Shopper Analytics · Kava OS Pro
Learn what people want without learning who they are.
Shopper Analytics is micro market analytics built from anonymous signals: how many sessions start, how long a door stays open, how often someone walks away empty-handed, and what ends up in the same basket. No faces and no profiles. Just the patterns that decide your next planogram.
- images of faces stored or sent
- 0images of faces stored or sent
- transactions in the last 12 months
- 11.8Mtransactions in the last 12 months
- per device per month on Pro
- $39per device per month on Pro
Micro market analytics in four measurements
Sessions
Taps at a cooler, scans at a kiosk, entries into a Kava Box, by hour and day. Counted, never identified.
Dwell
How long a door stays open, or a shopper stays in the store, before choosing or leaving.
Conversion
Sessions that end in a purchase vs. sessions that end empty-handed. The second number is the interesting one.
Baskets
What sells together, at which hours, at which sites. Pairings decide what belongs on the same shelf.
Counted, not identified.
Every measurement on this page is produced on the device and leaves it as a number. Recognition runs on the hardware's own processor and looks at products. No images of faces are stored or sent, and there is no facial recognition anywhere in Kava OS.
Reports are aggregated by device, site and time window. They answer "what do night-shift shoppers buy?" and are built so they can't answer "what did this person buy?" That trade-off is deliberate. Planograms don't need identities, and shoppers shouldn't have to wonder.
- Aggregated patterns only; no individual shopper profiles.
- Data hosted in US regions, encrypted in transit and at rest.
- Role-based access decides who sees which reports.
One example, start to finish
A lobby cooler in a hospital. The numbers are illustrative; the method is the point.
Spot the gap
Sessions between 2 and 4 a.m. are high, but a third of them end empty-handed. Daytime conversion at the same cooler is fine.
Read the baskets
Night buyers pair hot-ready meals with an energy drink. The eye-level shelf holds salads, which sell at lunch and sit untouched overnight.
Test on one device
Swap the meal and salad facings on this cooler only, and pick a similar cooler on another campus as the comparison.
Measure
Compare night conversion and revenue per session against the prior four weeks and against the comparison cooler.
Roll out
If it holds, push the planogram to similar night-shift sites. If it doesn't, roll it back and test the next idea.
What a sales report says vs. what Shopper Analytics says
| Question | Sales report | Shopper Analytics |
|---|---|---|
| Why did cold brew sales drop? | It shows the drop | Whether sessions fell, conversion fell, or the slot was empty |
| Is this site worth a second cooler? | Revenue per day | Peak-hour sessions, dwell and empty-handed walk-aways |
| What belongs at eye level? | Top sellers overall | Top sellers by hour and by basket pairing |
| Did the new planogram work? | Before-and-after revenue | Conversion per session against a comparison device |
A review rhythm that fits a busy week
- Every Monday
Check the walk-aways
Scan the ten devices with the most empty-handed sessions. Check stock first: most walk-aways turn out to be an empty slot.
- Every month
Run one facing test per region
One change, one comparison device, four weeks. Keep the ones that hold and write down the ones that didn't.
- Every quarter
Prune by site, not by fleet
Retire slow items site by site. A slow seller fleet-wide may be the reason people visit one particular cooler.
- New sites
Check the opening assortment
Compare hour-by-hour sessions with the Site Intelligence forecast during the first weeks and adjust the planogram before habits form.
What's measured on each format
Kava Cooler C2 / C2-XL
- Measured
- Sessions, door-open time, empty sessions, baskets
Kava Box B160
- Measured
- Entries, time in store, baskets across up to 700 SKUs
Kava Kiosk K1
- Measured
- Checkout sessions and baskets, including unpackaged items like fruit and bakery
Legacy machines with Kava Link
- Measured
- Sales by item and hour, time out of stock, payment mix
The most useful number in your report is the sale that didn't happen.
A shopper who opens the door, looks for a while and closes it again is telling you something no sales report can. Maybe their favorite was out. Maybe the price was wrong for 3 a.m. Maybe what they wanted was never on the shelf. Sales data only describes the people you already satisfied. Which of your sites has the most empty-handed sessions, and what would you stock if you knew why?
Shopper Analytics questions
Do you use facial recognition or store video of shoppers?
No. Recognition runs on the device and is trained on products. No images of faces are stored or sent. What leaves the device is counts, durations and baskets.
Can a report identify an individual shopper?
No. Shopper Analytics reports patterns across devices, sites and time windows. Payment and loyalty records live in their own systems with their own access controls and aren't joined into shopper profiles.
Does it work on legacy vending machines?
Partly. Machines connected through Kava Link report sales by item and hour, time out of stock and payment mix. Sessions, dwell and empty-handed walk-aways need 8Kava sensing hardware.
How does this change my planograms?
It replaces "top sellers overall" with "top sellers for this site, at this hour, next to these items." Most changes are small: a facing swap, a slot moved to eye level, a night-only item. You test on one device before rolling out.
Which plan includes it?
Shopper Analytics is part of Pro at $39 per device per month, or $33.15 with annual billing. Try it free for 30 days on up to 10 devices.
Where to go next
Find the sale that didn't happen.
Book a demo and we'll show you the session, dwell and basket views on a live device, and what they say about a location like yours.