The “What Can I Actually Make Right Now” Problem Is the Most Underrated UX Pattern in E-Commerce
Cross-border sellers spend enormous energy on discovery — keyword research, ad creative, TikTok hooks, Amazon A+ content — and almost none on the moment after the customer has already committed. That moment is where returns are born, where reviews turn negative, and where repeat purchase rates quietly die. So when I see a small indie product built around inventory-aware matching — “tell me what you already have, I’ll show you what’s possible” — my ears perk up, even if the product itself has nothing to do with commerce. Kooler, a rebuilt cocktail app from designer Eduardo Duccigne, is exactly that pattern in miniature. And the pattern is worth stealing.
The thesis here isn’t “go download a drink app.” It’s that the inventory-first matching loop is the same loop that separates a Shopify store with 4% return rates from one bleeding 18%, and the same loop that makes Amazon FBA bundles convert. Kooler is a clean, small, free case study in how to build it — and in where the logic breaks.
What Kooler Actually Solves (and Why It’s Not a Recipe App)
Strip away the cocktail theme and Kooler is a constraint-solver dressed as a content library. Duccigne’s pitch is blunt: you add what’s in your cabinet to “My Bar,” and the app shows every drink you can make from it. There are 100+ curated recipes, spanning classics and Brazilian drinks most cocktail apps skip — Rabo de Galo, Macunaíma, Caju Amigo — plus clean recipe pages with consistent measurements and photos, check-ins to rate drinks you had at a bar, and community recipe submissions on the roadmap.
The pricing detail matters more than it should: it’s free, with no ads. That’s a deliberate anti-pattern in a category where subscription fatigue is the default complaint. One commenter, Emma, put it plainly: recipe apps that lock everything behind a subscription “get annoying pretty quickly.” Duccigne agreed. That’s not just generosity — it’s a positioning decision that removes the single biggest objection in the category.
But the real product isn’t the recipes. It’s the matching. Gal Dayan, who builds Dial, nailed the insight in the comments: most cocktail apps assume you’ll go buy the one missing ingredient, which means they “quietly become a shopping list app instead.” Kooler inverts that. It starts from your reality, not the recipe’s ideal. That inversion is the whole ballgame.
Why Amazon sellers should care more than Shopify ones
If you sell on Amazon Seller Central, you already live inside a constraint-matching engine — it’s called “Frequently Bought Together” and “Customers Also Bought.” But that’s Amazon matching its catalog to your cart, not your inventory to the customer’s need. The gap is where the opportunity sits.
Consider the bundle problem. A seller with 40 SKUs across three categories has, mathematically, hundreds of viable bundles — but most sellers build five, chosen by gut. An inventory-aware matcher, even a crude one, would surface the bundles that (a) share ingredients or components, (b) ship in the same box profile, and © don’t cannibalize each other’s reviews. That’s a spreadsheet exercise today. It shouldn’t be.
On Shopify, the same logic applies to “complete the set” merchandising. Most stores implement this as a manual collection. The stores that do it well — think Etsy sellers in craft categories, where buyers routinely own half the supplies already — treat inventory-awareness as a merchandising primitive, not a nice-to-have.
How Kooler Differs From the Incumbents (and What That Teaches You)
The obvious comparison set is Yummly and Paprika on the recipe side, and Cocktail Flow and Difford’s Guide on the drinks side. Paprika is a recipe manager — powerful, but it assumes you’ll go shopping. Cocktail Flow has a “what can I make” feature, but it’s buried and the ingredient taxonomy is loose. Difford’s is a reference library, not a solver.
Kooler’s differentiator is that the matching is the front door, not a filter. You open the app, you see what’s possible. That’s a structural choice with real consequences:
- It reduces decision fatigue. Instead of browsing 100+ recipes and self-filtering, you see a short list of things you can make right now. That’s the same reason Amazon’s “Buy It Again” widget outperforms the homepage for repeat customers.
- It creates a natural upsell loop. Once you know what you can make, the missing ingredient becomes a want, not a chore. A vermouth-shaped gap is a much warmer lead than a banner ad.
- It rewards inventory accuracy. The more you tell it, the better it gets. That’s a retention mechanic disguised as a utility.
Where the math breaks
Dayan’s edge-case question is the one that matters and, notably, went unanswered in the thread: when a recipe needs something close to what’s in your bar — a different vermouth, gin instead of a specific bitters — does Kooler surface a “one swap away” match, or is matching strict on exact ingredients?
This is the entire difference between a toy and a tool. Strict matching is easy to build and easy to break: the moment a user’s bar doesn’t exactly match any recipe, the app returns an empty state, and empty states kill retention. Fuzzy matching — ingredient substitution graphs, “close enough” scoring, confidence thresholds — is where the real engineering lives.
For cross-border sellers, this maps directly onto product recommendation engines. A strict matcher says “you bought a Nikon body, here’s a Nikon lens.” A fuzzy matcher says “you bought a Nikon body, here’s a third-party lens with a 92% compatibility score, and here’s why.” The second one converts. The first one is a filter.
What Cross-Border Sellers Can Borrow From This
Three transferable patterns, in order of how fast you can implement them.
1. Build the “My Bar” equivalent for your category
Every product category has an implicit “what you already own” state. Skincare sellers: what’s in the customer’s routine? Pet supplies: what’s the dog’s size, age, and current food? Coffee: what grinder, what brewer, what roast preference? If you’re not capturing that state, you’re recommending blind.
The cheapest version is a two-question quiz embedded in the Klaviyo welcome flow. The expensive version is a logged-in “my kit” dashboard. Both beat a homepage carousel. The Kooler lesson is that the state capture has to feel like a service, not a form. “Tell us what you have and we’ll show you what’s possible” converts. “Complete your profile” does not.
2. Treat the empty state as a product, not an error
When Kooler has no match, what does it show? The source doesn’t say — which is itself a signal that this is an unsolved problem in the app. For your store, the equivalent is the “no results” page in your on-site search, or the “you might also like” module when a customer’s cart is thin.
The rule I’d steal: never show a dead end. Show the nearest match, labeled honestly. “No exact match — but if you swap X for Y, here are three options.” That’s a fuzzy-match philosophy applied to merchandising, and it’s worth more than most A/B tests.
3. Price to remove the objection, not to maximize ARPU
Kooler is free with no ads, and the maker explicitly frames that as a deliberate choice. For a cross-border seller, the analog isn’t “make your product free” — it’s “identify the single objection that’s killing conversion and price it out of existence.” If free shipping is the objection, eat the margin. If returns are the objection, offer free returns on first orders. If subscription fatigue is the objection in your category, offer a one-time purchase option even if it cannibalizes LTV slightly. Emma’s comment is a data point: users notice when you don’t nickel-and-dime them, and they say so publicly.
Where My Judgment Says Kooler Falls Short
I want to be fair to a solo designer who rebuilt a 2010 project with AI tools, from the database to the iOS app — that’s a legitimate build story, and the Instagram presence suggests there’s a real audience. But the source material exposes three gaps I’d flag to any operator studying this as a template.
First, the iOS-only constraint is a strategic ceiling. The source mentions an iOS app explicitly. No Android mention, no web mention. For a utility that lives or dies on inventory accuracy, mobile-only means you can’t quickly add 20 bottles from a laptop after a grocery run. The friction is fatal to the core loop.
Second, community recipes are “next,” not shipped. Duccigne says recipe submissions are coming. That means the 100+ curated recipes are the entire content moat today. Curated is good for quality control — the consistent photos and measurements are a real differentiator — but it doesn’t scale, and it doesn’t create the network effects that make a utility defensible. Untappd won beer check-ins because the community generated the content. Kooler is pre-that.
Third, the matching logic is unverified. Dayan asked the right question and didn’t get an answer in the thread. Until we know whether Kooler does fuzzy matching or exact matching, we can’t judge whether the core feature is a real solver or a filter with a nice UI. I’d bet on the latter, because fuzzy matching is genuinely hard and the maker is a designer, not a developer — which is not a knock, just a prior.
The “AI-built solo app” caveat
There’s a subtext in this launch that cross-border sellers should read carefully. Duccigne built the whole thing with AI tools. That’s a real capability shift, and it’s why we’re seeing a wave of small, polished, single-purpose apps from non-engineers. But “built with AI” is not a moat. It’s a speed advantage. The moat is still the data — the recipe corpus, the ingredient graph, the user’s bar state. Any operator who mistakes “we shipped fast with AI” for “we have a defensible product” is in for a bad year.
What I’d Watch / Test Next
This week, three concrete moves.
One: Audit your own “empty state” surfaces — on-site search no-results pages, thin-cart recommendation modules, post-purchase upsell screens. For each, ask: does this show the nearest match, or does it show nothing? Fix the worst one. Anchor the fix to a Shopify metafield or an Amazon bundle rule, whichever fits your stack.
Two: If you sell consumables or components, build the “My Bar” quiz. Two questions, embedded in your Klaviyo welcome flow, feeding a segmented recommendation. Measure the delta in first-order AOV against your control. I’d expect 8–15% lift, but measure it, don’t assume it.
Three: Watch Kooler’s next 30 days. Specifically: does Android ship? Do community recipes arrive? Does Duccigne answer the fuzzy-matching question publicly? Those three signals tell you whether this becomes a real solver or a nice portfolio piece. Either way, the pattern — inventory-first matching as the front door — is the takeaway, and it’s one you can apply to your store before the app store even updates.
The sellers who win the next 18 months won’t be the ones with the best discovery. They’ll be the ones who solved the “what can I actually make/use/complete right now” problem better than everyone else. Kooler is a small, free, imperfect proof that the pattern works. Steal it.
