tablet, ipad, touchscreen, gadget, wooden, table, coffee, cafe, business, meeting, brown business, brown meeting, brown cafe, brown company, brown conference, brown tablet, tablet, tablet, tablet, ipad, ipad, ipad, ipad, ipad, meeting, meeting, meeting, meeting. How to test coffee shop software before you buy
Photo by StockSnap on Pixabay

Guides

How to test coffee shop software before you buy

A buyer's guide to coffee shop software: the exit test, the capabilities to see demonstrated live, a demo script, and a two-week trial that measures the peak.

What to take away

  • Run the exit test on any coffee shop software firstexport format, exit cost, retention after cancellation. Vague answers end the conversation.
  • Twelve capabilities, each demonstrated live on a menu you supplied. Described features are not features.
  • Timestamped transaction export and recipe entry by weight decide whether the system is useful at all.
  • Keep the exported file from every demo. It outlives the sales conversation.
  • Two weeks with a real busy morning, then ask the one question vendors never raise: did the peak get slower.

The exit test, before anything else

Ask three questions and get written answers.

The exit test

  • How do I export everything, in what format, and does it include transaction level detail?
  • What does leaving cost, including any final export charge?
  • What happens to my data after cancellation, and how long is it kept?

A vague answer here is the most useful thing you will learn about the vendor. This costs one email and removes candidates before you spend a week on trials.

Twelve capabilities to see working

Ask for each one demonstrated live, on a test menu you supplied.

  1. A transaction export with a timestamp on every line.
  2. Sales mix by item and by hour, from the day's real menu.
  3. Recipe entry by weight, with yield and portion size.
  4. Discounts, comps and voids that record a reason and a name.
  5. Waste and staff drinks logged as stock movements rather than notes.
  6. A cash count and till reconciliation at every close.
  7. Menu and price changes scheduled by daypart.

Twelve capabilities to demo live

  • Transaction export with timestamps to the minute
  • Sales by item by block, arbitrary date range
  • Recipes by weight or volume, theoretical usage
  • Physical counts compared against theoretical usage
  • Waste log with a reason field
  • Modifiers that do not break item reporting
  • Separate reporting for pre order or pickup

Numbers one and three decide whether the system earns its keep. Without timestamps you cannot see your peak. Without recipes you cannot compare theoretical usage against a count, which is the comparison that makes stock control work at all.

A demo script that stops the sales pitch

A demo script

Say thisWhat it reveals
Load these five menu items with recipes nowWhether recipe entry is realistic or theoretical
Show me last Tuesday in five minute bucketsWhether timestamped reporting exists
Export that to a file and open itWhether export is real and readable
Add a modifier and show item level sales afterWhether modifiers break reporting
Create a user who cannot see reportsWhether permissions are granular
Show me the audit trail for a refundWhether accountability is built in

Ask for the demonstration on your own laptop where possible, and keep the exported file. It is the only artifact from the process that survives the sales conversation.

Demo script that stops the pitch

  • Load five menu items with recipes now
  • Show last Tuesday in five minute buckets
  • Export that to a file and open it
  • Add a modifier, show item level sales
  • Create a user who cannot see reports
  • Show the audit trail for a refund

Security is part of the shortlist

A shop taking cards and holding customer contact details is a target, and small businesses get targeted because they are assumed to be unprotected. The National Institute of Standards and Technology publishes small business quick-start guides written for exactly this situation.

Build your questions from them: separate accounts per person, multi factor authentication, updates, and a plan for a lost device. Then decide who sees what. A part time employee does not need reporting, and access levels belong in induction alongside the standard a new starter is taught.

What the system has to hold besides sales

Records first. The Internal Revenue Service page on what kind of records to keep sets out the general expectation that a business can support its figures with documents, and the system is where most of those documents now live.

Product information second. The FDA's questions and answers on food allergens covers the federal framework for labeled packaged food. What you must disclose about made to order drinks is set locally by your health department. Whoever is on the counter needs to look a specification up during service rather than guess.

The two-week trial, and the peak test

Give it a fortnight with your real menu and at least one real busy morning. Then check three things: did anyone on bar avoid using it, did the exports match the reports, and did the peak get slower.

The two-week peak test

  1. Run a fortnight on your real menu
  2. Include at least one real busy morning
  3. Check whether anyone on bar avoided it
  4. Check whether exports matched the reports
  5. Time the same morning before and after
  6. Count tickets through the door

The third is the one people forget. A system that adds two seconds to every ticket costs cups in the block where cups are limited. Time the same morning before and after, and count tickets through the door rather than trusting the feeling of the shift.

Run the test again before you commit to a second site, because expansion decisions turn on systems that work across more than one bar. The same demo script and the same exit test apply to each candidate.

Common questions

Three questions that come up on almost every shortlist.

Should I choose the system my payment provider offers?

Test it like any other. Bundling is convenient and it can make leaving harder, so run the exit test on the bundle rather than on the software alone.

Will a tablet based system stand up to a bar?

Ask about failure rather than durability: what happens when the network drops mid service, and what happens when the device does. Those two answers beat any specification sheet.

How long should a trial run?

Long enough to include a busy morning and a stock count. Two weeks is usually the minimum that holds both.

More in Guides

Latest from Reporting Desk