Calculating Legacy TCO

Across seven parts this series has named a cost in many forms: a switching cost, a licensing trap, a six-bucket maintenance bill, a drain on engineering capacity, a tilted budget ratio. This final part collects all of it into the one number an economic buyer actually decides on — the total cost of ownership of a legacy system. TCO is where the whole story lands, because it is the figure that converts everything this series has described into a comparison someone can approve: the cost of staying, assembled honestly, against the cost of acting.

It is also the number easiest to get wrong, in both directions. Undercounted, it keeps a system running long past sense. Inflated, it invites the skepticism that gets a whole case dismissed. So this part is as much about how to calculate TCO honestly as about what goes in it — because a TCO figure is only worth what it can survive under challenge.

TCO is forward-looking, or it is nothing

The single most important property of a legacy TCO is that it is a projection, not a snapshot. The cost of keeping a system is not what you paid this year; it is what you will pay over the horizon that matters — typically the same multi-year period over which you would amortize a modernization. A TCO that reports this year’s spend and stops is not a TCO at all. It is an invoice, and an invoice is exactly the rigged comparison this series has warned against from Part 3 onward.

The reason the projection matters is everything the earlier parts established: the costs do not hold flat. Capacity-based licensing climbs with the workload. Specialist labor climbs as the skills thin. The integration tax grows with every new connection, and the risk premium climbs as dependencies age out of support. A legacy system priced on today’s bill is being valued as though it is getting younger. It is not, and the trend line is where most of the cost the comparison usually ignores actually lives.

What goes into the number

A defensible legacy TCO assembles the six maintenance buckets from Part 3, projects each forward, and adds the cost that lock-in specifically forces.

• The six buckets, projected forward. Licensing and support, infrastructure, operations and incident response, specialist labor, the integration tax, and the risk premium — each totalled for the system and projected on its own real trend, not held flat. The capacity drain from Part 4 enters here, priced at the loaded cost of the engineering time the system consumes.
• The switching-cost premium. Lock-in adds costs that are not in the run-rate: the price of having no negotiating power at renewal, and the expected cost of end-of-life events you cannot avoid because there is no portable path off the platform. These are the costs of being unable to leave, and a TCO that omits them undercounts the very thing this series is about.
• The subtraction that keeps it honest. Not all of the above is decision-relevant, because a modern replacement carries costs of its own — infrastructure, operations, labor. The figure that belongs in the comparison is the difference between maintaining the legacy system and running a modern equivalent, not the gross legacy bill. Subtracting the cost you would pay either way is what separates a credible TCO from an alarmist one.

The line-item worksheet

Most legacy TCO estimates come out low for one reason: they count only what lands on an invoice. The categories below are the whole bill, with the line items people most often leave out named inside each. Put your own annual figure against each row, pick the trend it actually moves on, and you have the raw material for the projection. Do not reach for industry averages here. Use your own operating numbers, because the point of the exercise is to price your system, not the category.

Cost category	Count this, including what usually gets dropped	Your annual figure	Trend to apply
Licensing and support	Renewals, per-core or per-seat fees, mandatory support tiers, and the capacity-based increases that arrive as usage grows		Rising with workload
Infrastructure	Servers, storage, network, and the premium for hardware or OS versions old enough that support itself now costs extra		Flat to rising
Operations and incident response	On-call, patching, manual workarounds, and the incidents a fragile system generates that a modern one would not		Rising as it ages
Specialist labor	The loaded cost of the people who can still work on it, plus the premium you pay because that skill pool is shrinking		Rising as skills thin
Integration tax	The extra work every surrounding system does to talk to this one, and the projects it slows or blocks outright		Rising with each new connection
Downtime and risk	Expected cost of outages, security exposure, compliance gaps, and the end-of-life events you cannot avoid without a portable exit		Rising as dependencies age out
Opportunity cost	The engineering capacity this system consumes that would otherwise ship revenue work, priced at loaded cost		Rising with maintenance load
Switching-cost premium	The price of having no negotiating power at renewal, because leaving is not a move you can credibly threaten		Step increases at renewal

The last two rows are the ones a quick calculation almost always drops, and they are frequently the largest. A TCO that skips opportunity cost and the switching-cost premium is not conservative. It is simply wrong in the direction that keeps the system running. Once every row has a figure and a trend, subtract the modern-equivalent baseline from the total. Only the difference between running the legacy system and running its replacement carries into the decision.

Set that net figure against the cost of acting — the honest, ranged estimate of modernizing the system, which the business-case series prices in how much software modernization costs. The comparison, made this way, is the one the whole series has been building toward.

Projecting it over three to five years

A single year’s figure decides nothing. The comparison an economic buyer approves runs over the same horizon you would amortize a modernization on, usually three to five years. Take each worksheet row’s current figure, apply its trend, and sum the horizon. Then set that cumulative cost of staying against the cost of acting: a one-time modernization investment plus the lower run-rate of the modern system once it is live.

The shape of that comparison, with your own numbers in place of the labels:

Horizon	Cost of staying (legacy)	Cost of acting (modernize)
Year 1	Current net TCO, trended up	Modernization spend, plus the legacy run-rate still carried during cutover
Year 2	Higher, as the trends compound	Remaining modernization spend, plus a shrinking legacy tail
Year 3	Higher again	Modern run-rate only
Years 4 to 5	Still climbing	Flat, at the lower modern run-rate
Cumulative	A rising line with no ceiling	A one-time hump, then a lower floor that holds

The legacy column climbs because every trend in the worksheet points up. The modernization column front-loads a one-time cost, then drops below the legacy run-rate and stays there. The year the two cumulative totals cross is the payback point, and it is the single most persuasive number the exercise produces. If your trends are honest and the crossover lands inside the horizon, the case makes itself. If it lands outside the horizon, that is real evidence the system is worth keeping for now, and the caveats below apply.

Why the honest version usually wins anyway

There is a temptation, having assembled all this, to reach for the biggest number the buckets will support. Resist it. The honest TCO is more persuasive than the inflated one, for the same reason the cost-of-inaction argument is most powerful when it is calmest: its weight comes from being undeniable, not alarming. An executive who hears a worst-case headline reaches for the discount; an executive who hears a conservative, ranged, sourced figure engages with the math.

And the honest figure usually wins on its own merits, because the deck was never stacked toward modernization — it was stacked against it, by the rigged comparison that counts the legacy system at a fraction of its cost. Simply counting every bucket, projecting the real trend, and adding the switching-cost premium is enough to flip most comparisons, with no exaggeration required. The most persuasive thing you can do with a legacy TCO is refuse to overstate it, and then let the complete, conservative number do what the partial one never could. The case for acting does not need a big number. It needs an honest one.

What keeps a TCO figure credible

The discipline that makes a TCO credible is the same one that runs through this series. Present a range, not a point — false precision is a tell, and a figure with stated assumptions is one a skeptic can engage rather than dismiss. Date it, because every trend and every benchmark ages. Use your own operating data for the buckets and reserve industry figures — Deloitte’s run-the-business share, McKinsey’s locked-up-capacity range, BCG’s transformation failure rate — for establishing scale, never as a stand-in for your number. And keep Part 1’s boundary in view: if the system is genuinely quiet, stable, cheaply maintained, and not blocking the business, its honest TCO may be low, and that is a legitimate reason not to act. A TCO calculation that can only ever recommend modernization is not a calculation; it is a conclusion wearing a spreadsheet.

Where this leads — and where you do

That closes the series. You now have the full account of lock-in and the maintenance it produces: what lock-in is and its four dimensions, the licensing traps where it bills you, the maintenance cost and the engineering capacity it consumes, the budget ratio it tilts, the portable architecture that resists it, the incremental escape that gets you there, and now the TCO that authorizes the move.

The figures in this series size the problem in general. The only number that prices your system comes from examining it — your buckets, your trends, your switching cost. You can begin assembling that figure yourself with the legacy cost calculator, and turn it into the business case an economic buyer approves. When the question shifts from “what does lock-in cost in general?” to “what does it cost us, and what would leaving actually take?”, that is what a discovery answers. Book a 30-minute discovery call — no deck, just a conversation about your system, your switching cost, and the path off it.