Put this into practice
Run A/B tests and personalize your store without code. 14-day free trial, no credit card.
Start free trial →Run A/B tests and personalize your store without code. 14-day free trial, no credit card.
Start free trial →Real-time personalization adapts the experience during the current session, responding to what a visitor is doing right now rather than to a profile assembled from past visits. A shopper who has viewed three running shoes in four minutes sees running-specific content on their next page view — not on their next visit.
The distinction from segment-based personalization is one of latency. Segment-based approaches assign visitors to audiences built from historical data and serve each audience a prepared experience. Real-time personalization reacts within seconds, which means it works on first-time visitors who have no history at all.
The majority of visitors to most D2C stores are new, anonymous, and will not return. Personalization that depends on purchase history or a logged-in profile simply does not apply to them. Yet these visitors generate intent signals continuously — what they search, which filters they apply, how long they linger, what they compare.
Acting on those signals within the session is the only way to personalize for the largest share of your traffic. A first-time visitor who filters a category by "under ₹1,000" has told you something specific and immediately useful. Continuing to show them ₹3,000 products for the rest of the session ignores information you already have.
Intent also decays quickly. Someone researching a purchase now may be gone in ten minutes. The value of knowing they are interested in a category is highest during the session and much lower a week later, which is why session-level reaction outperforms next-visit retargeting for this group.
The trade-off is that real-time signals are noisier than historical ones. A single product view is weak evidence; three views in the same subcategory is much stronger. Reacting too eagerly produces an experience that lurches around based on accidental clicks.
A multi-category personal care brand found that visitors arriving on a hair oil product page from search frequently browsed two or three more haircare items and then left without buying — a pattern suggesting research rather than rejection.
They implemented a session-level rule: once a visitor viewed three products within one category, the page adapted to show a comparison view of those items, the category's most-reviewed product, and a "which one is right for me?" guide.
Conversion rate for visitors who triggered that rule rose from 1.2% to 2.6%. The trigger threshold mattered — an earlier version that activated after a single product view performed no better than the control, because one view carries almost no information.
Test the rule, not the concept. For each trigger — three views in a category, a price filter applied, a cart with a single low-value item — randomly assign qualifying visitors to the personalized experience or the default, and measure only within that qualified group.
This matters because visitors who trigger a rule are self-selected high-intent shoppers who would convert better regardless. Comparing them to overall site performance will make any rule look brilliant. CustomFit.ai evaluates behavioural rules in real time, renders without flicker, and reports each rule as a controlled experiment against its own qualified control group.
Run smarter A/B tests with CustomFit.ai — 14-day free trial, no credit card required.