I designed and built a B2B ordering tool for Delicorner with Airtable. I took the project from discovery through delivery.
One order, four manual entries
Delicorner is a Paris-based company founded in 2015. It delivers fruit and snacks to businesses.
The project covered orders for Meta, Salesforce and Datadog. Over 300 distinct products per week, around €100,000 per month. Each client had its own negotiated product list, budget and delivery rules.
Each order went through four manual entries. The client filled in a Google Sheet. Customer success copied the data into a second file, then entered it again in the internal CRM. During preparation, the warehouse recorded shortages in a fourth file, which customer success then used.
Then came the comparison. Customer success manually checked ordered quantities against delivered quantities, line by line, to prepare the report.
Every entry added a risk of invoicing errors. Between placing the order and receiving the report, the client had no visibility. And the entire process relied on one person.
The fieldwork settled it
I mapped the full order journey across customer success, purchasing and logistics. I also interviewed the client.
At the warehouse, I identified the main mismatch. Clients ordered individual units. The warehouse prepared packs.
For example, a client orders 55 cans of soda. The cans come in packs of 12. The picker takes 4 packs, or 48 cans, or 5 packs, or 60. Either way, the client does not receive what they ordered. The discrepancy goes back into the report and has to be explained manually.
I prototyped the interface with Figma Make. Three rules shaped the first version.
- Order full packs. Products supplied in packs of 12 must be ordered in multiples of 12.
- Never overwrite the ordered quantity. Invoicing is based on delivery; the difference must remain verifiable.
- One database for all 3 clients, with separate interfaces. I designed it to scale, so other clients using the same model could be added easily. I turned down specific requests to keep it adaptable.
From ordering to reporting
The client orders through a portal, with the total updating in real time. After delivery, they can see the difference between what they ordered and what was delivered.
They can also enable recurring orders. The same order repeats every week, so they no longer need to re-enter quantities.
Once an order is approved, the tool generates the logistics export and picking lines. It creates the next order and moves the portal to the following week.
Warehouse staff record shortages in the tool. Purchasing enters the restocking date. Customer success receives an alert if that date falls after the delivery date.
After delivery data is imported, the tool calculates discrepancies against the original order. The report is ready without manual work.
I supported the rollout with a demo, a client guide and a follow-up after 7 days. I documented procedures by role so the teams could work without me.
Five to eight hours given back
All 3 clients and customer success use the tool daily.
Previously, customer success spent 1 to 2 hours per client preparing reports. That work is gone. The weekly meeting used to take 1 hour per client; it now takes 15 to 20 minutes.
Across 3 clients, that gives customer success back 5 to 8 hours a week.
Clients can order independently, check their budget at any time, and no longer discover new information in the report. Fewer errors, more autonomy.