A residential developer had two projects to sell, to local buyers and to buyers abroad, in two languages. Google was blocked on day one by a slow tag install on the website side, so Meta had to carry the launch on its own. We turned it into a controlled experiment and let the data choose where the money went.
The situation
Two projects, two audiences, two languages, one channel
The client needed a quick, honest read before committing larger budgets: which project, which audience, which platform and which language would produce leads its sales team could actually work. Every lead had to come from someone who could plausibly buy a home.
What we did
A launch built as an experiment
-
Split everything that could be split
Campaigns by audience and platform, ad sets by project and language, and eight instant forms, one per project, language and platform, each routed to the CRM. The first budget was split evenly so nothing was decided by guesswork.
-
Read the data early
After nine days the ready-to-move-in project and its local-language creative were clearly ahead. Facebook with local buyers was the cheapest; Instagram with buyers abroad the most expensive.
-
Isolated the weaker project
The second project got its own campaigns and budget for five days. Its leads cost about $36 against about $4.70 for the lead project, so it was paused and the budget moved.
-
Let sales feedback decide the audience
Local audiences were narrowed for quality. Broad automated audiences were tested abroad for volume, then switched off when the sales team reported the leads were weak.
Same channel, same weeks: cost per lead by project
Results
Nine weeks, 896 leads
Across the full period the account produced 896 leads at an average of about $7.30 each. Leads from abroad ran about three to four times the local cost, which shaped where the second phase went.
The work
Lifestyle first
What made the difference
The ad platform could tell us which leads were cheap. Only the sales team could tell us which were real. Building that feedback into the weekly decisions mattered more than any bid setting.
