Sometimes It's Faster to Just Ask

01 /The Concept

The best salesperson in a store doesn't guess what you want – they ask. Online retail has spent two decades trying to infer preferences from behavioral signals and getting it consistently wrong. Wyng takes a more direct approach: when you want to know what a customer prefers, just ask them.

Wyng is a zero-party data platform that helps e-commerce operators collect explicit customer preferences and use them to drive personalized recommendations across the buying experience. Zero-party data (ZPD) is information users voluntarily provide about themselves – a meaningful distinction from the other data types retailers typically rely on.

First-party data captures what users do on a site: categories browsed, products purchased. But behavioral signals are famously ambiguous. A skincare purchase might indicate the buyer's own preferences, or it might be a gift for someone with entirely different needs. Second-party data is someone else's first-party data, bought or licensed. Third-party data is sourced from external databases and enrichment services – demographic inferences rather than stated facts.

Wyng cuts through the noise by asking the right question at the right moment. A visitor browsing skincare gets asked their skin type – shown as a short illustrated list – and immediately receives a curated selection and a purchase-day discount. The timing, question selection, and framing are optimized by the platform. Responses update recommendation blocks across all pages of the site in real time, and users can view and edit their preference profile at any point.

The data flows downstream too. Preference tags can be passed to email platforms like Mailchimp, enabling segmented campaigns built on what customers actually told you rather than what their click history implies. Integration is API-based and takes hours, not weeks.

The numbers back it up. L'Oreal more than doubled average order value using Wyng. Transaction conversion rates and email click-through rates both roughly doubled in documented cases. Wyng has raised $37 million across eight rounds, with annual revenue estimated at tens of millions of dollars.

02 /Why It Matters

Average basket sizes in physical retail consistently run higher than online – and a large part of the gap is the skilled sales associate who asks the right question and makes a recommendation you can't refuse.

E-commerce never replicated that dynamic. The pop-ups and banners that dominate online retail mostly irritate rather than assist. Recommendation engines built purely on purchase history get caught in feedback loops: a playlist dominated by 80s music after one party session keeps surfacing whether you want it or not.

The underappreciated insight here is that a well-designed question is itself a form of personalization. The act of asking signals attentiveness. The answer removes ambiguity that behavioral inference can't resolve. Wyng's results – doubling order values at the same price points and inventory – are hard to explain any other way.

A [recent review](/review/dumaete-im-vsjo-ponjatno) covered Sprig, which built a similar premise for product research: asking users direct questions at contextually relevant moments and gathering enough responses in hours to be statistically meaningful. The underlying principle is identical – knowing when and what to ask is more valuable than smarter inference.

The broader pattern: recommendation systems that only predict will keep hitting their ceiling. Systems that predict and occasionally ask – and know which moments call for which – compound their accuracy over time in a way that pure behavioral models cannot.

03 /Opportunities

Growing e-commerce revenue is one of the most contested spaces in B2B SaaS. The recommendation layer is a well-established category, and Wyng's closest competitors approach the cold-start problem differently.

Depict, [covered earlier](/review/podsunut-i-prodat), trains its recommendation engine on data from tens of thousands of stores, so even a small merchant gets recommendations calibrated against a large-scale behavioral corpus from day one. It's a scale-as-moat approach: aggregate enough behavioral data and individual stores don't need much of their own.

Wyng's angle is complementary rather than competing: replace inferential guesswork with direct stated preferences, particularly for attributes that behavior can't reliably reveal – aesthetic preferences, skin type, dietary needs, intended recipient. Neither approach makes the other redundant; they target different failure modes in the same recommendation stack.

The immediate entry point for anyone building in this space: categories where behavioral signals are systematically misleading. Gift retail, personal care, specialty food, and professional services all share the property that the buyer and the actual end-user are often different people, or that personal attributes not captured in purchase history determine what the right product actually is. In those categories, a well-timed question does work that no amount of click history can.