AI Coaching for Job Seekers: What the Evidence Says — and What It Means for Employment Commissioners

The challenge at the heart of employment support has never really been information. Most job seekers know roughly what they need to do: write a CV, apply for roles, prepare for interviews. The challenge is sustained behaviour change — taking the right actions at the right moments, week after week, in the face of rejection and uncertainty. That is a psychological and motivational challenge. And until now, delivering it at scale has required human coaches operating under significant time constraints.

The mechanism that makes AI coaching work

Coach Ava, developed by Big Pebble Change and evaluated in a Birkbeck Business School feasibility study (N=25, 2025), operates through a two-part behaviour change mechanism grounded in published research.

The first part is implementation intentions. Goal Setting Theory (Locke & Latham, 2002) and subsequent implementation intention research (Gollwitzer, 1999) consistently show that the strongest predictor of whether a goal is followed through is whether it specifies when, where, and how. Coach Ava enforces this specificity in every goal-setting conversation: 'Get a job' becomes 'Apply to Bluebird Care and Caring Hands by Wednesday.' That specificity is what converts intention into behaviour.

The second is named accountability. When a user sets a specific goal, Coach Ava follows up on that exact goal — naming what the user committed to and asking what happened. Not a generic reminder. A named check-in. In the V1 study, 18% of action-confirmation messages followed a Coach Ava accountability check-in — the commitment-accountability loop in action.

What the V1 study found

• 78% of respondents reported making measurable progress against their stated goals.

• 65% agreed Coach Ava supported them to take action on their job search.

• 57 discrete action confirmations across the cohort — applications submitted, interviews attended, LinkedIn outreach completed.

• Engagement was significantly associated with goal attainment (Spearman ρ = .587, p = .008).

• 100% onboarding completion — including participants who self-rated as low digital confidence.

What this means for employment commissioners

DWP's own published evaluation of the Job Finding Support programme benchmarks structured coaching at £63 per participant with fiscally positive employment effects. Coach Ava delivers a comparable — in some respects more intensive — coaching mechanism at £26–30 per participant at pilot scale. If the mechanism holds at programme scale (1,000+ participants), the fiscal case is straightforward: the per-participant cost is recovered if the product reduces mean benefit claim duration by one week for 15% of users under even conservative Universal Credit savings modelling.

Three things become possible with AI that cannot be replicated at scale with human-only delivery: personalised coaching adapted to each user's sector, constraints, and barrier type; disclosure that users do not share with human work coaches because Coach Ava sits outside the sanctioning relationship (making previously invisible risk visible); and named accountability at zero marginal cost as volume scales.

The ask: a funded pilot

Coach Ava is approaching first commercial deployment. Big Pebble Change is in active discussion with DWP-contracted Restart Scheme providers for a pilot anticipated in Q3/Q4 2026. Pilot pricing (100–300 participants, 12–16 weeks, approximately £22,000–£34,000) falls below the Procurement Act 2023 direct award threshold, enabling commissioning without competitive tender where appropriate.

The mechanism is validated. The evidence is honest. The ask is a funded pilot that proves — or disproves — the return on investment at scale.

If you are commissioning employment support programmes for DWP, a combined authority, or a Restart Scheme delivery partnership, we'd welcome a conversation. Contact us to discuss a pilot or to request a product demonstration.