Picking a monetisation model is an engineering decision

Founders usually treat monetisation as a pricing question and hand engineering a number. It isn't. Each model implies different infrastructure, a different refund story, and a different worst case when a stream drops.

We built MyPerico with three concurrent revenue streams at launch: paid live-stream entry fees, self-serve post advertising, and platform commission on marketplace transactions. Here's what each one costs to build.

1. Paid entry to a live stream

The model: viewers pay to enter a stream, optionally with a capped number of slots.

What it costs to build: more than you think, and the cost is concentrated in edge cases.

The happy path is easy — take payment, grant access. The hard parts:

  • Capacity races. Two people pay for the last slot at the same moment. You need the seat reservation and the payment to be a single atomic operation, or you refund an angry customer.
  • The "slot full" state. This needs real UX attention, not an error toast. A user who paid and can't get in is a support ticket and a chargeback.
  • Failure mid-stream. If the host's connection dies ten minutes into an hour, what does a paying viewer get? Decide the refund policy before you build, because it determines whether you capture payment up front or hold an authorisation.

When revenue arrives: immediately, per stream. That's the appeal — it's the fastest of the three to produce real money.

2. Self-serve advertising on posts

The model: users pay to promote their own posts.

What it costs to build: the streaming part is free — this doesn't touch video at all. The cost is in the admin surface:

  • Ad creation, targeting and budget controls for the advertiser
  • A review queue, unless you're comfortable with unmoderated paid placement
  • Delivery accounting — impressions served against budget, without double-charging on retries
  • Reporting the advertiser trusts enough to spend again

When revenue arrives: slowly at first. Self-serve ads need supply and demand: enough audience to be worth buying, enough advertisers to fill inventory. It's the highest-ceiling model and the slowest to start.

3. Marketplace commission

The model: take a percentage of transactions between users.

What it costs to build: the payments work is the real scope — split payments, payouts to sellers, tax handling, and a dispute process. Commission itself is one line of arithmetic; everything around it is not.

When revenue arrives: proportional to marketplace activity, which is usually the last thing to warm up.

Why three, and not one

Three models at launch sounds like scope creep. The argument for it is that they fail on different timelines.

Paid entry produces revenue immediately but is capped by how many streams happen. Advertising has the highest ceiling and the slowest start. Marketplace commission scales with a behaviour you don't fully control. Ship only advertising and you have months of near-zero revenue while inventory builds.

The build cost is also less additive than it looks — all three need the same payments substrate. Once you have taking money, holding money, paying out and refunding, each additional model is mostly product surface, not new infrastructure.

What that implies for sequencing: build the payments layer as a shared capability, not inside the live-streaming feature. If your first revenue model owns the payment code, the second one starts with a refactor.

The thing that decides it

None of this matters if the streaming experience is poor. In MyPerico's beta, paid live streaming was the feature reviewers cited most often as the differentiator — across 180 reviews at a 5★ average. People pay for live video when the live video is good. The monetisation model determines how you capture that value; it never creates it.

The MyPerico case study covers the full build, including the ad manager and the streaming stack.