Scaling a marketplace from one million to 2.2 million monthly active users

The situation

eBay Italy was a general marketplace: everything for everyone. One million monthly active users, a €15M P&L, and a forty-person team. Traffic was not the problem. Users arrived, browsed, sometimes bought, and did not come back with any reliability.

The monetisation question was the one the business could not answer. A marketplace makes money on transaction volume and margin, and both depend on frequency. A user who visits once a quarter is not a customer, whatever the acquisition dashboard says about them.

The diagnosis

Segmentation was demographic — age, region, income band. That approach describes who someone is and tells you nothing about what they came to do. Two users with identical demographics had completely different intent, frequency and value, and the marketing spoke to neither of them.

The second problem followed from the first. Because nobody knew what users came for, the platform could not build depth anywhere. Every category was shallow. A user looking for something specific found a thin selection and went to a specialist site that had made the effort.

What I did

1. Rebuilt segmentation around behaviour

Demographics were replaced with observed behaviour: what people browsed, how often they returned, what they abandoned, what they bought repeatedly. That produced entirely different groups from the demographic model — and, critically, groups that could be spoken to with a proposition rather than a message.

2. Built vertical propositions with real depth

Rather than marketing the marketplace, I marketed specific destinations within it. The sports vertical was the clearest case: enough inventory depth, enough brand presence and enough dedicated experience to become a place people came to deliberately rather than arrived at by search.

Each vertical was treated as a small business with its own supply target, its own demand target and its own economics.

3. Used partnerships to solve supply and credibility at once

More than 500 brand partnerships were built to give the verticals genuine depth. The commercial logic is efficient: a partner brings their own audience to the platform while simultaneously fixing the platform's inventory problem in that category. One negotiation, two constraints solved.

4. Grew both sides in sequence, not in parallel

Two-sided growth run simultaneously with equal effort tends to break liquidity — buyers arrive to thin inventory, or sellers arrive to no demand. Supply depth came first within each vertical, then demand was driven into it. That sequencing is why the GMV figure moved alongside the user figure rather than lagging it.

The results

What transfers to other businesses

• Behaviour beats demographics wherever intent varies. If your segmentation cannot predict what someone will do next, it is describing your customers rather than helping you serve them.
• Depth in one vertical beats presence in ten. A platform that is the obvious destination for one thing acquires more efficiently than one that is an option for everything.
• Partnerships solve two problems per negotiation. Supply and credibility usually constrain platforms at the same time and can be addressed in the same move.
• Sequence two-sided growth. Fix the constrained side first. Parallel growth with equal effort usually breaks liquidity on one side or the other.