Key takeaways
- Size a site in daily transactions, not headcount. Headcount rules of thumb break at hybrid offices and 24/7 sites.
- Unsupervised hours push you toward locked formats. Open shelves at 3 a.m. are an honor system with nobody around to be honorable for.
- Assortment sets the floor: snacks and drinks fit a spiral, fresh meals want a cooler, a full pantry with fruit and coffee needs a micro market.
- Restocking visits often cost more than the hardware, so model the route before you pick the box.
- Most strong sites end up hybrid: a kiosk for breadth, a locked cooler for high-value fresh food.
Most format decisions are made backward. Someone sees a sleek smart cooler at a trade show, buys three, and then goes looking for hallways to put them in. Six months later one is a star, one breaks even, and one is a very expensive refrigerator full of expired wraps. The format should fall out of the site, not the other way around. Four variables do most of the work: how many transactions the site can produce, when people are there, what they want to buy, and how much you can trust the room. A fifth, the cost of servicing it, quietly decides whether the first four add up.
The three formats, described honestly
Before choosing, be blunt about what each format is good and bad at.
Spiral vending machine. Product sits behind glass on coils. The shopper pays, the coil turns, the item drops. It has the smallest footprint per selection, the lowest shrink of any format, and well-understood servicing. It’s limited to products that survive a drop and fit a coil, and its fixed capacity per coil means best-sellers sell out first. A connected machine with a drop sensor can refund automatically when nothing falls, which removes the most common complaint.
Smart cooler. A locked, refrigerated cabinet. The shopper taps a card, phone or badge, the door opens, they take what they want, and they’re charged when the door closes. Sensors (computer vision, shelf weight, RFID or a mix) work out what left. Because the shopper holds the product before buying, coolers handle fresh meals, salads, dairy and odd-shaped packages well. The door doesn’t open without a payment credential, so shrink stays low.
Micro market. Open shelves, open-front coolers and a self-checkout kiosk. It offers the widest assortment, including fruit, bakery, coffee and frozen items. It’s also an honor system: nothing physically stops someone walking out without scanning. That’s fine in a room with a strong sense of community and steady traffic. It’s expensive in a room that’s empty 14 hours a day.
| Spiral vending | Smart cooler | Micro market | |
|---|---|---|---|
| Typical assortment | Dozens of selections, mostly packaged snacks and drinks | Up to a few hundred items, fresh-friendly | Hundreds of SKUs, including unpackaged items |
| Fresh food | Limited; needs a chilled zone and drop-tolerant packaging | Strong | Strong |
| Shrink exposure | Very low | Low (credential first, locked door) | Highest (open shelves) |
| Floor space | Smallest | Small | A room or alcove |
| Service effort per visit | Low | Medium | Highest (more SKUs, merchandising, cleaning) |
| Best fit | Small and mid-size sites, snacks and drinks | 24/7 sites, fresh meals, unsupervised hours | Large populations, cafeteria replacement |
Variable 1: transactions, not headcount
You’ll hear rules of thumb that tie each format to a minimum employee count. They’re proxies for what actually matters, daily transactions, and they break in two common situations: hybrid offices, where the badge count is double the Tuesday population, and 24/7 sites, where a small staff buys three times a shift. (We take the headcount rule apart in our site-scoring guide.)
Estimate transactions directly:
Daily transactions ≈ people on site on a typical day × share who buy on a given day × purchases per buyer
Then compare the revenue with what the format costs to own and service. Here’s an illustrative example for a single smart cooler. The equipment figure comes from our price list; the margin and visit cost are assumptions you should replace with your own.
- Equipment: a Kava Cooler C2 on Vending-as-a-Service is $189 a month on a 36-month term, including the hardware, Kava OS Pro and service.
- Restocking sits outside that. Assume a visit costs $30 in labor and vehicle time, twice a week: about $260 a month.
- Monthly cost to cover: $189 + $260 = $449.
- Margin: assume 50% gross margin after product cost. Covering $449 takes about $900 a month in sales, or roughly $43 a day across 21 working days.
- Break-even: at a $4.25 average ticket, that’s about 10 transactions a day.
Notice that the visits cost more than the cooler. That’s normal, and it’s why the service model matters as much as the box. A site that looks marginal on two visits a week can be healthy on one, if forecasting and pre-kitting make that one visit count. Our pre-kitting and dynamic routing guide shows how.
Variable 2: hours and supervision
The second question is when people are there, and who else is there at the same time.
- Daytime, supervised sites (a single-shift office with a front desk) tolerate open shelves. Social pressure works when colleagues share the room.
- 24/7 sites with thin overnight staffing (hospitals, distribution centers, residence halls) need locked formats for anything valuable.
- Public or transient sites (lobbies, transit, gyms with guest passes) should be locked, full stop. Anonymous traffic is the strongest predictor of open-shelf loss we know of.
Hours also shape fresh food. A cooler in a Monday-to-Friday office has a two-day hole every week, so a short-life sandwich delivered on Thursday may have one real selling day left. Either deliver fresh on a Monday-and-Wednesday rhythm, or carry a longer-life fresh range and let markdowns handle the tail. A 24/7 hospital floor sells fresh food around the clock, which is exactly why coolers shine there.
Variable 3: what people actually want
Assortment sets the floor for the format.
- Snacks, candy, canned and bottled drinks: a spiral machine does this cheaply and well. If one is already on site and mechanically sound, connecting it is usually the best first move. A Kava Link retrofit adds telemetry, remote pricing and cashless reader support without replacing the machine.
- Fresh meals, salads, yogurt, sushi: a smart cooler. These items are fragile, irregular and expensive, which makes them bad candidates for a coil drop and good candidates for a locked door.
- A full pantry: fruit, bakery, coffee, frozen meals and hundreds of SKUs. That’s a micro market, ideally with a kiosk that can recognize a banana without a barcode.
A quick test: list the 30 items the site would buy most. If more than a third are fresh or unpackaged, a spiral machine alone will disappoint.
Variable 4: how much you can trust the room
Shrink risk isn’t about whether people are honest. It’s about how anonymous they are and how often the room is empty. Score each site on four questions:
- Is traffic mostly the same people every day, or strangers?
- How many hours a day is the space unsupervised?
- Are there high-value items (energy drinks, premium meals) that are easy to pocket?
- Does the location host care, and will they act on a monthly report?
Sites that score badly can still have fresh food. It just needs to be behind a door. Across our deployments, shrink fell 62% when sites moved from open honor-system micro markets to Kava Coolers with locking doors. That’s not a judgment on the people at those sites. It’s what happens when the default changes from “pay if you remember” to “pay before the door opens.” Theft is also only one source of loss; where unattended retail shrink comes from breaks down the others.
The decision table
| Site profile | Lead format | Add-on worth considering |
|---|---|---|
| Single-shift office, around 10 transactions a day or fewer, snacks and drinks | Connected spiral machine (retrofit if one exists) | Nothing until demand proves out |
| Office with steady fresh demand, mostly regulars | Smart cooler | Spiral machine for snacks and candy |
| 24/7 site with overnight staff (hospital, plant, DC) | Smart coolers near each floor or shift area | Heated lockers for cafeteria-made hot meals |
| Large, mostly supervised population replacing a cafeteria | Micro market with self-checkout kiosk | A locked cooler for premium and high-shrink items |
| Public lobby, transit, gym with guest traffic | Locked formats only (cooler or spiral) | Outdoor-rated units where there’s no indoor space |
| Existing machines with sound refrigeration and mechanics | Retrofit first | Replace only on condition or demand evidence |
Hybrids beat purity
The strongest sites rarely run one format. A common winner at large workplaces is a kiosk with open shelving for chips, fruit and coffee, plus a locked cooler for $9 salads and energy drinks: the honor system for low-value impulse items, a door for the items that walk. At a three-shift plant, a cooler near each breakroom plus a spiral machine for snacks often beats one central micro market that night shift has to cross the building to reach.
Hybrids only work if the formats share a platform. If the cooler, the kiosk and the spiral machine report to three different systems, nobody can see the site as one store, forecast it or price it consistently.
A worked example
Take a regional distribution center: about 220 people on site on a typical weekday across three shifts, fewer on weekends, two breakrooms at opposite ends of the building, no cafeteria, and a pair of 12-year-old spiral machines.
- Transactions. If 35% buy on a given day and buyers average 1.3 purchases, that’s about 100 transactions a day (illustrative). Plenty for more than one device.
- Hours. 24/7 with a thin overnight crew. Open shelves are a risk after 10 p.m.
- Assortment. Second and third shift want real meals, not just chips. Fresh demand is high.
- Trust. Mostly regulars, but temporary labor rotates through in peak season.
- Service. One route stop covers both breakrooms.
Result: retrofit the two spiral machines for snacks and drinks, add one smart cooler per breakroom for fresh meals, and skip the open micro market. Revisit a kiosk if a staffed first-shift break area appears later.
Mistakes we see often
- Choosing the format before estimating transactions
- Putting open shelves in a space that’s empty at night
- Stocking fresh food at a Monday-to-Friday site without a weekend plan
- Replacing working machines when a retrofit would do
- Leaving the visit cost out of the model, when it’s often the largest line
How 8Kava approaches this
We start with the site, not the catalog. Site Intelligence models traffic, shifts, dwell time and existing equipment to recommend a format and an opening assortment, and it’s perfectly happy to recommend a $349 retrofit over a new cooler. Every format, from a smart cooler to a micro-market kiosk, runs on the same Kava OS, so a hybrid site behaves like one store. And if the numbers don’t work for a site, we’d rather say so before anything gets installed.