CPG websites should be built like Software. In tech, we obsess over "latency." Every millisecond of delay is a drop in user retention. In CPG, the biggest latency isn’t page load speed. It’s Cognitive Latency. It’s the time it takes for a customer to ask: “Is this for me?” “What does it actually do?” “Why is it $60?” Most brand websites we reviewed last month at Veratempo Product Lab are beautiful, but they are "High Latency." They make the customer work too hard to buy. We’ve found that 40% of conversion drops happen because the "Add to Cart" button is competing with a 3-paragraph story about the founder's childhood. Storytelling builds the brand. But Clarity builds the bank account. Our Conversion Snapshots aren't about making your site prettier. They are about reducing the cognitive latency so your traffic actually flows. Want us to look at your "Latency"? Comment SNAPSHOT below. #ConversionRate #SystemsDesign #CPG #GrowthOps

3

🚨 This seems to have slid a bit under the news radar today. Walmart's products will now be purchasable directly through ChatGPT using Instant Checkout (as well as Sam's Club products!) Just 2 weeks ago there was a study published showing 20%+ of Walmarts referral traffic was coming through ChatGPT searches and climbing rapidly. No surprise that Walmart wants to capitalize on this partnership now in the same way that Etsy and Shopify have already done so for their sellers. There was a big question over the past year of if AI Search & LLMs would disintermediate retailers in the process and this is a clear picture that it WILL NOT. ChatGPT is the top of funnel that will drive traffic and enable better purchases TO the retailers that they partner with - not cut them out completely. The reality is capturing customer intent is only step one of the purchase lifecycle, and OpenAI doesn't want to deal with the intricacies that follow suit (payments, supply chain & fulfillment, returns, etc). This is exactly what retailers are great at and have build deep expertise in. If you're a retailer / brand and you're not thinking about how to feed your catalog and make it visible to ChatGPT and Instant Checkout, you're going to miss out a huge wave of growth - especially if you're seeing decline from your organic google search traffic. We've already published a field guide for our customers on what they need to do NOW to prep their catalogs for Instant Checkout - DM me for it and get started right away or comment "Guide" below and I'll shoot it over to you.

42

10 Comments

Midweek musings: I spend a lot of time talking to retailers about agentic commerce, data structure/feeds and onsite agents often come up in conversation. Retailers broadly share that feed investment is positive/high ROI and my pov is you can't monetize if you don't show up. 💬 Retailers are also increasingly exploring AI to make the site experience more interactive (think natural language prompts). As the bar for the onsite customer experience continues to rise, here are a few insights from large-scale AI agent deployments at the MegaCaps: 🛒 Walmart’s Sparky is driving 35% higher average order values by solving "contextual problems"—like planning a themed birthday party—rather than just finding single items. 📦 Amazon’s Rufus delivered nearly $12B in incremental sales last year; CEO Andy Jassy highlighted the "Buy For Me" feature, which allows the assistant to shop tens of millions of items and purchase them on behalf of the customer. 🛠️ The Home Depot's Magic Apron is seeing an 8% lift in online comp sales by transforming into a project companion that summarizes thousands of reviews and provides aisle-level precision for in-store wayfinding. Consumers enjoy and will expect convenience at their fingertips across channels. Personalization is becoming real. At Stripe, we’re helping the world’s most innovative retailers prepare for the future of commerce, on their own site and with AI agents. #agenticcommerce #stripe #stripeforretail #data

55

7 Comments

Quick tip for AI automation builders: Stop letting your sales team handle first-response scheduling. If you’re building AI tools for healthtech or sales, you know the bottleneck isn’t usually the tech—it’s the handoff. When a lead is ready to talk, every minute of delay drops conversion rates significantly. Here’s the actionable move: Automate the “next best step” immediately after a lead is captured. Don’t just log the contact; trigger a scheduling link or a personalized follow-up call within 5 minutes. This shifts the burden from human speed to system reliability. Eric Porter — SmithRx, you’re deep in the AI business automation space, so this is likely a pain point you’ve seen in your own operations or your clients’ workflows. The gap between “lead captured” and “meeting booked” is where most revenue leaks out. Lieutenant James Henderson handles this end-to-end. We automate phone answering, lead capture, and scheduling so that the moment a prospect engages, the system acts. No manual triage, no missed calls, no waiting for a human to check their inbox. It’s designed to save you roughly $1,400/month in admin overhead and reclaim hours your team can spend on high-value work. Setup takes about two weeks. If you want to see how we handle the operational side of AI automation, learn more here: https://lnkd.in/euvUEfkF #AIautomation #HealthTech #SalesOps #SmallBusiness

1

eCommerce brands doing millions in revenue Still losing money? - Not because of bad products   - Not because of low traffic But because -Backend quietly bleeding - Hidden fees eating margins   - Broken funnels losing customers   - Leaky checkout flows   - Inventory mistakes   - Refunds piling up - Looks like growth from the outside - Profits shrinking internally Fixing 2–3 backend issues. Can recover more profit than spending $10k on ads Running an eCommerce brand? Numbers feel off? DM me for a free setup review   - No pitch   - No deck

9

7 Comments

Marketers, there is no excuse not to prepare before you reach out. There never was but it is especially more pronounced in the AI era. Read chapter 9 of my first book, “The Entrepreneur’s Essentials”, for free - link in comments.

13

4 Comments

Here's a stance that gets me disagreement from CRO folks I respect. Your friction-reduction backlog might be quietly costing you conversions. Every optimization team has one. Shorter forms. Faster load. Fewer clicks. Cleaner nav. Worthy work, most of it. But it runs on an assumption borrowed from e-commerce. That friction is always the enemy. Remove it wherever you find it, and people move forward. For high-consideration purchases, "reduce friction" is the wrong playbook. When someone is choosing a health plan, a financial advisor, or a $5,000 trip, some of that "friction" is the customer doing exactly what a hard decision requires. Slowing down. Comparing. Re-reading the policy. Sitting with it before they commit. Strip that out and you don't speed them to yes. You speed them to a decision they're not ready to make, which usually means no. So before anything goes on the backlog, it should have to answer one question. Is this procedural friction, the kind that wastes the customer's effort, or cognitive friction, the kind the customer needs in order to feel sure? Remove the first. Design around the second. Treat them the same and good optimization programs plateau for years while every test comes back flat. So I'm curious, and I mean this as a real question, not a gotcha. Look at what's on your optimization backlog right now. How many of those items are removing friction the customer actually needed in order to trust you?

9