Beverage distributor aligned preseason buys to velocity phases

A regional beverage distributor with roughly 1,900 SKUs, one central depot and two cross-docks lived and died by seasonality: summer terraces, December, and a calendar of sports events. Every spring it over-bought lines that had looked strong the previous year but were already fading. With four years of Odoo sell-through read into Flowra and a 90-day what-if run before the preseason order, seasonal inventory planning stopped being a slide deck argument. Preseason write-downs fell 35% year over year.

The situation

Three teams brought three forecasts to the preseason meeting. Marketing wanted depth on the brands it had promotional money behind. Finance wanted turns and a hard cap on the spring purchase order. Buyers referenced last year's sell-through in slides, not in ranked system output, and last year's number was usually the total, not the shape.

The warehouse filled in April with the wrong mix. Some lines sold out by the first hot weekend; others were still on the racks in October and went out at a write-down before the December reset.

What the data showed, and didn't say

Four years of weekly sales in Odoo held far more than year-ago totals. They held phase shifts: categories entering a seasonal upswing, brands whose 4-week velocity was already below their 12-week average by March, and promo-linked SKUs whose demand spike arrived two weeks before the campaign in the ERP calendar because the trade ordered early.

Odoo did not label any of that. It reported quantities sold per period. Which SKUs were entering an upswing and which were carrying dead weight from the previous cycle was a question the reports could not answer, so the loudest forecast won.

Connecting Flowra

The team connected Odoo 16 read-only through the XML-RPC connector in an afternoon. Flowra ingested sales orders, deliveries, purchase receipts and the product catalogue, reconstructed the stock trajectory per SKU from the net flow, and delivered the first seasonal risk view the next morning. Webhook-triggered syncs keep it fresh; full re-scoring runs nightly or on demand.

Six weeks before the spring order, purchasing ran the Flowra simulator on a 90-day horizon with two scenarios: the marketing-led mix and a velocity-led mix. Both produced a projected stockout list, an overstock list and the cash tied up at the end of the horizon.

What Flowra surfaced

• A preseason reorder list weighted toward SKUs in an upward velocity phase, with recommended quantities based on cover targets and measured supplier lead times.
• A clear-down list for 74 lines still carrying more than 16 weeks of forward cover from the previous season, with the annual carrying cost attached.
• Week-by-week stockout risk for the 38 promo-linked SKUs, with the lead-time cut-off date by which each order had to be placed.
• Dead-stock onset flags on eleven products that were still selling and still being restocked, but were likely to go silent for two months once the season turned.

What the team did

Leadership capped preseason PO spend to the velocity-led reorder list. Marketing exceptions were allowed, but each one had to be signed with the evidence from the simulator run attached: which SKU, how much extra cover, and the projected end-of-season position. Four exceptions were signed; nine requests were withdrawn once the projected overstock was visible in cases and euros.

Purchasing placed every order in Odoo themselves. The clear-down list went to the sales team as bundle and on-trade deals in April, before the previous season's stock competed with the new arrivals for rack space. During peak weeks, the 14-day stockout list for promo SKUs arrived in Slack each morning, and two early reorders were approved from the phone.

Results

"For years the preseason meeting was three forecasts and a shouting match. This year it was one list, two scenarios, and the marketing exceptions were signed with the numbers next to them."

— Managing Director, regional beverage distributor

What made it work

• Phase, not total. A SKU's year-ago volume says nothing about whether it is rising or fading now. The 4-week versus 12-week comparison in reading sales velocity trends in ERP exports is what separated the upswing lines from the dead weight.
• What-if before the PO, not after. Running both mixes over 90 days turned the marketing debate into a comparison of projected end positions. Exceptions became rare because they became visible.
• Clear the old season first. Treating prior-season stock as a slow-mover problem in April rather than a dead-stock problem in October is where most of the write-down reduction came from.