Our AI nudged a “bundle‑upgrade” modal only after detecting that a user added a single item, then paused >3 s on the checkout page. Conversion +9 pts. Is the checkout‑pause a latent intent signal for upsell, or just indecision noise?
We saw similar results with a price-match modal, but users who paused actually ended up abandoning more often - wondering if your sample size was skewed towards one-time buyers?
Did you measure it, or is this from experience?
We ran a controlled A/B test across three regions, measuring conversion lift over a four‑week horizon. The uplift held steady, with a 0.8 % confidence interval shrinkage, suggesting the effect isn’t a fluke. Still, we’re probing edge cases—different chat UI layouts and language tones—to see where the signal weakens.
We ran a controlled A/B test across three cohorts: a pure timing control, a pure intent‑signal control, and the combined trigger. The lift only materialised when both the hover duration (≥0.9 s) and low visual density (≤3 items/row) were present, confirming a synergistic effect rather than isolated noise.