B2B Distribution

From a photo of the menu to a sales package: how a distributor closes on the first visit

A rep photographs the restaurant's menu and minutes later has a quote, a kitchen map, a coverage report, and a sales video in the distributor's brand.

~3 min

from the menu photo to the three sales documents

System capability, not a client business result.

4

deliverables per restaurant, all carrying the distributor's brand

System capability, not a client business result.

2,111

catalog products normalized and priced to the day

System capability, not a client business result.

Selling supplies to a restaurant is guesswork. The rep walks the kitchen, estimates what that client buys, and comes back days later with a hand-built quote that is almost always incomplete. By then the chef has bought from someone else. Cross-referencing a restaurant’s menu against a catalog of two thousand products does not fit inside a visit — and it is exactly what decides the sale.

For a supply distributor we built the shortcut, and it is running today. The rep sends up to eight photos of the menu from their phone and the system does the rest: it reads the dishes, infers the inputs each one requires, cross-references them against the real catalog at current prices, and returns four documents in the distributor’s brand, in about three minutes.

The problem: the quote arrives late, or never

A twenty-dish menu hides a hundred inputs. Some are written down — the salmon, the avocado — and others never appear: the oil everything is fried in, the salt, the napkin, the takeout container. That second group is where the distributor’s margin lives, and it is exactly what a rep forgets when quoting from memory.

Building the full list by hand means someone cross-referencing the menu against thousands of SKUs, checking what exists, in what presentation, at what price. That does not happen during a visit. It happens days later, if at all. And the restaurateur, who already had a supplier, did not switch suppliers over a late email.

The four documents

The quote. One optimal SKU per ingredient: the operational list, without twenty-five variants of the same oil. This is the one that gets signed.

The kitchen map. The opposite: every available variant per ingredient, category by category. It is the “I can supply all of it” argument turned into paper, and it serves the conversation that follows the first sale.

The coverage report. How much of that restaurant the catalog covers, and what it is missing. This is the one nobody asks for.

The sales video. Sixty seconds with voice-over and the distributor’s identity, made for that restaurant. The full package, video included, takes about ten minutes.

How it was built

The hard part was not reading the menu: it was the catalog. A real commercial catalog arrives with broken characters, units buried inside product names, and duplicates nobody had noticed. Before cross-referencing anything, 2,111 products had to be normalized into a structure you can actually search.

The second decision was the interface. The system lives in an access-controlled messaging channel, not a new application: the rep sends photos from where they already work and receives the PDFs in the same place. Nobody installs anything or learns a dashboard.

The third was the live catalog. We started with manual exports and replaced them with an API connection to the store, because a stale price on a quote costs more than not quoting at all.

What changes for the team

The rep stops promising to “send some information” and delivers at the table. The tone of the conversation shifts: they are no longer asking for a chance, they are showing how much of the kitchen they can supply.

And leadership gets something that did not exist before: every visit produces market intelligence. The coverage report turns each quote into a list of the products the catalog needs in order to become a genuine single-source supplier. The sales operation pays for itself and, along the way, tells purchasing what to buy.

The figures on this page are system capability — generation times, catalog size, number of deliverables — not client business results. We do not publish our clients’ metrics, and we are wary of anyone who does without permission.

Does your team quote by hand?

If your reps take days to build a proposal that should come out of the visit, there is a design that fixes it. Book a 30-minute session and we will go through your catalog and your process — with explicit scope, like this one.

Frequently asked questions

How accurate is reading a photographed menu?

Good enough to work from, which is why the rep reviews before handing anything over. The system reads up to eight photos of one menu, infers the inputs each dish needs — including the ones nobody writes down, like oil, salt, or takeout containers — and flags what it could not find in the catalog instead of inventing it.

Where do the prices come from?

From the distributor's real catalog, synced by API against their store. The quote carries today's price, not whatever was in the last Excel export somebody remembers making.

What is the coverage report for?

It tells you what that restaurant needs and the distributor does not stock. Nobody asks for it, and it is the document that pays the most: every quote that goes out comes back as a shopping list for the purchasing team.

Is the video generated automatically too?

Yes. Sixty seconds with voice-over and the distributor's visual identity, made for that specific restaurant. The three PDFs take about three minutes; the full package with video, about ten.

Does this work outside restaurants?

The pattern works wherever there is a large catalog and a customer whose consumption can be inferred from something they already published. A menu is the cleanest case; a service list or a spec sheet works the same way. The initial diagnosis tells you whether yours qualifies.

Ready to scale your infrastructure?

Book a 30-minute strategy session to assess your current workspace.