Who Owns Your Billing Logic?

Here is a question most companies never ask: who owns the rules that determine how you charge your customers?

Not the invoices. The rules. The tier boundaries, the proration behaviour, the minimum commitments, the discount logic, the decision about what happens when someone upgrades mid-cycle.

For a lot of businesses the honest answer is a vendor, and the arrangement is usually priced as a percentage of the revenue flowing through it.

The shape of the deal

Modern subscription billing platforms are genuinely good products. They handle metered usage, tiers, proration, plan changes, dunning and tax integration, and they do it well enough that building any of it yourself would be hard to justify at the start.

The pricing model is the part worth examining. A percentage of processed revenue means the cost scales with your success rather than with your usage of the product. At a quarter of a million in revenue it is immaterial. At five million, the same percentage is a significant operating line, and you are paying it for a system whose cost to serve you barely changed.

The exercise I would recommend to anyone currently on one of these platforms: work out what you will pay at three times your present revenue. Not as a criticism of the vendor — the model is transparent and the product is good. Just so the number is known rather than discovered.

The second thing you are renting

Cost is the visible part. The less visible part is expressiveness.

Your pricing model has to fit the platform's data model. Mostly it does, and that is the whole value proposition. But when you want to do something the platform does not model — a contract with a rate that steps up in year two, a parent company paying for several subsidiaries with separate usage breakdowns, a hybrid of platform fee and usage with different proration on each — you have two options. Change your commercial terms to fit the software, or add a manual step.

Companies pick the manual step, because the commercial terms closed a deal. And then the manual step scales with the number of unusual contracts, and somebody is doing an hour of arithmetic every Friday.

When owning it is the right call

Not always. Most businesses should configure a good platform properly and never think about this again. Billing is not a competitive advantage for a plumbing company.

Owning it makes sense in three specific situations.

When your contracts cannot be expressed in any candidate platform's data model without manual steps. Not most of your contracts — any of them, because the manual step is the problem.

When the manual work has already become substantial. Forty person-hours a month is roughly a quarter of a role, permanently, and you are paying for a custom system in salary without receiving the asset.

When how you charge is part of why you win. If your pricing model is genuinely differentiated — usage-based in a market that charges per seat, outcome-based in a market that charges per hour — then the system expressing it is strategic rather than operational.

What changed the arithmetic

For a long time none of this mattered, because building was unaffordable. A credible billing system meant a year of engineering on rating, tax, payment adapters, customer portals and admin tooling, none of which is your product.

AI has not changed the core of that. Rating logic and money arithmetic still need a human who understands the contracts, specifying carefully and reviewing line by line, because a rating bug does not crash — it silently charges the wrong amount, and the customer finds it before your tests do.

What changed is the surrounding seventy percent. Adapters, migration scripts, test fixtures, reconciliation harnesses, admin screens, document templates. Schema-heavy, pattern-heavy work that is now quick to produce and straightforward to review.

The result is that the break-even moved substantially. A build that was a year-long commitment is now often a phased project measured in months, with revenue flowing through the new path well before the end.

The one question

Can every one of your live contracts be expressed in your current platform's data model with no manual step?

If yes, you are in the right place and should stop reading about this. If no, then you are already running a hybrid system — software plus a person — and the only question left is whether you would rather own the whole thing.

Full version with cost ranges, architecture and migration approach: TechCirkle's guide to small business software for billing. Our SaaS development practice is where most of these conversations begin.

Frequently Asked Questions

Why does percentage-of-revenue billing pricing become a problem?

Because the cost scales with your success rather than with your use of the product. A percentage that is immaterial at small scale becomes a significant operating line at ten times the revenue, for a service whose cost to deliver has barely changed.

What does it mean to own your billing logic?

Holding the rules that define how you charge — tier boundaries, proration, minimum commitments, discount behaviour — inside a system you control, rather than configured into a vendor's data model. You can still use their payment processing; the distinction is about where the rules live.

Should most companies build their own billing system?

No. Most should configure a good platform properly. Building makes sense when your contracts cannot be expressed in any platform without manual steps, when manual billing work already exceeds roughly forty hours a month, or when your pricing model is part of your competitive advantage.

How do I know if my contracts fit a billing platform?

Take your ten most unusual live contracts and write out how each would be represented. Any contract that needs a manual calculation, a free-text note or a spreadsheet is a gap, and the number of such contracts tends to grow rather than shrink.

Has AI made building billing systems practical?

It has made the plumbing practical — adapters, migrations, fixtures, reconciliation, admin tooling, which is most of the effort. The rating logic still needs careful human specification and review. The net result is a phased project of months rather than a year-long commitment.