# Recommendations for first time visitors: useful suggestions when there is no history to personalise from | KilterMuse Source: https://kiltermuse.com/blog/2026-10-23-recommendations-first-time-visitors-no-history/ KilterMuse answers October 23, 2026 **Short answer:** A first time visitor is not a blank slate. They arrive with a landing product or category, a search term or referring context, a device, a rough location, and within seconds, a pattern of viewing. Good first visit recommendations use exactly those signals, in that order of reliability, and resist pretending to know more. The goal is modest and crucial: be visibly relevant once, so the visitor grants the store a second session to learn from. ## Start from the landing context The strongest first visit signal is what the visitor came for: the product or article they landed on and whatever query or campaign brought them. Recommendations anchored tightly to that context, same category, complementary items, the products other buyers of this item genuinely needed, outperform any storewide popularity row, because they continue the visitor's own sentence instead of starting a new one about the store's priorities. Referral context sharpens the anchor when it is available honestly: a visitor from a review of a specific product line, a gift guide, or a comparison page has declared an intent the recommendations can serve. Keep the use proportionate and invisible. The visitor should feel the store is relevant, not that it has been reading their journey. Relevance feels like service. Surveillance feels like a reason to leave, and the line between them is presentation. ## Use the session as it builds Within a handful of views, a first session develops a shape: a price frame, a style direction, a size or specification interest. Recommendations should update from that shape quickly, narrowing rather than restarting with each click. The common failure is the opposite, every product page showing its own isolated row, so the visitor who has viewed three items in one clear direction is still being shown the generic row on the fourth, as if nothing had been learned in the last four minutes. Cart and search behaviour are the highest intent in session signals and should override browsing inference immediately. A first time visitor who searches a specific term or adds an item has handed the store a brief. Recommendations that ignore the brief to keep pushing the landing category's row are not personalisation failing for lack of data. They are personalisation ignoring the data it was given, which is worse, because the visitor can see it happening. ## Bestsellers, with adult supervision Popularity data is the fallback, and it should be a curated fallback, not a raw feed. Bestsellers within the visitor's current category, filtered for availability, season, and price coherence with what they are viewing, give a first session something proven to stand on. Raw storewide bestsellers drag a visitor researching a specialist product into the store's most generic lines and announce that the system knows nothing about why they came. Label honestly where it helps. A row that says popular with buyers of this item, when that is true, earns trust a silent row does not, because the visitor can judge the basis of the suggestion. As the visitor's own behaviour accumulates, the basis should shift visibly from popularity to their session, so by the second visit the store's memory feels earned rather than assumed. People accept being learned about when they can watch the learning happen from things they actually did. ## Judge first visit recommendations by return The success metric for a first session is not only immediate conversion, which will always be modest, but return: did the visitor come back, engage deeper, or buy on the second visit at a better rate than an unrecommended control. Stores that judge first visit rows purely on same session clicks optimise them into clickbait, bright, broad, and forgettable, and then wonder why first impressions do not compound into a customer base. Keep the system humble in configuration: strong contextual anchor, fast session learning, curated popularity fallback, honest labels. That stack will not demo like magic, and it will outperform magic steadily in the only place that matters, the second visit rate. Personalisation's hardest problem was never the hundredth visit, where data is rich. It is the first ten minutes, where a store either proves it is paying attention or proves it is not, to a visitor deciding whether there will ever be a second chance to learn more. Keep reading: [Price sensitivity in recommendations: when showing the expensive option helps and when it kills the click](https://kiltermuse.com/blog/2026-10-23-price-sensitivity-recommendations), [Replenishment recommendations: reminding repeat buyers at the moment they actually run out](https://kiltermuse.com/blog/2026-10-23-replenishment-recommendations-repeat-buyers), and [Context aware recommendations: using season, weather, and place without being creepy](https://kiltermuse.com/blog/2026-10-23-context-aware-recommendations-season-place) cover the nearby cases in this series. ## Frequently asked questions ### What signals does a first time visitor provide? The landing product or category, the search or referral context that brought them, device and rough location, and within seconds their viewing pattern, price frame, searches, and cart actions. Used in that order, they support genuinely relevant suggestions. ### Why do generic bestseller rows fail first visits? Because they ignore why the visitor came. A specialist researcher shown the store's most generic lines learns the system knows nothing about them. Popularity works only as a curated fallback inside the visitor's category, price frame, and season. ### How fast should recommendations learn within a session? Within a few views. As a price frame and style direction emerge, rows should narrow to reflect them, and search or cart actions should override browsing inference immediately, since they are the clearest statement of intent. ### How should first visit recommendations be measured? By return and second visit behaviour as well as immediate clicks: did visitors come back and engage or buy more than a control group. Judging only on same session clicks optimises the row into forgettable clickbait.