Customer intent signals are the observable cues, drawn from behavior, product affinity, and customer profile, that predict how likely a person is to make a purchase or repurchase. In ecommerce they fall into three broad families: behavioral signals (what someone does, such as browsing, cart adds, email opens, and order frequency), product-affinity signals (which categories and price tiers someone gravitates toward), and profile signals (who someone is, including their email domain, shipping address, and identity type). Combine the three families and you can forecast which customers will buy again, and roughly when, instead of waiting for the next order to surprise you.
Purchase intent prediction for ecommerce works by scoring each customer against these signals and ranking them by likelihood to convert or repurchase. A first-time buyer who ships to an affluent zip code, uses a corporate email domain, and bought your highest-margin product carries far more predicted future value than a one-off discount-code buyer who never returns. Most Shopify merchants only see the second customer's order total and miss the first customer entirely. This guide breaks down each signal type, how to capture it, and how to turn it into retention revenue you can actually act on.
The Three Families of Intent Signals
Intent signals are easiest to reason about when you sort them into three buckets, because each bucket comes from a different source and answers a different question.
Behavioral signals answer "what is this person doing?" They include browsing sessions, time on product pages, cart additions and abandonments, email open and click rates, review submissions, and, crucially, order recency and frequency. Behavioral data is the backbone of classic retention modeling and the foundation of RFM segmentation, which scores customers on recency, frequency, and monetary value. Behavior is powerful because it is directly observed, but it has a cold-start problem: a brand-new customer has almost no behavioral history, so behavior alone tells you nothing on a first order.
Product-affinity signals answer "what does this person want?" They are derived from which categories, collections, and price points a customer engages with. Someone who repeatedly buys your premium line, adds full-price items rather than waiting for sales, and gravitates to a specific category has a clear affinity profile. Affinity predicts not just whether someone will buy again, but what they will buy, which is what makes it so useful for merchandising and email targeting.
Profile signals answer "who is this person?" This is the family most merchants neglect, because Shopify does not surface it natively. Profile signals include the email domain (a personal Gmail versus a corporate or agency domain), the shipping address and its neighborhood, social and professional presence, and identity type such as founder, investor, press, creator, or affluent buyer. Profile signals solve the cold-start problem that behavioral data cannot, because they are fully available on the very first order, before any browsing or repeat-purchase history exists. For a deeper look at the raw inputs, see our guide to identity data in ecommerce.
Why Profile Signals Beat Behavior on the First Order
The hardest moment to predict repeat purchase is the first order, and it is also the most valuable moment to get right. Behavioral models need history they do not yet have. This is exactly where profile-based intent signals shine, and it is the core idea behind first-order VIP detection.
Consider two new customers who both place a $120 order. Customer A used a Gmail address, shipped to a mid-tier suburb, and redeemed a 20 percent welcome code. Customer B used an email at a venture firm's domain, shipped to one of the most affluent zip codes in the country, and paid full price. Behaviorally they look identical at checkout. By profile, they could not be more different. Customer B carries dramatically higher predicted lifetime value, higher repurchase probability, and the potential to become a partner, an investor, or a press mention. A model that ignores profile signals treats them as twins. A model that uses profile signals flags Customer B for a hand-written thank-you and a VIP track.
The shipping address is the single richest profile signal because it reflects where someone actually lives, which is a strong proxy for buying power. SonarID weights the shipping address over the billing address for exactly this reason, since the residence is the better signal of disposable income and long-term value. Our guide to address verification in enrichment explains how that signal is validated and scored.
How These Signals Combine Into a Prediction
No single signal is decisive. The art of purchase intent prediction is combining signals so that strong cues reinforce each other and weak cues get discounted. A practical scoring approach layers the three families.
Start with the free signal layer, which costs nothing per lookup: email-domain classification, spend and order-value analysis, and affluent-zip matching. This layer alone separates the obvious high-intent customers from the crowd. Then add behavioral weighting from order recency and frequency. A customer who bought twice in 60 days at full price is signaling strong repurchase intent. Finally, for the customers who score high on the free layer, add paid enrichment to resolve the full identity and confirm whether the corporate domain belongs to a founder, the affluent address belongs to a public figure, or the buyer is a creator with real reach.
This tiered logic mirrors how a predictive scoring model forecasts which customers will become VIPs. The goal is not a single magic number but a ranked, explainable list: here are the customers most likely to buy again, here is why, and here is what they are likely to buy. For a deeper treatment of building the model itself, see our guide to predictive VIP scoring.
Turning Signals Into Retention Revenue
Intent prediction only matters if it changes what you do. The whole point is to act before the next purchase decision instead of reacting after it. Here is how merchants convert intent signals into retention.
Each of these actions depends on knowing who is likely to buy again before they do. That is the practical payoff of intent signals: not a dashboard you admire, but a sequence of timely, personalized moves that lift repeat rate and lifetime value.
Where Repeat Behavior Confirms the Prediction
Profile signals predict on the first order, but behavior confirms and refines the prediction over time. As a customer accumulates a history, fold that history back into their score. Two paid orders at full price within a quarter is a stronger repurchase signal than any profile cue. This is where repeat customer analysis becomes essential, because it tells you which customers have actually demonstrated loyalty rather than merely looking likely to.
The most accurate intent picture comes from running profile and behavior together. Profile gives you an instant read on a stranger. Behavior validates or corrects that read as the relationship matures. A customer who scored high on profile and then bought three more times is your confirmed VIP. A customer who scored high on profile but never returned tells you something too, perhaps that the product was a one-time gift. Tracking both keeps your predictions honest and your retention spend efficient.
Putting Intent Signals to Work Without a Data Science Team
Most Shopify merchants do not have a data scientist to build a custom intent model, and they do not need one. The signals described here are already sitting in your order data; the gap is that Shopify does not enrich, score, or rank them for you. This is the gap SonarID closes. It reads each order in real time, applies the free signal layer of email-domain matching, spend analysis, and affluent-zip matching, then enriches the high-potential customers at $0.05 per enrichment to confirm exactly who they are. Every plan carries a concrete enrichment cap, so the cost is always knowable.
The result is a continuously updated, ranked view of who in your customer base is most likely to buy again and most worth your attention, surfaced on a VIP dashboard with real-time alerts to Slack and Klaviyo the moment a high-intent customer orders. If you want the broader strategic context, our overview of customer intelligence explains how these signals fit into brand growth. Intent signals are not a forecasting toy. They are the difference between guessing who matters and knowing, on the very first order, who is going to come back.