The situation
Ardem Ahşap makes bespoke furniture. One showroom, an average sale of 18,400₺, and a decision cycle of 3-8 weeks.
They were spending 62,000₺ a month on Meta ads. Based on website purchase data, ROAS looked like 0.9 — in other words, the advertising was losing money.
When they came to us, the item on the agenda was stopping the ads entirely.
The problem
Measurement only saw purchases on the site. But nobody buys 18,000₺ of bespoke furniture in one click. The customer messages on WhatsApp, comes to the showroom, has measurements taken, and then the sale closes.
None of the steps in that journey reached the ad dashboard.
WhatsApp traffic was invisible. There was a WhatsApp button on the site but the clicks were recorded nowhere.
Showroom visitors weren’t asked. “How did you hear about us?” wasn’t being asked, and when it was, it wasn’t written down.
A short attribution window was in use. A 7-day window is far too short for a product with a 3-8 week decision cycle. A customer who saw the ad and bought five weeks later was credited nowhere.
What we did
Week 1 — We started measuring the intermediate steps. WhatsApp clicks, form submissions and “get directions” clicks were each turned into a conversion event. Each was also sent server-side (Conversions API).
Week 1 — We set up a showroom log. Every visitor was asked one question and it was written into a simple sheet: how did you hear about us. Not a technology job — a discipline job.
Week 2 — We lengthened the attribution window. Evaluation moved from a 7-day to a 28-day click window.
Weeks 2-3 — We changed the campaign objective. The objective moved from “purchase” to “qualified enquiry”. Meta was now looking for people who would make contact, not people who would check out on the site.
Weeks 3-8 — We moved the creative from product-led to process-led. Instead of photographs of finished furniture: production in the workshop, the measuring visit, before and after installation, and footage from homes already delivered. In bespoke manufacturing the real objection is “will it come out the way I want?” — showing the process answers it.
Weeks 8-10 — We steered towards higher-value products. Once measurement was fixed, it became clear which product group actually made money. Budget moved to the kitchen and wardrobe categories.
The result
Over two months total enquiries from advertising reached 312, and 47 sales closed from them. In the previous two-month period that number was 29.
The real change was in measurement: the share of completed sales that could be measured rose from 38% to 91%. The advertising had been working all along; it simply wasn’t visible.
Because budget moved to higher-value products like kitchens and wardrobes, the average sale value also rose from 18,400₺ to 24,900₺.
What we learned
If the sale isn’t completed online, you can’t make decisions by looking at online data.
In this case, what was won before a single line of ad copy changed was simply seeing the right number. Any optimisation made before measurement is fixed is like giving directions from the wrong map.