Here’s the idea in one sentence. Every order is one record with two sides: the buyer sees it as an order they placed, the vendor sees it as an order they received, and when either side moves it, the other side sees the move. That’s it. No portal to register on, no second copy to keep in sync, no ‘their system says confirmed but they told me Thursday on the phone’.
It sounds obvious written down. It isn’t how any of the software works.
Two portals pointing at each other
Take a café and the roaster who supplies its beans. The café runs a purchase order system, or a spreadsheet called Orders 2026 v3, and in it there’s a line: 30kg house blend, ordered Monday, due Thursday. The roaster runs a sales system, and in it there’s a different line with a different reference number for the same 30kg, sitting in a queue behind eleven other wholesale customers.
Two records. Neither one knows the other exists. The connecting tissue between them is a WhatsApp thread, a phone call on Wednesday morning and a delivery note that gets signed at the back door by whoever is on shift.
Now put a portal in. If the café is big enough to build one, the roaster gets a login and is asked to keep the café’s record updated, which means the roaster now maintains their own system plus somebody else’s. If the roaster builds an ordering portal instead, the café is asked to place orders in it, on top of the nine other suppliers they order from in nine other ways. Either way one party hosts and the other is a guest, and the guest is doing the work twice.
Two portals pointing at each other is just email with extra steps and a password policy. The WhatsApp thread keeps doing the real work because it’s the only thing both sides are actually in.
One record, two views
A shared order book inverts the hosting. There’s one record for the order, and each side looks at it from their own angle.
Walk the 30kg through it.
Monday, 4pm. The café creates the order: 30kg house blend, 6 x 5kg bags, £11.20 a kilo, delivery Thursday, back door before 10am because that’s when the manager is in. On the café’s board it appears in a column called Placed.
The same second, on the roaster’s board, it appears in a column called Received. Same quantity, same price, same Thursday. The roaster didn’t retype anything and nobody sent a PDF. It’s not a copy that arrived, it’s the same order seen from the other chair, which is why the two boards can’t drift apart.
Tuesday, 8am. The roaster taps Acknowledge. The café’s board moves the order from Placed to Confirmed and stamps it with the time, so when the owner asks on Wednesday whether the beans are coming, the answer is a glance rather than a phone call. If the roaster needed to change something, short by 5kg because the Guatemalan hasn’t landed, they change it on the same record and the café sees the amended quantity rather than discovering it at the back door.
Thursday, 6:40am. The van loads and the roaster marks it dispatched. The café’s board updates while the café is still closed.
Thursday, 9:15am. Six bags arrive. The manager counts them, marks five received against the order because one bag is missing, and the discrepancy is now on the record both sides are looking at rather than in a text message that ends with a question mark. The roaster sees it before lunch.
Friday. The invoice attaches to the same order, for the five bags that actually turned up. Two weeks later the payment marks it closed, and the order is done: one record, seven timestamps, both names on it, nothing reconciled by hand at any point.
The thing worth noticing is what never happened. Nobody logged into anybody else’s system. Nobody typed the same 30kg twice. There was no moment where one side’s screen said one thing and the other side’s screen said another.
What happens when the other side isn’t on the app
This is where portals fall over, so it’s worth being specific about.
Say the roaster isn’t on the app. Hasn’t heard of it, isn’t about to sign up for software because a café asked them to, and honestly that’s a reasonable position. Most of your suppliers are in exactly this state and they will stay there for a long time.
So the order goes to them as a WhatsApp message. It’s a clean message, the way a good order message should be: what, how much, at what price, by when, and where. At the bottom there’s a link. Tapping it opens a page with that one order on it and a button that says acknowledge, and tapping the button is the whole interaction. No account, no password, no app install, no commodity code. About four seconds, done from the phone that was already in their hand.
On the café’s side, the board moves to Confirmed by itself. Not because the café’s buyer retyped a WhatsApp reply at 5pm, but because the acknowledgement landed on the record directly. The café gets a structured order history without the roaster ever having to maintain one.
And if the roaster does sign up eventually, maybe a year later, maybe because a second customer started sending orders the same way, the history is already sitting there waiting. Every order the café placed, every acknowledgement, every short delivery. They’re not starting from an empty screen, which is usually the moment new software gets abandoned.
The honest limit: a tap is a lighter commitment than a full confirmation with amended quantities, and some suppliers will still just reply “ok” in the chat and ignore the link. You’ll be chasing those the old way. The difference is that it’s some of them rather than all of them.
Why both sides matters more than it sounds
Nearly every small business is on both sides of this. The café buys beans, milk, cups and a window cleaner, and it also sells catering: forty rounds of sandwiches and two urns to an office on Bishopsgate every other Tuesday. The roaster buys green coffee, packaging and courier services, and sells to ninety cafés.
A vendor portal covers one of those directions. Whatever you spend on it, half your order flow is still living somewhere else, usually in a different tool with a different vocabulary and a different person maintaining it. When the office on Bishopsgate rings to change the Tuesday order, that conversation happens in a completely separate place from everything else the café does.
On a shared order book those are two columns of the same shape. Orders you placed, orders you received. Same statuses, same acknowledgement action, same board, same Tuesday. The café’s catering customer gets the exact experience the café gives its roaster, because it’s the same mechanism turned around. For the businesses whose main problem is inbound orders arriving as voice notes, the wholesaler’s version of this goes into more detail.
It also changes who you can ask to adopt something. You’re not asking a supplier to join your system as a guest. You’re both on a record that belongs to the trade between you, which is a much easier sentence to say to someone you also want a favour from in March.
What it isn’t
A shared order book is not a marketplace. Nobody is matching you with new suppliers, there’s no discovery, no ratings, no bidding. You bring the relationships you already have.
It’s not EDI. There’s no 850 or 855, no AS2 connection, no mapping project, and if your customer is a supermarket demanding EDI this won’t satisfy them. It’s for the far larger number of trading relationships where EDI was never realistic on either side.
It’s not a compliance suite. No insurance certificate expiry tracking, no supplier risk scoring, no tax document vault, no approval chains routed through three managers. If you genuinely need those, you need a procurement platform and you should buy one.
It’s also not a replacement for talking to people. Prices get negotiated on the phone and problems get sorted out in person. The record is what the conversation leaves behind, not a substitute for having it.
The rule we run on
WhatsApp when they’re not on the app, structured workflow when they are. One board for every order you place and every order you receive, with the other party as an equal on the record rather than a guest in your system. That’s the model, and OrderBookApp is built on exactly it, which is the only time we’ll say so on this page.
The part still being worked out, and it’s a real open question rather than false modesty, is what happens as more of both sides join. A café with nine suppliers on one board and a roaster with ninety customers on one board are different problems, and the second one gets harder faster. We think the shape holds. Ask us again in a year.
Keep reading
- Why vendor portals fail (and what your suppliers do instead) explains the problem this model was built to answer.
- Customer order portal for wholesalers is the same idea from the seller’s side, for anyone whose orders arrive as voice notes.
- Vendor portal for small business: what you actually need compares this against building your own portal on a no-code tool.