CleanBee: launching a cleaning marketplace from scratch in Florida

CleanBee is a cleaning platform (app + web version) in Florida, USA. It’s a two-sided marketplace: customers find cleaners locally and buy their services, while cleaners get orders. The logic is like a freelance marketplace, only instead of freelancers there are cleaners. We joined the project about two years ago, when the brand had just launched.

Task

A two-sided marketplace that at the start has no audience on either side. So the main task wasn’t a KPI on cost per lead, but to test the platform: deliver the first traffic and the first economics, see how the audience interacts with the app (where the bugs are, how users behave), bring in real people — and understand how viable the idea is and where to take the product next. The initial economics were needed so the client could go to investors and present numbers to raise the main investment. We still set benchmarks, though: cost per install ≤ $5, cost per acquired customer ≤ $100.

Our approach

1. Cleaners first, then customers. The logic is simple: if a customer opens the app and finds no cleaners, they’ll delete it and go to competitors; whereas a cleaner without orders still stays in the app, because they can get an order at any time. So we first gathered cleaners across Florida — via Facebook groups and paid targeting.
2. Local targeting by cleaner density. Then we targeted customers precisely in the regions of the state where the most cleaners had already registered.
3. Deep audience research → ICP (ideal customer portrait). We targeted a narrow, affluent segment in specific wealthy areas of Florida — where people have the means to order cleaning regularly.
4. A mix of channels: app + website + social media — to work around the reach limitation (below).

Challenges

1. Zero trust and zero materials. A freshly launched brand: no reviews, no «before/after» cases, no people willing to speak on camera. Two years ago — long before the current hype — we started using AI image generation and sourced «before/after» materials to at least somehow demonstrate the service’s value and build trust through content.
2. iOS privacy limitations. Because of Apple’s data-privacy policy, out of more than 1M iPhone users in Florida, targeting could technically reach only about 5,000 people. So we built funnels through the website and social media as additional install sources — to reach the audience that in-app advertising couldn’t get to directly.

Result

In about 4 months we delivered the needed traffic volume and trained the campaigns. Installs came out cheap — $3–5 per install. As planned, we reached ~$150 per registered customer: at first glance not cheap, but factoring in LTV (the customer’s lifetime in the app) it’s more than worth it — and for launching a brand-new product from scratch, it’s a strong result.

Start from zero a marketplace with no audience on either side

Cleaners first gathered cleaners across Florida (Facebook groups + targeting)

Then customers precise targeting of affluent areas by ICP

$3–5 / install cheap installs, trained campaigns

~4 months → investors initial economics to attract investment

It was exactly these numbers and the initial economics that let the client keep refining the app and attract investment. The budgets were small (a few thousand dollars a month) — because the mission of this stage was precisely that: to validate the MVP and gather the economics for investors. With it, a brand-new product gained the arguments for its next round of growth.