Software pricing models are not really about how vendors charge. They are about which risk you take on as the buyer. A model that looks cheap on day one can be the most expensive one by year three - and the difference is predictable before you sign.
Per-seat pricing: predictable, and it punishes growth
You pay a fixed price for each user. The appeal is obvious: the maths is trivial and the invoice is stable. The problem is that cost scales linearly with headcount, so the model taxes exactly the thing a healthy company does.
A 50-person team growing 20% a year ends up with almost 87 licensed users by year three. If the per-seat price never changes, the licence bill has grown 73% - and that is before the vendor's annual uplift. Per-seat pricing is a slow, compounding cost increase that nobody notices because each individual annual step feels small.
Usage-based pricing: cheap to start, hard to forecast
You pay for what you consume - API calls, contacts, gigabytes, transactions, emails sent. The attraction is that you only pay for value received, and small teams start nearly free.
The trade-off is variance. Usage-based bills scale with your success, your traffic spikes and occasionally with a bug in your own code that loops an API call ten thousand times overnight. Worse, usage is the line item most likely to change: vendors regularly reprice units, redefine what counts as a billable event, or introduce minimum commitments at renewal.
Usage-based pricing is the right choice when your usage is genuinely unpredictable and the unit definition is written into the contract. It is the wrong choice when you cannot put a ceiling on it.
Flat-rate tiers: the middle path, with a ceiling problem
You buy a package that includes a set number of seats, contacts or features. Cost is predictable and the value is easy to reason about. The catch is what happens when you outgrow the tier: the jump to the next one is rarely proportional, and vendors know you are not going to migrate mid-year to save money.
Hybrid: what almost every vendor now sells
Most modern software stacks a platform fee, a per-seat component and a usage component. This is not accidental. It lets vendors compete on whichever number the buyer looks at, while recovering margin in the other two.
When you evaluate a hybrid quote, separate the three parts and model each one with your own growth assumptions. Vendors will happily show you a blended figure that hides the component that is about to explode.
The three clauses that matter more than the model
The pricing model sets the shape of the cost. These clauses set the size.
1. Overage pricing
What you pay when you exceed your allowance, and whether you get warned first. Overage rates are often deliberately punitive - they exist to push you into the next tier, not to be a fair price for extra units. Ask for the overage rate in writing, and ask whether exceeding the limit triggers an automatic tier change.
2. Minimum commitment and true-up
Many usage contracts include a minimum annual spend. If you under-consume, you pay the minimum anyway. If you over-consume, you pay the difference at the end of the period. Both directions are your risk, not the vendor's. Model the worst case, not the expected one.
3. Renewal uplift
The cap - or absence of one - on how much the price can rise at renewal. A 5% annual cap and no cap at all are completely different contracts over three years. Vendors resist caps on the headline price but will often accept them on the total contract value. Push for the cap that holds.
How to compare two vendors on different models
You cannot compare a per-seat price with a usage price directly, so do not try. Instead:
- Model the same business scenario across both vendors - same headcount growth, same activity volume.
- Extend both to three years, including a realistic renewal uplift.
- Include implementation and add-ons, which differ by vendor independently of the pricing model.
- Compare the three-year total, and only then look at which one is easier to budget month to month.
The model you choose is a bet on your own growth. Per-seat is a bet that headcount stays flat. Usage-based is a bet that your consumption is predictable. Choose the one that fails gracefully when you turn out to be wrong, because you will be.