Retail Media · Kava OS Autopilot
The screen that sees the sale.
Most in-store screens count impressions and then lose the shopper. Kava OS runs vending machine advertising on the same device that rings up the purchase, so a campaign on a cooler door is measured against what came out of that cooler, in the same session.
- door-side display on Kava Cooler
- 10.1"door-side display on Kava Cooler
- touchscreen on Kava Vend V4
- 21.5"touchscreen on Kava Vend V4
- connected retail points
- 3,400+connected retail points
Vending machine advertising, in six formats
Cooler door display
The 10.1" display on Kava Cooler is the last thing a shopper sees before choosing. Short loops, no audio by default.
Vending touchscreen
The 21.5" touchscreen on Kava Vend V4 runs creative between selections and on the idle screen.
Sampling
Free or discounted trial units at chosen sites and hours, redeemed at the device and counted as they leave the shelf.
Brand-funded promotions
Discounts paid by the brand, executed through Dynamic Pricing rules with start and stop times.
Launch pilots
Place a new product in a set of devices for a fixed period and read real sell-through before a wider rollout.
Venue targeting
Choose by site type and hour: hospitals at night, campuses during finals, workplaces at breakfast. Places and times, not people.
Impression and purchase, on the same device.
Retail media's weak spot is the gap between the screen and the register. The ad plays in aisle four, the sale happens at a checkout three aisles away, and measurement means matching the two across weeks of loyalty data.
On an 8Kava device there is no gap. The device knows when the creative played, and the same device records what left the shelf in that session. Brands see lift against holdout devices and baseline periods, by site type and hour, in a read-only brand partner view. Shoppers stay anonymous: it's counts and baskets, never faces or profiles.
- Lift measured against holdout devices, not modeled.
- Live inventory, so a campaign can pause when its product sells out.
- Brand partners see their campaigns only, read-only.
A typical in-store screen vs. an 8Kava device
| Question | Typical in-store screen | 8Kava device |
|---|---|---|
| Did anyone see it? | Estimated from store foot traffic | Plays and sessions counted on the device |
| Did it sell anything? | Matched later, often across stores and weeks | Same device, same session |
| Is the product in stock right now? | Unknown when the ad plays | Live inventory per slot |
| Who sees results? | The retailer's report, eventually | A read-only brand partner view in Kava OS |
| Does the location host benefit? | Rarely | Revenue share on campaigns at their sites |
From brief to readout
Brief
Product, regions, site types, dates and the question you want answered: awareness, trial, or switching from a competing item.
Select devices and holdouts
Pick devices that match the brief, and keep a comparable set without the creative so lift has something to be measured against.
Stock it first
The product is added to the planogram on every campaign device. An ad for something you can't buy is just noise on a door.
Run
Creative plays on door displays and touchscreens on the schedule you set. Sampling and promotions run as device rules.
Read
Plays, sessions, units and lift against holdouts, by site type and hour, in the brand partner view.
Our brand ran a sampling campaign across 60 coolers and could see which screens actually drove purchases.
Your door, your rules, your share
A revenue share. Campaigns that run on devices at your site share revenue with you. The terms are set per program and written into your agreement before anything plays.
Category approval. You approve the categories that can run at your sites. A hospital may decide energy drinks don't belong on a ward cooler; a gym may want them front and center.
Quiet by default. Creative runs without audio unless you choose otherwise, and the price and payment information on the display always take priority over the ad.
What's the one product your shoppers would buy if they knew it was there?
Behind a glass door, a new item is invisible until it's familiar. Most launches in unattended retail don't fail because the product is bad. They fail because nobody tried it once. A sample at the device where it's sold turns a guess into a test, with the answer counted in units the same week. Which item on your shelf deserves a second introduction?
Related
For brands
Sampling, launch pilots and robotic storefronts.
Retail media on vending screens
Formats, closed-loop measurement and pricing models.
Shopper Analytics
The anonymous measurement behind every campaign readout.
Kava Vend V4
The smart vending machine with a 21.5" touchscreen.
Retail media questions
How is lift measured?
We compare sales of the advertised item on campaign devices against holdout devices and baseline periods, by site type and hour. Because the screen and the shelf are the same device, there's no cross-store matching or modeling in between.
Is any personal data used for targeting?
No. Targeting is by place and time: site type, region and hour. Recognition runs on the device, no images of faces are stored or sent, and results are aggregated.
Do location hosts get paid?
Yes. Hosts receive a revenue share on campaigns that run at their sites, on terms set per program. Hosts also approve which product categories may run.
What does a brand see?
A read-only brand partner view of its own campaigns: plays, sessions, units and lift. Brands don't see other brands' data or the operator's wider business.
Which plan includes Retail Media?
Retail Media is part of Autopilot at $69 per device per month. Operators on Autopilot can sell campaigns across their fleet; brands work with us or the operator directly. See pricing.
Where to go next
Test a launch where the purchase happens.
Tell us the product, the regions and the question. We'll come back with the devices, the holdouts and what a readout will show.