Making Good: the value the frontline was owed all along

Making Good, part 4 of 4. Part one starts here.

I promised the math in part one, so here it is, with a source and a label on every figure.

The short version: for thirty years, connecting the frontline lost the budget argument. The cost was real and the return was modest, so the line item got cut. Agentic AI didn't lower the cost. It raised the return.

How underserved is the frontline?

About 2.7 billion people, roughly 80% of the global workforce, work without a desk, about 100 million of them in the US (Emergence Capital; McKinsey; BCG). By industry estimates they have received about 1% of enterprise software spend (Emergence Capital; Gartner), and only 23% of them say they have the technology they need to be productive (Deloitte).

Most of the people who keep the brand promise got the least software.

What does a disconnected frontline cost?

Survey research puts frontline friction at $80.6 billion a year, with about 376 hours lost per frontline worker per year (Unily and ScreenCloud, 2025; survey-derived). Those hours go to chasing schedules, hunting for the current policy and relaying answers that someone already wrote down.

Turnover compounds it. Hospitality runs at 70 to 80% annual turnover, and replacing one hospitality worker costs about $5,864 (Cornell Center for Hospitality Research).

What changed on the value side?

The connection now carries agents. A connected worker used to mean a reachable worker: the broadcast arrived, maybe it got read. With agentic AI, the same connection answers the policy question and routes the call-off to the manager who has to approve it.

Research on frontline experience points the same way. BCG associates a better frontline experience with 17% higher productivity, 25% lower turnover and 41% lower absenteeism. Those figures are correlational, so they show where value sits rather than proving cause.

Why do most AI pilots miss it?

MIT's Project NANDA reported in 2025 that 95% of enterprise generative AI pilots showed no measurable P&L return. The method is contested, so read it as directional. Parts two and three of this series cover why: the agent didn't know the organization, or the people it was meant for never opened it.

Reach is measurable. Across RedeApp customer networks, adoption runs 80 to 96%, and 44% of monthly users come back every day against a 13% industry average (RedeApp network telemetry, 2026). At one deployment, 75% of the workforce was active within seven days of go-live.

What should an operator do with this?

Start where the cost is already visible to a CFO and the record is strongest, which for most operators means one site and one shift. Pick one workflow, such as call-offs on nights. Measure it before and after, then expand on evidence.

That is what Making Good looks like in practice. The brand promise stays the same. The people keeping it finally get the tools it assumed they had.

I'm writing the long form next: a research brief on the frontline value inversion, with every figure sourced and every limit stated. The RedeApp and Capgemini alliance page will carry it when it's ready.