Calculate the ROI of Customer Discovery Step-by-step

For Product Managers, everything you do needs to drive product improvements that (ideally) have positive business impact.

For UX Researchers, that connection has always been a bit more vague.

But whichever camp you fit in, calculating the business impact of running customer discovery is crucial. It helps you:

  • Show the value of your own work (and keep your job)

  • Ensure the team understands the value of customer discovery

  • Learn to estimate and prioritize the most impactful customer discovery over time

But even the best product management books and research how-to guides don’t tell you how to calculate customer discovery’s impact in business terms.

This guide will.

I’ll walk you through the simple “napkin math” that anyone can use to calculate the ROI of your project.

You’ll be able to:

  • Generate a range of reasonable $$ amounts that you likely saved and earned the business

  • Sum up the impact in clear terms that resonate with executives

  • Show 3 types of financial impact

  • Check whether research is worth doing by comparing costs with and without research

Quantify customer discovery impact

Let’s use an example scenario:

Your team developed a new feature. But there were a few different ways you could build it.

You decided to test three versions of the feature in a prototype before building any of them.

The results show that presenting Variant A generates more interest and converts more users in tests than Variant B and Variant C.

The team builds A.

The business has $552k more money as a result.

Here’s how to get that estimate, and do your own.

1. Collect the numbers from your team

If you don’t have the numbers, ask someone knowledgable in your team to estimate:

  • The average salary of your team’s engineers

  • The average/typical time to develop a feature/product change like this one

  • An example of the time it took to fix a feature after launching (especially if it wasn’t converting as planned)

Example:

🧑‍💼 The average engineer earns $12k per month = $600 per day

⏳ The average feature dev time = 30 days

💸 Total cost of a feature: $18k (30 days x $600)

2. Savings from avoiding the build-fail-build again cycle

You probably wouldn’t build all three versions of the feature and launch them all.

But you might build one, launch it, and find that it completely failed.

Then the team has to fix it. And it’s not always clear what to fix.

The time to build, then launch, then figure out what’s wrong, re-build and launch a second time is a waste of money.

Here’s how to calculate that build-fail-build cycle’s cost:

  • Estimate the time to fix a feature and relaunch (assume 15 days)

  • Add that to the dev cost of a feature (because you’re developing a new one all over again)

Example:

🛠️ Average “feature fix time” cost = $9k (15 days x $600/day)

⏳ The feature build cost = $18k (30 days x $600/day)

💸 Build-fail-rebuild-launch cycle cost = $9k to 27k

Building the wrong thing first and then fixing it costs 9k. But it probably costs more.

There are a few ways to look at this. The option you choose for your estimate should be based on the option that most resembles how you’ve seen your team operate in the past.

👉 On the high end: you saved $27k if your team has previously built features that didn’t work out, you tried to fix them, and eventually had to start over.

👉 In the middle: you saved $18k if you ditched the first failed feature and built a new one.

👉 On the low end: you saved $9k by not needing to revise, redevelop and relaunch the feature again later

3. Adding revenue by converting sooner (saving opportunity cost)

Your team built Version A instead of the other two versions.

It converted real usage in the product.

You can say you’ve actually earned money in the past month instead of spending money by delaying the build of a better feature.

Example:

👯 The additional feature converted 1000 new users in the month since release (nothing else was added or changed to the product pitch)

💳 Each new user account has a lifetime value of $500

🤑 Total revenue gain = $500k (1000 users x $500)

What’s our estimate range here?

👉 On the high end: you earned the company $500k by launching a feature that converts

👉 On the low end: you earned the company $500k, but might lose some to churn, so let’s say you end up with $375k net

4. Reducing the cost of customer support

How much does it cost your team to have customer support?

Each support ticket has a cost in time and financial value.

Developing the right feature that does not increase customer support requests is yet another way to save time and money, while generating revenue.

For an accurate calculation, collect:

  • the average salary of your customer support team

  • the number of tickets they solve each day

  • the total number of current users you expect to use the new feature

  • some measure of testers’ uncertainty for each of the variant tests (ex: how many questions they asked, degree of uncertainty, etc)

Example:

  • Customer support team members earn $5k/month ($250/day)

  • The team closes 10 tickets per person, per day ($25 per support ticket)

  • You expect about 1/3 of your 30k users to use the feature (10k users)

  • If you launched a less understandable feature version, and 1 of every 10 users tried the feature and asked for support, you would get 1000 new support tickets.

  • 1000 x $25 = $25k cost of support for a new, lesser feature

Add everything together

If your team had launched an untested, unvalidated feature version:

Total cost $52k (that likely didn’t result in new revenue).

  • $25k Cost of support for a new, suboptimal feature version

  • $27k Cost of launching the wrong feature version, then revising, redeveloping and launching a new feature a second time

If your team tested and launched a feature that converted from the start:

Total value saved + generated is $552k.

  • You earned the company +$500k

  • You saved the company $25k cost of support

  • You saved the company $27k Cost of revising

These calculations are not perfect, and depend on the accuracy of your input.

But you can always use a range to provide more accuracy and believability to your stakeholders.

Using a range

On the high end of the range, we’ve ended up with $552k of combined revenue and savings from launching a feature we validated first.

On the low end of the range, we can try to be more conservative:

  • We’ll say that 1 in 4 new users acquired with this feature add churns (500k * 75% retained = $375k)

  • We use $9k for “feature fix” savings instead of the cost of redeveloping a feature a second time

  • We use $20k of support troubleshooting saved

In that conservative estimate, we still get >$400k of value.

“But it costs money to run tests!”

That’s true! But not a lot when we look at the numbers.

Let’s do the napkin math there.

Our test wasn’t run by minions, so the company paid you or a UX Researcher to carry this out.

⏳ It took 2 weeks to test 3 variants (15 days)

🧑‍💼 A PM and a UX Researcher/other product person carried it out, with an average salary of €10k/month ($500/day for 20 workdays)

🧰 You needed a platform to run the tests on at $100/month

💳 You paid $900 for participants to test (€60 x 15 testers)

💸 Total cost of running the test = $8.5k

(7.5k labor + $100 test platform + $900 for testers)

If we’re on the conservative side, and subtract the cost of running a 2-week test, we’ve still saved $395k by testing and finding the best version of our feature before building it.


Want support to build ROI estimates like this into your discovery process?

Previous
Previous

How can we figure out what customers are willing to pay?

Next
Next

UX Research Impact Tracker (Template)