Grill Bill is our oldest shop: running since 2018, premium barbecues and accessories, with a brutal seasonal peak every spring. It was also our messiest: twelve separate tools, tied together with Zapier, exports and good intentions.
Why now?
The last straw was the 2023 seasonal peak. A campaign kept running for five days on a sold-out Kamado. €1,700 of ad budget spent on a product we could not deliver. Not because anyone was asleep, but because the ads dashboard simply did not know our stock.
That is not a tool problem, that is an architecture problem. You can patch it with yet another Zapier connection, or you can tear down the silos.
The approach: channel by channel, never big-bang
We migrated in six weeks, in this order:
- Weeks 1-2: Inbox. Email, chat and WhatsApp first, because that is the lowest risk and the fastest win. Trengo kept running in parallel for two weeks.
- Week 3: Customer data. Order history and customer profiles merged from six sources. Deduplication was where most of the manual work sat.
- Week 4: Stock. The nerviest step: three sales channels on a single stock truth. One afternoon a counter was off by 3 units; more on that below.
- Week 5: Email flows. Klaviyo flows rebuilt, and improved where the data was now richer.
- Week 6: Advertising. Last, and deliberately: it only becomes valuable once stock and customer data are in place.
Lesson 1: migrate the channel with the fewest dependencies first, and let the old system run alongside for two weeks. The overlap costs a few extra euros; the safety net is worth it.
What went wrong
An honest account, so this too. In week 4 the stock counter for one SKU was off by three units against the physical count. The cause: a bol.com return that the old system still had as "in transit" while it was already physically in the warehouse. The import took the system status as truth, not reality.
The fix was simple (take the physical recount as the source at cutover), but the lesson is bigger: a migration is the moment to confront your data with reality. Schedule a stock count.
The numbers after twelve months
| Metric | Before | After |
|---|---|---|
| Tool costs per month | € 1,480 | € 129 + € 41 modules |
| ROAS (on margin) | 2.9× | 3.8× |
| Manual work in peak week | ± 14 hours | ± 5 hours |
| Ads on sold-out products | 3-5× per season | 0 (paused automatically) |
"The May barbecue peak was always two weeks of stress. This year purchasing was in place three weeks ahead, and the ads paused themselves. I simply took that weekend off."
Daan Daalder, founder Grill Bill & Neuramerce
What this means for you
Grill Bill was our guinea pig; your migration does not have to repeat the learning phase. The order above is now our standard migration playbook: channel by channel, run in parallel, physical count at the stock cutover. For waiting-list customers we do the whole migration free of charge.