Lesson 6 of 7
An honest TCO and ROI model
8 min
Listen to this lesson
This course discusses ACCELQ by name. Matt Angerer, who runs The Sign-Off Table, works at ACCELQ. Disclosure
Automation ROI is where TCoE credibility is won or lost. Vendor ROI slides compare a perfect automated world with a clumsy manual one; finance has seen them before. This lesson builds a five-year model for Contoso's 240-test portfolio that compares a home-grown framework with a platform, states every assumption, and shows the case in which the platform loses.
Assumptions
| Assumption | Value | Source or reasoning |
|---|---|---|
| Scope | 240 automated tests: 150 web, 60 SAP GUI, 30 API | Lesson 5 portfolio |
| Blended rate | $95 per hour | Agreed with finance in sap-tcoe-401 |
| Change events touching the suite | 6 a year: 2 stack updates, 4 enhancement releases | TCoE calendar |
| Suite runs | 50 a year (weekly plus event runs) | TCoE calendar |
| Manual execution avoided | 2.5 h per test incl. data and evidence; full pack at each stack update, 40% at each enhancement release | Contoso actuals; only runs that would really happen manually |
| Platform subscription | $80,000 a year, illustrative | A placeholder for a quote; not any vendor's price |
| Maintenance ratios | Home-grown: 20% of tests touched per event at 1.5 h; platform: 12% at 0.75 h | Hypotheses the PoC change event measures |
Every assumption in the model, with its source. Replace each with your own data.
The five-year comparison
| Line (expected case) | Home-grown framework | Platform |
|---|---|---|
| Framework or platform setup | 800 h | 250 h |
| Build: web / SAP GUI / API hours per test | 7 / 9 / 3 | 4 / 5 / 2 |
| One-time cost | 2,480 h = $235,600 | 1,210 h = $114,950 |
| Maintenance per year | 432 h (72 h × 6 events) | 130 h (about 22 h × 6 events) |
| Framework upkeep or platform admin per year | 300 h | 100 h |
| Failure triage per year | 300 h (6 h × 50 runs) | 150 h (3 h × 50 runs) |
| Evidence assembly per year | 180 h (30 h × 6) | 48 h (8 h × 6) |
| Running cost per year | 1,212 h = $115,140 + $10,000 infrastructure | 428 h = $40,660 + $80,000 subscription |
| Key-person replacement (year 3) | 400 h = $38,000 | n/a |
| Five-year total cost | $899,300 | $718,250 |
| Break-even subscription | n/a | About $116,000 a year |
Contoso's illustrative five-year TCO. The platform is cheaper while its subscription stays below about $116,000 a year.
Return against manual testing
The benefit is manual execution that would really have happened: the full pack (240 × 2.5 h = 600 h) at each of the two stack updates and 40% of it (240 h) at each of the four enhancement releases, 2,160 hours or $205,200 a year. Year 1 gets half, because the portfolio is built during the year. Weekly runs add protection, but they are not counted as savings, because Contoso would never have run the pack weekly by hand.
| Five years | Home-grown framework | Platform (expected) | Platform (pessimistic) |
|---|---|---|---|
| Benefit (manual hours avoided) | $923,400 | $923,400 | $923,400 |
| Total cost | $899,300 | $718,250 | $987,475 |
| Net | $24,100 | $205,150 | −$64,075 |
| ROI (net ÷ cost) | About 3% | About 29% | Negative |
| Payback | Year 5 | Year 3 (about month 31) | Not within 5 years |
Return over five years. The pessimistic platform case assumes 5.5 / 6.5 / 2.5 build hours, 16% of tests touched at 1.1 h, 4 h triage per run, 10 h evidence per event and a $110,000 subscription.
- Year 1−$133,010 cumulativeBuild and setup $114,950 plus running cost $120,660, against half a year of benefit ($102,600)
- Year 2−$48,470 cumulativeFull benefit $205,200 against running cost $120,660
- Year 3+$36,070 cumulativePayback, about month 31
- Year 4+$120,610 cumulativeRunning cost stable if maintenance stays at the measured ratio
- Year 5+$205,150 cumulativeAbout 29% return on five-year cost
Read the last column as carefully as the middle one. If the platform's maintenance advantage is small and the subscription is high, the home-grown framework wins on cost. That is why the PoC change event exists: the maintenance ratios are the assumptions that decide the case, and they can be measured before signing. CT-TAS frames ROI the same way: savings divided by investment, first in hours and then in money, with average script maintenance time and the share of failing scripts inside the investment.
Rules that keep the model honest
- Count only manual runs that would really happen; label extra protection separately.
- Include triage of false failures and evidence assembly; both are invisible in vendor slides and large in practice.
- Put key-person risk in the home-grown column as an expected cost, not a footnote.
- Never use a vendor's benchmark as an input; use PoC measurements and your own actuals.
- Show the case where your recommendation loses, and say what you would do then.
What would you do?
Contoso's CFO asks Priya: "Your platform case assumes a subscription of $80,000. What if the quote comes in at $120,000?" What is the best answer?
- State every assumption with its source; label the subscription as a placeholder, not a vendor price.
- Contoso's expected case: platform $718,250 against home-grown $899,300 over five years; break-even subscription about $116,000.
- Count only manual runs that would really happen: $205,200 a year at steady state.
- Show the pessimistic case where the platform loses; the PoC maintenance ratio decides which case is real.
Check your understanding · not graded
Lesson check
1. Why does Contoso's model count only two full-pack manual runs a year as savings?
2. In the expected case, about what subscription is break-even against home-grown?
3. Which assumption most decides whether the platform wins?
4. Why should the model show the pessimistic case where the platform loses?