How to evaluate vending machine software subscriptions with a twelve check trial

What to take away

Why products are named but not ranked

A ranked list of vending platforms would attach commentary to real, identifiable businesses, and nobody writing this page has run their software on a route. Capabilities and pricing change, integrations depend on the machines and readers you already own, and support quality varies by region and by month.

Products and providers you will encounter include VendSoft, Cantaloupe Systems (Seed and Seed Pro), Nayax (VendSys and Moma), Parlevel Systems, Televend, Vendon, and 365 Retail Markets. This page still does not rank them, because the checks below decide the fit.

So the deliverable here is the protocol. It takes a couple of weeks and it produces an answer about your route rather than about an average one.

The twelve checks

Run every one of these against each candidate. The order matters, because a failure on the first three usually ends the evaluation.

Turning the checks into a trial

Do not evaluate from a demonstration. A demonstration is designed to work.

Put the system on a small number of machines that are representative rather than convenient: one busy, one quiet, one with poor signal. Run it for at least a full service cycle and preferably two.

During the trial, do three things you will not want to do. Physically count a machine and compare it to the report. Deliberately let a machine go offline and watch what the data does. And call support with a real question, at the hour you would actually be calling.

What the trial usually reveals

Three findings recur, and none of them appears in a comparison table.

Reported sales and physical counts often disagree slightly at first, usually because of how the system treats a failed vend or a test vend. Find out how before you rely on the number, not after you have set par levels from it.

Signal quality is a site property, not a product property. The same system that reports perfectly at one building may not at a plant with thick walls, and that is worth knowing before you commit the whole route.

Support response is the variable that changes the experience most and is never in the specification. A system with fewer features and someone who answers beats the reverse on a route where a dead reader means lost sales.

Cost, compared properly

Build the comparison per machine per period, and include everything.

Ask for a written quote that states the recurring cost per machine per month, the connectivity cost per device per month, and the payment fee per transaction. Public list prices are uncommon, so that quote is the figure to compare.

The subscription is only part of it. Add any connectivity or data line, any per transaction charge, any hardware you have to buy to make it work, and any minimum that bites when you have few machines. Then compare that total to what the system saves: trips avoided, and the lost contribution avoided by finding faults sooner.

That saving is calculable from numbers you already hold, and the method sits with the wider software and measurement framework. On a clustered route with busy machines it usually comes out positive. On a small route with everything nearby, waiting is a legitimate answer.

The access question people skip

Whatever you choose holds your sales history, your site list and, indirectly, your payment arrangements.

The National Institute of Standards and Technology: NIST Small Business Quick-Start Guides cover account access and recovery for businesses with no IT function. The Cybersecurity and Infrastructure Security Agency: Cyber Guidance for Small Businesses pages cover what to do when an account is compromised.

Ask your payment processor separately what merchant obligations sit with you, since that answer depends on your arrangement rather than on the software.

Software does not discharge the record keeping

A platform makes records easier and does not satisfy the obligation. Sales, purchases, expenses and inventory all have to be documented, and the Internal Revenue Service: What kind of records should I keep? guidance sets out what a small business is expected to hold.

Check whether the system holds two records that often live outside it: temperature logs for refrigerated machines, and removals with a reason recorded. If it cannot, keep them with your other operating records instead of assuming someone will remember.

Also check what happens to all of it if you add machines in a new area with different connectivity or a different processor.

Common questions

Should I choose the system my machine manufacturer recommends?

It is a reasonable starting candidate because compatibility is likely, and it is not automatically the right answer. Run the same twelve checks on it that you would run on anything else, particularly the export and exit questions.

How long should a trial run?

At least one full service cycle at every trial machine, and two is better, because the first cycle is where you find out how the system counts and the second is where you find out whether it stays accurate.

Is it worth switching systems later?

It is, and check two and check three decide the cost of switching. A system you can export from is one you can leave.

That is why those questions sit near the top of the list, not the bottom. They matter more if you are still setting up the route and will choose again as it grows.