Group Detection

A family of four is one purchase decision, not four

KSI identifies the groups that walk in together, re-identifies them on the way out, and treats them as a single conversion unit. This dramatically changes how your real traffic, conversion, and customer base composition are measured.

Request my demo
KSI Vision · Visit composition
Live
Group visits
61%
+4.5% vs. yesterday
Avg. size
2.4
Group size
Families detected1 group = 1 decision
01The problem

Conventional sensors treat every person as an independent opportunity

A family of four walks into your store, browses, looks, and buys once. To a conventional sensor, that's four conversion opportunities and one sale - a 25% conversion rate, when reality is 100%.

The same happens with couples, friend groups, and customers with companions. Without grouping, every conversion, average ticket, and behavior metric is systematically underestimated or distorted.

Group = one purchase unit, not several visits

02How it works

Temporal grouping + re-identification

1. Proximity-based detection

KSI detects groups at access points by combining physical proximity with joined trajectories inside the store.

2. Anonymous re-identification

On exit, KSI anonymously re-identifies the group. No personal data stored.

3. Group-level dwell and conversion

Computes dwell time and conversion of the group as a single purchase unit.

03What it's used for

What working at the group level shows

Real conversion per group

Conversion rate per purchase unit - not per person - reflects what actually happens in your store.

Family / couple / solo segmentation

Compare behavior, ticket, and conversion across visit types.

Service sizing

Serving a family of four doesn't take the same time as serving a single customer. Plan staffing on real units.

Social and behavior analysis

Analyze how companions, kids, and group dynamics influence purchases and shopping center demand.

04Business impact

Metrics that reflect reality

~95% validated accuracy

Algorithm trained and validated against real datasets. Accuracy decreases at extremely dense entrances (>600 people/hour per entry).

Compatible with anonymous re-identification

Combined with re-ID, measures dwell and conversion at the group level without storing personal data. GDPR compliant.

Stop counting people - start counting purchase decisions

Request my demo

Share your case to get a personalized product demo.

* Required fields

Your data is handled carefully. By filling out and submitting this form you accept the Privacy Policy.