What a licence quote leaves out
Total cost of ownership is the sum of everything a system consumes between the decision to buy it and the day its data is safely somewhere else. The licence is the visible part. Around it sit four categories that arrive as internal effort rather than invoices, which is exactly why they escape the business case.
Getting it working. Configuration, data loading, permission models and the vendor's professional services. This is the single largest one-off item on most enterprise systems and the one most often estimated by the vendor rather than by you.
Connecting it. Identity, finance, the data warehouse, the ticketing system. Integration work is where implementations overrun, because each connection has an owner on the other side with their own schedule.
Running it. Someone administers access, configures new workflows, handles upgrades and answers questions. It is rarely a whole role and it is never zero, and at a quarter of a loaded role it can exceed the licence fee on a mid-sized contract.
Leaving it. Data extraction in a usable format, archiving what you must keep for compliance, and a period running two systems side by side. Teams that have never migrated off a major system consistently estimate this at zero, and it is never zero.
Adding these up changes decisions rather than just numbers. Two products whose licence quotes differ by 20% routinely have total costs that differ in the opposite direction, because the cheaper licence needs more integration work or more administration.
The four terms, and why the licence compounds
The licence term is a geometric series, not a multiplication. Year 1 is the quoted price; every year after that carries the uplift, so the total is L × ((1+u)^Y − 1) / u. At 7% over five years that multiplier is 5.7507, not 5 — the uplift alone adds 15% to the licence line before anything else is counted. When the uplift is zero the expression collapses to L × Y.
Administration is FTE × loaded salary × years. Use fully loaded cost, not base salary: benefits, payroll tax and overhead typically add a substantial multiple, and using base salary understates the largest recurring non-licence item. If administration is split across three people at 10% each, enter 0.30.
One-off costs are added once. Implementation, integration and training land in year 1; exit lands at the end. They do not recur, which is why a longer ownership period lowers cost per user per year even at a constant licence — the same setup effort is spread over more years.
Cost per user per year, TCO / (users × years), is the figure to compare between products and against alternatives. It is the only output here that is directly comparable across systems of different sizes, and it is what a finance team will ask for.
One deliberate omission: this model does not discount future cash. A dollar in year five is treated as a dollar today. That is the right default for comparing two similar subscription profiles, and the wrong one for comparing a large upfront purchase against a subscription — for that comparison, discount both streams at your cost of capital first.
Worked example: a $60,000 licence over five years
A vendor quotes $60,000 a year with a 7% annual uplift. Implementation is $45,000, integrations $25,000, training $12,000. You expect to spend a quarter of a role administering it, at a loaded salary of $140,000, and to budget $15,000 to get out at the end. There are 150 users and a five-year horizon.
- Licence over five years. $60,000 × ((1.07⁵ − 1) ÷ 0.07) = $60,000 × 5.750739 = $345,044.34. Year by year that is $60,000, $64,200, $68,694, $73,502.58 and $78,647.76.
- Administration. 0.25 × $140,000 = $35,000 a year, × 5 = $175,000.
- One-off costs. $45,000 + $25,000 + $12,000 + $15,000 = $97,000.
- Total cost of ownership. $345,044.34 + $175,000 + $97,000 = $617,044.34.
- Year-one cost. $60,000 + $35,000 + $82,000 of upfront one-offs = $177,000, nearly three times the licence quote.
- Share that is not licence. ($617,044.34 − $345,044.34) ÷ $617,044.34 = $272,000 ÷ $617,044.34 = 44.1%.
- Cost per user per year. $617,044.34 ÷ (150 × 5) = $617,044.34 ÷ 750 = $822.73.
- Multiple on the quote. $617,044.34 ÷ $345,044.34 = 1.79× the licence spend.
Two things follow. Negotiating 10% off the licence saves $34,504 across five years, which is 5.6% of total ownership — worth having, but not the main lever. And the year-one figure of $177,000 is the number to put in a budget request, because a request built on $60,000 will be back for more before the first quarter ends.
How to use the numbers
Compare products on cost per user per year, never on licence. It normalises for user count and horizon, and it is the figure that survives a change in either. When two products are within about 10% of each other on this measure, the decision belongs to capability and risk rather than to cost, because the estimate is not precise enough to separate them.
Read the non-licence share as a guide to where effort belongs. Below about 30%, the licence dominates and negotiation is the highest-value activity. Above 50%, the licence is the smaller half and you should be scrutinising the implementation quote, the integration scope and the administration assumption instead — a 10% licence discount is a rounding error against a 40% implementation overrun.
Watch the year-one figure separately from the total. It is usually two to three times the steady-state annual cost, and it is the number that breaks budgets. A department that approved the licence and then discovers implementation, integration and training in the same fiscal year will slow the project down while it finds the money.
Then sanity-check the administration input, which is the one people guess. Ask the team that runs the comparable system you already own how much of their week it consumes. If the answer is more than a day, 0.25 FTE is too low, and administration is about to become your largest non-licence cost.
Licence-only cost against full ownership cost per user
| Users | Licence-only per user per year | Full TCO per user per year | Ratio |
|---|---|---|---|
| 50 | $1,380.18 | $2,468.18 | 1.79× |
| 100 | $690.09 | $1,234.09 | 1.79× |
| 150 | $460.06 | $822.73 | 1.79× |
| 300 | $230.03 | $411.36 | 1.79× |
| 500 | $138.02 | $246.82 | 1.79× |
The ratio is constant across the table because user count divides both columns equally. That is the point: however you scale the deployment, this system costs 1.79 times what its licence says, and only a change in the one-off or administration inputs moves that multiple.
Inputs people get wrong
- Using base salary for administration. Loaded cost includes benefits, payroll taxes and overhead, and it is substantially higher. Using base salary understates the largest recurring non-licence line.
- Accepting the vendor's implementation estimate. It covers the vendor's work, not yours. Your data cleansing, your approvals, your testing and your change management are all additional.
- Setting exit cost to zero. Every system is eventually replaced. Budget the extraction, the archive and the parallel-running period, and check the contract for what format your data comes out in.
- Ignoring the uplift. Over five years a 7% escalator adds 15% to the licence total against a flat assumption. Over ten years it adds 38%.
- Counting training as the course fee only. The larger cost is the time users spend not working, which is real even though it never appears on an invoice.
- Forgetting infrastructure the software needs. Storage, egress, a data warehouse seat or an integration platform licence are all ownership costs of the decision, even when a different budget pays for them.
Key terms
- Total cost of ownership
- The sum of acquisition, operating and disposal costs across a system's whole life, rather than its purchase price alone.
- Loaded cost
- An employee's full cost to the organisation: salary plus benefits, employer taxes and allocated overhead. Always higher than base salary.
- Run rate
- The steady-state annual cost once one-off costs have passed — here the final-year licence plus administration. It is the figure to compare against next year's budget.
- Exit cost
- What it costs to leave: data extraction in a usable format, compliance archiving, and running the old and new systems in parallel during cutover.
Where the ownership view changes the decision
The clearest case is a low licence price attached to heavy configuration. Platforms that are cheap per seat and expensive to implement can carry a higher total cost than an expensive product that works out of the box, and only an ownership view catches it. Ask any vendor for a reference customer of your size and ask them how many hours the implementation actually took.
The second is the build alternative. Once administration and integration are on the ledger, a bought system's cost stops looking like a subscription and starts looking like an ongoing commitment — which is the right basis for comparison against writing it yourself, and exactly what the build versus buy calculator models.
The third is the renewal. Ownership cost is the reason a renewal negotiation should not start with the licence rate. Check the seats you actually use first, model the escalation with the uplift calculator, and consider whether consolidating removes an entire administration burden rather than shaving a percentage off one line. And where the product's price scales with usage rather than seats, the pricing model comparison belongs in the same analysis.