The missed order that stopped happening

What changed

• The order gets placed on time.The product that would have run out in six weeks is on the list today, with the materials it needs. The forgotten order stopped turning into an empty shelf.
• The afternoon came back.A weekly spreadsheet afternoon became one refresh, one folder drop, one click. The team reads a finished sheet instead of building one.
• The argument ended.Every number shows its working, so the production lead questions one cell instead of distrusting the whole sheet. Decisions get made on the call, not deferred to a re-check.

The brand

Cocosolis is a skincare brand from Sofia with around seventy products in eight categories, sold through retailers across Europe, the UK and the Gulf. Production is in-house. Raw materials take two to four months to arrive and every production order needs a deposit up front.

The problem

"Someone in the chain forgets to order something on time, and we run out of stock six weeks later."

That is the owner's own description. The maths behind it is simple and unforgiving. If a product runs low today, the materials to make more arrive in two to four months. So the moment to order is long before the shelf looks empty, and nobody has time to check seventy products every week by hand.

When someone did check, it was a spreadsheet: join warehouse stock to sales, project ahead, find the risks. Slow, and one dropped row meant one missed order.

What runs now

Every week the system reads five things:

• stock in four warehouses, from the ERP
• the production orders that are already open
• the materials on hand, with what they cost
• the work orders of the year so far, product and material lines
• three years of monthly sales, with the current year refreshed as each month closes

It then does the analyst's job and writes four sheets:

1. Forecast. Three months ahead for every product, each with its own season, with a flag when a product is growing faster than its history and a separate flag for products too young to trust.
2. Stockout. Each product marked covered, at risk, or needs a look, month by month.
3. Materials buy. The shopping list, only what the forecast actually needs, with the materials that need a human decision marked.
4. Materials proof. The working behind every number, so the production lead can question one cell instead of distrusting the sheet.

Products get renamed, discontinued, bundled. The system carries those rules in plain config tabs the team edits themselves, so a legal rename does not split a product's history in two.

What the team does each week

One sales refresh. One folder drop. One click. The output lands in the same sheet they already use, one tab per week, so a year later there is a record of what was true when and what was done about it.

How it was built, and who decides

I built it with AI tools; that is the part that got cheap. The part that did not get cheap is the judgement: what to read, how to join it, which products to flag and which to leave alone, what the production lead needs to see first. That came from sitting in the analyst's chair at Cocosolis for a year and a half. It is the same judgement I bring to your desk.