What's the ROI of digital humans?

‍

Emma works around the clock. She speaks fluent Spanish, never takes a lunch break, and saved her employer a projected $1.8 million in her first year on the job. Her approval ratings have run as high as 98% — the kind of number most of us would frame and hang above our desks.

She's also, technically, not a person.

Emma is a digital human built for the City of Amarillo, Texas, by Dell Technologies and UneeQ. And she might be the clearest answer yet to the question every CFO asks when a digital human project lands on their desk: okay, but what's the return?

Let's answer that properly, with real numbers, a real live case study, and a framework you can take away.

What is the ROI of a digital human?

Digital human ROI is the measurable business value a digital human generates, compared to the total cost of building and running it. The formula is the one you already know: (annual value created − annual cost) ÷ annual cost. The value side includes hard savings, like deflected support requests, plus compounding returns like language access, 24/7 availability, and customer satisfaction.

If you're looking at digital humans for immersive training, you'll be looking at metrics such as ramp time, time to proficiency, CSAT, etc. Regardless, the equation remains the same: (annual value created − annual cost) ÷ annual cost.

Let's look at this through the use case of a digital human customer-facing assistant.

‍

What makes the math different from other automation organizations might buy is the "human" part. A digital human pairs conversational AI with a lifelike face, voice, and emotional expression. That combination takes on work that used to need a person, while delivering an experience people genuinely like using.

Both halves show up in the ROI, as one Texas city proved.

Who is Emma, and how did she save Amarillo $1.8 million?

Emma is the City of Amarillo's digital human assistant, launched in December 2024 and built with Dell Technologies on UneeQ's digital human technology. Within a year, she was answering around 16,000 resident questions a month — roughly half the city's total inquiry volume — allowing Amarillo to defer a $1.8 million investment in new call center hires.

Here's the situation Amarillo faced. The city of 200,000-plus fields about 32,000 questions a month through its call center. Its residents speak as many as 62 languages, and around 24% don't speak English. That's a real barrier when city services — paying a water bill, booking a park, using the library — mostly live behind an English-language website and a phone queue.

So the city gave residents someone to talk to. Emma launched in English and Spanish, built to scale to many more languages over time, and available any hour a resident happens to need her.

The results, twelve-ish months in:

• 16,000 of 32,000 monthly resident questions answered by Emma — with plans to grow that share
• $1.8 million in projected annual savings from deferred call center hiring
• More than $1 million effectively saved since launch, per GovTech's reporting
• Approval ratings from 90% to 98% across reporting periods — a floor most help desks would call a ceiling
• City staff freed up to focus on complex, higher-value resident cases

And the returns aren't only financial. "Our approach is to be a conversational city," Assistant City Manager and CIO Rich Gagnon told Route Fifty — framing Emma as a way to rebuild trust between residents and their government, one conversation at a time.

How do you measure digital human ROI? The four returns

Measure digital human ROI across four layers: the cost you avoid, the capacity you reclaim, the reach you unlock, and the trust you earn. The first two are hard savings you can drop straight into a spreadsheet. The last two compound over time — and they're usually the reason the project gets renewed.

1. The cost you avoid

This is the CFO's favorite layer: contacts your digital human resolves that would otherwise cost staff time, plus hiring you no longer need.
‍Metric to track: containment rate (conversations resolved without human escalation) multiplied by your cost per contact. Emma's version: half of Amarillo's monthly question volume handled, and a $1.8 million call center expansion left unspent.

2. The capacity you reclaim

Every routine question your digital human absorbs is time your team gets back for work that actually needs a human.
‍Metric to track: staff hours redirected, and what happens to complex-case resolution once people have room to breathe. Emma's version: Amarillo's agents now focus on the tricky, high-stakes resident cases instead of repeating office hours 400 times a day.

3. The reach you unlock

A digital human works at 3 a.m., speaks multiple languages, and never puts anyone on hold. That means serving people your current setup quietly excludes.
‍Metric to track: after-hours conversations, non-primary-language sessions, and first-time users. Emma's version: a front door to city services for the 24% of Amarillo residents who don't speak English.

4. The trust you earn

Satisfaction, repeat usage, and brand (or civic) goodwill. Soft to name, easy to measure, and the layer that turns a pilot into a program.
‍Metric to track: approval or CSAT scores, sentiment, and returning users. Emma's version: approval between 90% and 98%, and a city willing to build its whole engagement philosophy around her.

How do you build the ROI case before you launch?

To build a digital human ROI case, baseline your current contact volume and cost per contact, set a containment target, measure satisfaction from day one, and report against all four returns every quarter. In practice:

1. Baseline everything. Contact volume, cost per contact, wait times, satisfaction scores — captured before launch. If you don't measure the before, you can't prove the after.
2. Price a conversation. Work out what one resolved inquiry costs you today, in staff time and overhead. That single number turns every contained conversation into countable savings.
3. Set a containment target. Decide what share of conversations your digital human should resolve without escalation, and by when. Amarillo reached roughly half its inquiry volume within a year.
4. Measure delight, not just deflection. A deflected-but-furious customer is negative ROI wearing a disguise. Track satisfaction from the very first conversation.
5. Report quarterly against all four returns. Lead with the hard savings, then show the compounding layers, so it survives budget season.

Where do digital human projects lose ROI?

Digital human projects lose ROI when there's no baseline data, when the deployment is AI for the sake of AI, or when the digital human is treated as a launch-day gimmick instead of a team member with a job description and KPIs.

The most common leaks:

• No before-numbers. Without a baseline, even a great deployment produces a shrug instead of a business case.
• Innovation theater. A digital human deployed to look cutting-edge, with no actual job to do, returns exactly what it's given: nothing.
• A face with no brain (or a brain with no face). Real ROI needs the full stack — conversational intelligence, lifelike expression, and integrations into the systems that actually resolve things. That's the job of a complete Digital Human API, and it's the difference between a helpful colleague and a chatbot in a costume.
• Set-and-forget. Emma keeps evolving — new capabilities, growing question coverage, continuously reviewed performance. ROI is a habit, and the returns compound for teams that keep tuning.

Frequently asked questions about digital human ROI

If you have a question about setting ROI objectives and hitting them, check out the FAQ below. Cant find an answer to your specific questions? Book a call with our team and we'll help.