What a cafe counts that a restaurant doesn’t
Ask a chef where their food cost hides and they’ll say the proteins. Ask me and I’ll say milk, and then I’ll say the modifiers, and then I’ll admit I spent two years not really knowing. The tools sold to restaurants assume the plate on the ticket is the plate in the recipe book. In a cafe that assumption breaks before the second customer of the morning.
The drink on the ticket isn’t the drink in the recipe
A 12oz latte has a cost. A 12oz oat latte with vanilla and an extra shot has a different one, and in most shops both ring up as the same menu item with modifier lines hanging off it. Software that costs only the base item quietly prices every one of those drinks as if it were made with whole milk and two ounces of espresso. Charge for the oat or eat the cost, and the report reads the same either way.
We don’t charge extra for oat, or almond, or any of it. I’d make that call again. But it does mean every oat latte earns what a whole-milk latte earns and costs more to pour, and for years I could not have told you how much more. It was never a knowledge problem. Nothing in the stack was doing the arithmetic. Now the swap carries its own cost, and a week of POS sales adds it up, so the free upgrade has a number on it instead of a shrug.
Milk goes out the door in small amounts, all day
A restaurant’s waste arrives in events: a bad delivery, a walk-in door left open, a botched prep. A cafe’s waste is a hundred tiny withdrawals: steamed too much, wrong size called, wrong milk, remake, the pitcher someone dumped at close. None of it is dramatic and all of it is real, and it shows up as a gap between what your sales say you poured and what actually left the fridge.
Small basket, high volume, no back office
Our tickets are a fraction of a dinner check and there are a great many more of them, so a few cents of drift per drink is a real number by the end of the month. Meanwhile nobody in the building has an office. Counting happens standing in the walk-in between rushes, or at the bar at 3pm with one hand, on whatever phone is in an apron pocket. If a count takes a laptop and a login, it does not happen.
How ShelfCount handles each one
Modifiers get real ingredient effects
Every modifier on your POS can be told what it actually does to the recipe: add an ingredient (an extra shot), swap one for another (oat for whole), change the quantity of one, change the cup, or do nothing to cost at all (“extra hot”). A modifier can carry several effects, applied in order, and each one shows its net cost impact next to it (+$0.18, −$0.20), so you can check an upcharge against what it actually pours. Once configured, sales are costed as the drink that was really made, and past sales re-cost themselves.
Theoretical vs. actual, so the milk gap has a number
Once drinks are costed, your Square or Clover sales tell you what you should have used. Your counts tell you what you did use. The usage variance report puts those side by side, per item, which is how a hundred invisible half-pitchers turn into one line that says you are three gallons light this week. That is a number you can do something about, usually by changing which pitcher gets used for a 12oz.
Invoices get scanned, and you approve every price
Photograph the invoice at the register and ShelfCount reads it. What it does not do is push those prices into your costs on its own. Receipt ingestion is review-first by design, so every price change lands in a queue for you to approve or reject. When you approve, the recipes that use that ingredient recost themselves, all the way up through prep recipes, so your drink costs move the day your dairy invoice moves rather than the next time someone remembers to update a spreadsheet.
Anyone on the floor can count, from any phone
Counts open in a browser on the phone your baristas already carry. No app install, no per-seat pricing, no separate password to reset when someone new starts. You can count one storage area at a time (the bar, the back fridge, the dry shelf) and each area’s on-hand is attributed to it, so counting the bar between rushes is a real count rather than a half-finished one. Par levels turn a finished count into a purchase order your dairy vendor can read, in one click.
What we don’t do
Multi-unit enterprise reporting is not yet here. There is no group-level rollup across a dozen locations, because the product is built for independents and small groups first. Square and Clover are live today; Toast is in certification and not connectable yet. And the invoice scanner is deliberately never automatic, so it costs you a few seconds of approval per invoice. Those are the honest edges, and I would rather you read them here than find them in week three.
What it costs, and what everyone else costs
ShelfCount is $50 a month, flat, one plan, month to month, with a 14-day free trial you start yourself. Unlimited team members, unlimited storage locations, waste tracking included rather than parked in a higher tier.
For context, here is what the category publishes on its own pricing pages, checked August 2026:
| Published price | How you buy it |
|---|---|
| ShelfCount — $50/mo flat | Sign up yourself, 14-day trial |
| MarketMan — $249 / $299 / from $449 per month | Book a demo; no free trial |
| MarginEdge — $350 per location per month ($500 with Freepour) | Book a demo; no free trial |
| WISK — $249–$399 per month for one venue, plus a one-time onboarding fee of $750–$9,500+ | Book a demo; 60-day money-back guarantee instead of a trial |
| Craftable — no published price | Book a demo |
Every figure above comes from the vendor’s own public pricing page in August 2026; Craftable doesn’t publish one at all. Two things stand out for a single cafe. The first is that the entry price of this category is five to seven times ours before anyone adds a second location. The second is that you cannot try any of them without a sales call. WISK also publishes a one-time onboarding fee starting at $750 for a single venue, which is fifteen months of ShelfCount before the subscription even begins.
None of this makes those products bad. They are built for operators with more locations and more back office than I have, and priced for them. It does mean an independent cafe is buying a machine sized for somebody else. If you want the line-by-line version, we keep a detailed comparison with MarketMan, and the full plan lives on our pricing page.
A note on cafe benchmarks
You will find articles claiming the average coffee shop runs 25% to 35% cost of goods. That range is trade-press estimate rather than survey data. The one industry body that does break out the coffee-and-snack segment keeps the figure behind a paywall, and the blogs quoting percentages disagree with each other anyway. A band ten points wide is too loose to steer by. Your own number, measured against your own sales, is the only benchmark that will ever change a decision, and that is the number this software exists to give you.
Cafe costing guides are on the way. In the meantime the calculators are free and want no email address: food cost percentage, recipe and drink cost, and prime cost.
Cafe questions we get
- Does ShelfCount cost drinks with modifiers?
- Yes. Each modifier gets its own effect: add an ingredient, swap one for another, change a quantity, change the cup, or no cost impact at all. Those effects apply to sales as they sync, so an oat latte with an extra shot is costed as an oat latte with an extra shot rather than as a plain latte. Each effect also shows its net cost impact, which is how you check an upcharge against what it actually pours.
- Can baristas count without installing anything?
- They open a link on whatever phone is in their apron and start counting. No app, no seat licence, no separate password to reset. Team members are unlimited on the one plan, so putting the whole floor on it costs nothing extra.
- Do I need a POS to use it?
- No. Counts, recipes, vendors, and purchase orders all work on their own. Connecting Square or Clover adds sales, which is what turns recipe costs into theoretical vs. actual food cost — worth doing, but not required on day one.
- How is $50 a month possible when everyone else charges $250 and up?
- There is one plan instead of three, no sales team, and no demo call to sit through, so you aren't paying for a funnel. Waste tracking and unlimited users are in the base plan because there is no other plan to put them in. The tradeoff is real: multi-unit enterprise reporting is not yet built, since the product is aimed at independents rather than at groups rolling up twenty locations.
- What if my inventory is already in another system?
- Export your item list and upload it. The importer maps your columns and you approve the mapping before anything is written. If you would rather not do it yourself, done-for-you migration is a one-time $299.