In wholesale food distribution the working day has a hard edge at both ends. Orders keep arriving until the small hours. Vehicles have to be loaded and gone by six. Everything that determines whether the day goes well happens in the gap between those two points, and in most businesses that gap is filled by one person working from memory.

It is worth being precise about what that person is doing, because it is not one task. It is four, and they have different answers.

01 Orders arrive in four formats and none of them are structured

A regular customer texts a photo of a handwritten list. Another leaves a voicemail. A third emails a purchase order. A fourth uses the ordering portal properly, which is why they are everyone’s favourite customer. A fifth says “the usual, plus twenty kilos”, and the usual exists only in someone’s head.

This is the actual bottleneck, and it is the one most people skip past because it feels unfixable. It is not. Normalising orders from mixed inputs into one structured queue is now genuinely reliable work, including reading a photographed list or a voice message and turning it into line items against your product codes. It does need a person to confirm anything ambiguous. That is a thirty second review, not a thirty minute transcription.

02 The run sheet is a rules problem pretending to be an art

Sequencing the day looks like judgement, and some of it is. Most of it is constraints written down for the first time.

Vehicle capacity and cold chain. Delivery windows. Customer specific rules that exist nowhere in writing: the back gate before seven, the restaurant that will not take a delivery during service, the account that must go before its competitor, the driveway a rigid cannot enter. The person building the sheet holds forty of these. When they are away, the day runs long and nobody can explain exactly why.

Written down, most of those become rules a system can apply, with the sequence proposed and the person adjusting it. The value is not that the machine routes better. It is that the knowledge stops living in one head.

03 Catch weight breaks the paperwork

Meat, poultry and seafood are sold by weight, which means the invoice cannot be finalised until the product is picked and weighed. So the office produces a docket, waits for actual weights, then rebuilds the invoice.

In a lot of operations that loop runs twice a day, every day, and it is manual on both sides. It is also the point where an error costs real money, because a weight typed wrong is either margin given away or a credit request from a customer next week.

Weights captured once at the pick, flowing into the docket, the invoice and the dispatch pack without being read off a screen and typed into another one, removes a whole category of daily work.

04 The four separate places nobody wants to look

When a customer calls at nine to ask where their delivery is, someone checks the run sheet, then the driver’s phone, then the order system, then the invoice. Four places, one question, and the answer is usually “let me call you back”.

This is worth measuring in your own business before doing anything about it. Count those calls for a week. Most operators are surprised, and the number is what makes the case.

What this actually costs

The straightforward reading is labour: two hours of a senior person every morning, plus the interruptions. That is real, but it undersells it.

The bigger cost is that the six o’clock window sets the size of the business. New accounts get taken on cautiously, not because demand is missing or margin is thin, but because nobody is confident the morning can absorb them. Growth ends up limited by a two hour window rather than by the market, and that is a decision nobody consciously made.

Fixing the window does not save two hours. It removes the ceiling.

Where to start

Order capture, every time. If orders arrive in four formats, everything downstream stays manual no matter what you do to it. Get them into one structured queue first and the routing, the documents and the customer questions all become tractable.

Running a distribution operation with a morning like this? Send us a week of orders in whatever form they arrive and we will show you what a structured queue would have looked like.

Related reading: Where the hours go in a third party logistics operation and seven processes worth automating before you add more staff.

Frequently asked questions

Can delivery routing be automated for a small fleet?

Yes, and small fleets often benefit more than large ones because the constraints are customer specific rather than purely geographic. The practical approach is a proposed sequence that a dispatcher adjusts, not an unattended optimiser. The gain is that the rules are written down and survive someone taking leave.

What about orders that come in by phone or text?

Voice messages, photographed order sheets and free text messages can be converted into structured line items against your product codes reliably enough to be useful, with a short human confirmation on anything ambiguous. This is usually the highest value place to start because everything downstream depends on it.

Do I need a transport management system?

Not necessarily. A full TMS makes sense at scale and where routing is the core problem. For most distributors under one hundred million in turnover the binding constraint is order capture and document generation, not route optimisation, and those can be addressed without replacing the systems you already run.

How long does this kind of build take?

Order capture into a structured queue is typically two to four weeks. Run sheet generation and document flow add a similar amount, depending on how many systems have to be read from. Scoping the sequence properly matters more than the total, because the first piece has to work on its own.