Lesson 7.3: Measuring value and proving return
A project the customer cannot defend to finance will not be renewed. Learn to turn results into rupees and report ROI, payback and NPV honestly.
By the end of this lesson you will be able to show what a customer project was worth, in terms the customer's finance team will accept. You will learn the main kinds of value and how to convert each into money, how to separate the effect of your work from everything else that changed, three measures of return (the return on investment, the payback period and the net present value) and how to write a one-page value report that states its assumptions plainly. You will see three value cases, one of which was challenged by the customer's finance team, and use a simulator that puts the three measures together. This lesson follows 7.2, where you learned to carry customer needs back to the product team. Here you learn to carry results forward, to the people who decide whether the customer keeps paying. You need to be comfortable with percentages and simple arithmetic, and no more.
A pilot succeeds when the metric moves. A customer relationship succeeds when someone senior is willing to say, in a budget meeting, "this was worth it". Those are different events. The metric, say the share of accounts paying within seven days, belongs to the project. The senior person's statement depends on a number that finance can check: rupees saved, rupees earned, risk avoided, against what was spent. Many technically successful projects are not renewed because the engineer stopped at the metric. The customer's sponsor then has to build the financial case alone, usually at the last minute and with guesses. You can do it far better, because you have the data.
Why prove value
There are four reasons to measure value deliberately, each with a different audience.
For the sponsor, it is a defence. Budgets are cut and priorities shift. A sponsor with a clear, credible figure can protect the project. One without it is exposed.
For renewal and expansion, it is the case. A customer deciding whether to extend or widen the work will ask what they got. A clear answer turns a pilot into a programme.
For your company, it is evidence. Verified results are the most persuasive material for the next customer, and for the product team deciding where to invest, as Lesson 7.2 described.
For you, it is learning. Measuring what worked, and what did not, shows where your effort pays most.
There is also a reason not to avoid it. If the value is real, a fair measurement will show it. If it is not, you want to know sooner than the customer does.
Kinds of value
Value comes in several forms, and the first step is to name which ones your project produced. A rough taxonomy covers most cases.
Cost saved. Less spent on something: fewer calls handled, less rework, lower error costs, fewer paper forms. This is usually the easiest to count.
Revenue gained. More income: more accounts collected, more customers retained, a higher conversion rate. It is the most persuasive to a sponsor, and the hardest to attribute.
Time freed. People spend less time on a task. This is valuable, but only turns into money if the time is used for something else, or if headcount is reduced or growth is absorbed without hiring. Be clear which.
Risk reduced. A lower chance of a costly event: a regulatory penalty, a security incident, an error that harms a customer. Valued as the probability reduction times the size of the loss, with great care.
Quality or experience improved. Faster service, fewer errors, happier customers. These matter, and are often measured by a proxy, such as complaints or satisfaction scores, then connected to money only with caution.
A project often produces several, and a good value report lists them separately and says how firmly each is measured. A figure built on cost saved, with the method shown, is worth more than a larger figure that mixes in guesses about risk and goodwill.
Hard and soft numbers
It helps to divide value into two groups by how solid the evidence is.
Hard value is measured directly from the customer's records and can be checked by their finance team: fewer calls handled, a higher share collected, fewer hours of overtime paid. The figures appear in the customer's own systems, and the chain from the project to the money is short.
Soft value depends on assumptions or estimates: time freed that has not changed headcount, risk avoided, improved experience. It is real, and harder to defend.
A practical rule is to lead with the hard value, report the soft value separately and clearly labelled, and never add the two into one headline figure without saying so. A sponsor who defends a figure built mostly of soft estimates, and is challenged by finance, loses credibility, and so do you. A conservative claim that survives scrutiny is better than a larger one that does not.
Separating your effect from everything else
The central difficulty in measuring value is attribution. Numbers change for many reasons: the season, the economy, a new law, another team's initiative. If the share of accounts paying on time rose by five points while you were running a pilot, how much of the five points was yours?
Lesson 2.3 gave you the main tool, a comparison. Three approaches, in order of strength:
A control group. Compare the group that received the change with a similar group that did not, over the same period. The difference is the best estimate of your effect, because everything else affected both groups equally.
A matched comparison. Where a control is impossible, compare with a similar group, such as another branch, after checking that they looked alike beforehand.
Before and after. Compare the same group over time. It is the weakest, since anything else that changed is mixed in. If you must use it, show the trend before the project, so that a rise that was already under way is not credited to you.
Whichever you use, write down the method, and let the customer's analysts check it. Two further points keep the numbers honest.
Do not count the same benefit twice. If faster calls and fewer calls both reduce the cost of the same hours, count the saving once.
Use the customer's own figures. The cost of a call, the value of a collected account and the hourly cost of staff should come from the customer's finance team, not your guesses. Figures that the customer supplied and signed off are far harder to dispute.
Predict first: a pilot avoided about 1,800 calls a month. The sponsor asks you what that is worth. What do you need to know before you give a number, and what would you not do? Decide before reading on.
You need to know what a call costs the customer, and whether avoided calls turn into money. The first is a figure from their finance team, perhaps ₹90 a call including staff, systems and overhead. The second is subtler: if the callers are paid the same whether they take 10,000 calls or 8,200, the cash saving is zero until the team is smaller or growth is absorbed without hiring. The honest answer separates the two: "1,800 fewer calls is about ₹1.6 lakh a month at your cost per call. That is a cashable saving only if you reduce hours or avoid hiring. Otherwise it is capacity, which you can use for something else." What you would not do is multiply 1,800 by a cost per call and present the result as money saved without the caveat.
Three measures of return
Once the value is in rupees, you can set it against the cost. Three measures are standard, and finance teams will expect at least one.
Return on investment (ROI). The gain from the investment, minus its cost, divided by its cost, usually as a percentage. If a project cost ₹40 lakh and produced ₹90 lakh of net benefit over three years, the ROI is (90 − 40) ÷ 40, which is 125 per cent. It is simple and widely understood. It ignores timing: a return that arrives in year one counts the same as one that arrives in year five.
Payback period. The time it takes for the benefits to repay the cost. With a cost of ₹40 lakh and a net benefit of ₹33 lakh a year, the payback is about 1.2 years, or fourteen to fifteen months. It answers the question executives often ask first, "how soon do we get our money back?" It ignores what happens afterwards, and, in its simple form, it assumes the benefit is the same each year.
Net present value (NPV). The value today of all the future benefits, minus the cost. The idea behind it is that money later is worth less than money now, because money now can be invested. A discount rate, such as the customer's cost of capital, is used to shrink future amounts. A positive NPV means the project is expected to create value after allowing for the time value of money. It is the most complete measure, and the one that finance teams trust most, though it depends on the discount rate and on the forecast of benefits.
Use the language the customer's finance team uses. Ask which measure they prefer and what discount rate they apply. A report that gives payback, ROI and NPV, with the assumptions behind each, covers most readers.
The simulator: put them together
The simulator below computes the three measures from a handful of inputs. Set the yearly benefit you can show, the yearly cost of running the solution, the one-off cost of building and deploying it, the discount rate, the number of years you count and the share of the claimed benefit you would be willing to defend to finance. The last input is a haircut: a deliberate reduction of the benefit, as a margin of safety.
It shows the payback in months, the ROI over the period, the net present value and the cumulative net benefit year by year. The model assumes equal benefits each year, and ignores growth, tax and the cost of delay, which a real finance team would handle. It is a way of getting a first, honest answer quickly.
Try three experiments. First, keep the defaults and read the three measures. Second, set the share you would defend to 100 per cent and see how much the haircut was costing the headline. Third, raise the discount rate to 20 per cent, which describes a customer with expensive capital, and see how NPV falls while payback and ROI do not change.
At the defaults, a claimed yearly benefit of ₹60 lakh, with the 80 per cent haircut applied, gives ₹48 lakh. Less the running cost of ₹15 lakh, the net is ₹33 lakh a year. Against a ₹40 lakh build, payback is about 14.5 months, the three-year ROI is 147.5 per cent and the net present value at 12 per cent is about ₹39 lakh. The haircut cost the headline ₹12 lakh of benefit a year, and it bought a figure that finance is far more likely to accept.
The lender example turns a comparison into rupees in a few lines of arithmetic, and the same steps apply to any pilot with a control group. Set your own figures: the baseline, the control group's result, the pilot group's result, how many accounts a month, the average amount and the share of it that finance values.
At the defaults, five more points on 1,200 accounts is 60 more accounts paying in the first week each month, worth about ₹3.1 lakh a month, or ₹37 lakh a year. Had the engineer compared with the old baseline instead of the control group, the gain would have looked the same only because the control also stayed at 41. If the control had risen to 44, the effect, and the value, would have been much smaller.
Writing the value report
The value report is the document that carries your measurement to the people who decide. Keep it to one or two pages, in the same spirit as the one-page plan and the weekly update. A useful structure has six parts.
1. The headline. Two lines: what the project delivered in terms the sponsor cares about, with the main figure and how firm it is. "The pilot raised first-week repayments by five points, worth about ₹48 lakh a year in hard value, at a one-off cost of ₹40 lakh and ₹15 lakh a year to run."
2. The result against the target. The metric from Lesson 2.3: baseline, target and result, with the comparison used and its margin.
3. The value, by kind. A short table of hard and soft value, each with its source, such as "customer finance: cost per call".
4. The return. Payback, ROI and NPV, with the discount rate and the period, and the haircut if you used one.
5. The assumptions. A short list in plain words: where each figure came from, what was assumed, what would change the answer, and what is not counted.
6. The recommendation. What you propose next, such as to roll out to more teams, with the expected effect and what it would take.
Two habits make the report credible. Show your method, so that someone else can recompute it. And invite the customer's finance team to check it before it goes upward. A figure the finance team has already reviewed becomes theirs, and is defended by them.
When the numbers are challenged
Sooner or later a finance person will question a figure. It is a good sign, since it means the report is being taken seriously. A few principles help.
Welcome the challenge. Say, "That is a fair question. Here is how I got it." Do not defend.
Show the source. Point to the figure the customer supplied, and the method, and let them check it.
Concede what is soft. If the challenge is to a soft figure, agree that it is an estimate, and show what the result looks like without it. A conclusion that survives the removal of the soft part is stronger.
Offer to re-run with their assumptions. If finance prefers a different discount rate or a bigger haircut, run it. If the project still pays back, they will believe it. If not, you will have learned something.
Do not argue about what you cannot prove. A claim you cannot support is a claim to drop.
Worked example 1: the lender's value
The lender's pilot reached 46 per cent against a baseline of 41 for first-band accounts paying within seven days, with a control group that stayed at 41. The engineer works with the lender's finance analyst to convert this.
The analyst provides the figures: the first band has about 1,200 eligible accounts a month, the average overdue amount is ₹26,000 and a first-week payment avoids a month of follow-up costs and some write-off risk, which finance values at 20 per cent of the amount. Five more points of 1,200 accounts is 60 additional accounts paying in the first week each month. At ₹26,000 and a 20 per cent value, that is about ₹3.1 lakh a month, or ₹37.4 lakh a year. The call minutes saved, from callers no longer chasing those accounts, come to about ₹11 lakh a year at the lender's cost per minute.
Those two are hard, and come from the lender's own data and rates. Freed caller time that has not yet been redeployed, ₹14 lakh, and a lower complaint risk, ₹6 lakh, are shown separately as soft. The headline is about ₹48 lakh of hard value a year, to be set against ₹40 lakh to build and ₹15 lakh a year to run. The finance analyst checks the figures, and the sponsor takes the one-page report to the budget meeting.
Worked example 2: the hospital, careful with claims
The hospital group's bed screen cut the median time to find a bed from 18 minutes to 9, with a matched comparison. The engineer is tempted to claim a benefit in patient outcomes, since faster beds probably help patients.
The engineer resists. What is measured is time. The hard value is the bed managers' time: about 40 requests a day across the two hospitals, saving about 9 minutes each, roughly six hours a day. At the staff cost the hospital supplies, that is a measured saving of staff time, though not a cash saving unless it changes rostering. The patient effect, faster admission from the emergency department, is plausible and not measured. The report lists it as a possible benefit, with a recommendation that the hospital's clinical team study it, and does not put a rupee figure on it.
The restraint is correct. In health, claims about outcomes need clinical evidence, and an engineer's estimate is not evidence. The report remains credible because it says what it measured and what it did not. The clinical team later takes up the study, with the engineer's data.
Worked example 3: the figure that was challenged
A support assistant pilot reports a saving of ₹60 lakh a year: it cut the time to a first reply from 42 minutes to 28, across forty thousand tickets a year, and the engineer multiplied the 14 minutes saved by every ticket by the agents' hourly cost.
The finance team challenges it. Agents are salaried, they point out, and no one will be let go, so the saving is capacity and not cash. Second, the time to a first reply is not the same as the agent's working time, since part of it is waiting in a queue. Third, the comparison was before and after, and a seasonal drop in ticket volume had also helped.
The engineer concedes all three. The report is rewritten: the cashable value is nil, the capacity freed is about a third of what was claimed, and a control group, which exists for part of the period, shows that about two thirds of the speed-up is due to the assistant. What remains is a smaller, defensible figure, along with a clear statement of what the freed time could be used for. The sponsor's trust in the engineer's later reports rises, because she saw the first one corrected honestly. The first figure, though, cost a month of credibility and a delayed renewal decision, a price that a haircut and an early check with finance would have avoided.
The challenged figure shrank in three steps. The chart shows what was claimed, what survived as capacity and what survived as cash.
Common mistakes
Stopping at the metric. Convert the result to money, with the customer's finance team, in terms they accept.
Counting time freed as cash. Say whether it is a cashable saving or capacity, and what it could be used for.
Mixing hard and soft. Report them separately, and lead with the hard value.
No comparison. Use a control or matched group, and show the trend if you must use before and after.
Your own cost figures. Use the customer's rates and have them sign them off.
Double counting. Count each benefit once.
Hiding assumptions. State the source of every figure and what would change the answer.
Defending a weak figure. Concede what is soft, and show what survives.
Claiming outcomes you did not measure. In particular for health or safety, say what was measured and what was not.
Key takeaways
A project succeeds when the metric moves, and a customer relationship succeeds when a senior person can defend the spend with a number that finance can check.
Name the kinds of value: cost saved, revenue gained, time freed, risk reduced, quality improved. Lead with hard value, and report soft value separately and clearly.
Separate your effect from everything else with a control group, a matched comparison or, weakest, before and after. Use the customer's own figures, and count each benefit once.
Time freed is capacity, not cash, unless it changes headcount or avoids hiring. Say which.
Report payback, ROI and NPV in the language of the customer's finance team, with a haircut for safety and the assumptions in plain words.
Welcome challenges, show your sources, concede what is soft and show what survives. A modest claim that stands is worth more than a large one that falls.
Check yourself
1. Name the five kinds of value.
Cost saved, revenue gained, time freed, risk reduced, and quality or experience improved.
2. What is the difference between hard and soft value, and which should lead a report?
Hard value is measured directly from the customer's records and can be checked by finance. Soft value rests on assumptions. Lead with the hard value and report the soft separately.
3. A project costs 50 lakh to build and gives a net benefit of 30 lakh a year. What is the payback and the three-year ROI?
Payback is 50 ÷ 30, about 1.7 years or 20 months. The three-year ROI is (90 − 50) ÷ 50 = 80 per cent.
4. Why is "1,800 fewer calls, so ₹1.6 lakh saved a month" an incomplete claim?
The saving is only cash if hours or hiring change. Otherwise it is capacity. The claim should say which, and use the customer's own cost per call.
5. Why apply a haircut to the claimed benefit?
It builds a margin of safety, so the figure is more likely to survive the finance team's questions, which makes the whole report more credible.
Try this
This exercise takes about forty minutes and gives you a value report for a project you know, or an imagined pilot.
1. Write the result in one line: the metric, the baseline and the change, and how you compared.
2. List the kinds of value it produced. Mark each as hard or soft, and write where its figures would come from.
3. Convert the hard value into rupees a year, showing the arithmetic. Note which figures you would ask the customer's finance team to supply.
4. Write the soft value separately, with a reason it is soft.
5. Write the costs: one-off, and yearly to run.
6. Use the simulator to get payback, ROI and NPV, with a haircut you would defend. Write them with the discount rate and the period.
7. Write five assumptions in plain words, and what each would change if wrong.
8. Write the headline, and the recommendation for what to do next. Ask a colleague to challenge the three least firm figures.
A good result is a one-page value report with a headline, a result, a value table split into hard and soft, three measures of return, assumptions and a recommendation. It is the document that turns a technical success into a renewal.
Sources
The return measures are standard financial definitions. Return on investment is the net gain divided by the cost. Simple payback is the cost divided by the yearly net benefit when benefits are equal each year. Net present value is the sum of future benefits discounted at a chosen rate, minus the cost. I checked these against several course-note and calculator summaries, which differ in details such as whether the first cash flow is at the start or the end of the first year, so confirm the convention with the customer's finance team.
The three cases, the figures and the simulator's defaults are the author's illustrations for teaching, not data from real companies. The classification of value, the haircut and the report structure are practical guidance written for this lesson, not industry standards, and nothing here is financial advice.
Tags: Forward Deployed Engineer ROI Business value Measurement Customer success