The number to put in front of the board is the five-year total, not the go-live cost. Build it from one figure, the annual software cost, using a formula a CFO can rebuild on a single slide: five-year cost equals five years of software, plus implementation, plus run cost for the four years after go-live, plus your own people on top. For a tier 2 mid-market program the fully loaded five-year total runs about $2.8M to $5.3M at list prices. For a tier 1 global program it runs about $29M to $54M. Both are several times the implementation figure, and the gap between them is why presenting only the go-live number sets the program up to look failed by year three.
Present the five-year total, or the board judges a success as a failure
The board remembers the number you gave it at approval and measures every later invoice against that number. If you present the go-live cost, that is the figure that sticks. The run cost, the subscription renewals, and the internal team all arrive after the board has set its expectation, and each one reads as an overrun even when it was always going to be spent.
Run the arithmetic on a mid-market program. Suppose you take an implementation quote of $3M to the board and call it the program cost. Over five years the same program carries five years of subscription, four years of run cost, and an internal team. The fully loaded total lands near $8M. Every invoice came in on budget, yet by year three the program has spent more than double the number the board approved, and it looks like a failure.
Nothing in that example went wrong. The only mistake was the number presented at approval. The five-year total is the figure the board can hold you to without the program looking out of control in year two, which is the case for building it before you ask for the money. The cost pillar sets out the ratios this model rests on; this article turns them into a number you can defend.
The formula builds the five-year total from one input
The whole model runs off the annual software cost. Call that S, the subscription or the maintenance bill plus the amortized licence. Everything else is a multiple of S. Written for a five-year horizon, the external cost is:
\text{Five-year external cost} = 5S + mS + 4r(mS)The three terms are the three things you actually pay for. The first, 5S, is five years of software. The second, mS, is implementation, where m is the implementation multiple of 1 to 3 times first-year software, from ERP Research, an independent research firm. The third, 4r(mS), is run cost for years two through five, where r is the annual run cost as a share of implementation, 15 to 25% per Ecosire and SysGenPro. The multiplier is 4, not 5, because support in the first weeks after go-live is usually inside the SI's implementation fee.
Two costs sit outside that line and must be added by hand. The first is dual-run, the period when you pay for both the old and the new system at once. Costbench puts that period at 12 to 24 months, and for a migration it can be the single largest swing in the total. The second is your own organization. The internal team, backfill for the business leads pulled onto the program, and change management add 15 to 30% on top of the external spend, per ERP Research's hidden-cost estimate. Apply that uplift to the external total to get the number the board should see.
If you run perpetual licences with maintenance rather than a subscription, S is the annual maintenance bill plus the licence spread over its useful life. For SAP, maintenance is about 22% of licence value per ERP Research, so a $5M licence base carries $1.1M a year before the amortized licence is added.
Five years of subscription is already 5x, so the 3x-to-5x rule fits a licence
A common rule of thumb puts the five-year total at 3 to 5 times software cost. That rule was built for a one-time licence, and it breaks under a subscription. Five years of subscription alone is 5 times the first-year fee, before a dollar of implementation or run cost is added. Any five-year number an SI shows you that sits at 3x to 5x of a subscription has quietly dropped four years of the subscription itself.
The worked totals below come out at 6.6x to 11x the first-year fee once run cost is in. That is not an overrun. It is what the formula produces when S recurs every year instead of being paid once. The reason to carry the formula rather than the rule of thumb is that the formula does not care whether S is a subscription or a maintenance bill; it counts the same cash either way.
Two worked totals at list, one tier 2 and one tier 1
Here is a tier 2 program at published list prices. ERP Research lists Oracle NetSuite at about $999 a month plus $99 per user. For 300 users that is 300 times $99 times 12, or $356,400, plus $11,988 base, so S is about $368,000 a year before any discount.
Here is a tier 1 program at list. Dynamics 365 Finance lists at $210 per user per month on microsoft.com. For 1,500 full users that is 1,500 times $210 times 12, so S is about $3.78M a year. Tier 1 contracts mix full users with team and self-service users priced lower, so treat list as the ceiling and the discount section below as the correction.
The table runs both through the formula, term by term, at the full 1x to 3x and 15 to 25% ranges. The internal uplift is 15 to 30% on the external total.
| Line in the model | Tier 2, NetSuite, 300 users | Tier 1, Dynamics 365 Finance, 1,500 users |
|---|---|---|
| Annual software, S | About $368,000 | About $3.78M |
| Five years of software, 5S | $1.84M | $18.9M |
| Implementation, mS (1x to 3x) | $368,000 to $1.1M | $3.78M to $11.34M |
| Run cost years 2 to 5, 4r(mS) | $221,000 to $1.1M | $2.27M to $11.34M |
| Five-year external total | $2.4M to $4.05M | $24.95M to $41.58M |
| Fully loaded with internal (plus 15 to 30%) | $2.8M to $5.3M | $29M to $54M |
| As a multiple of first-year software | 7.6x to 14x | 7.7x to 14x |
Two readings matter. First, the tier 1 external total of about $25M to $41.6M lands inside the $20M to $50M range published for a global tier 1 program, which is the check that the formula is calibrated. Second, both programs cost between 7 and 14 times their first-year software fee fully loaded, which is why the 3x-to-5x rule understates a subscription program by half or more.
A discount scales the whole total, because every term is a multiple of S
Every line in the model is a multiple of S, so a discount on the subscription flows through the entire five-year total, not just the software line. This is the one point to make to the board when the negotiated price comes in below list. A 40% discount on the subscription cuts software, implementation, and run cost together, so it cuts the five-year number by about 40% as well.
Take the tier 1 example. The external total at list is about $25M to $41.6M. Apply a 40% discount to S and every term scales down with it, so the external total falls to roughly $15M to $25M. The same proportion holds for the tier 2 program: a 40% discount moves its external total from $2.4M to $4.05M down to about $1.46M to $2.43M.
One caution belongs with that point. The discount scales the software-linked terms cleanly, but dual-run and internal cost are driven by time and headcount, not by the licence price. A deep discount on software does not shorten a 24-month dual-run or shrink the internal team. Model those two lines on their own drivers, not as a percentage of S, or the discounted total will read lower than the program actually costs.