AI visit scoring vs double visits:
what the numbers say.

Every field organisation already has a quality tool: the double visit, where a manager accompanies a rep and scores what they see. It is also the tool everyone quietly distrusts — and the numbers explain why.

The coverage math

Take a typical first-line manager with 10 reps, each making around 10 visits a day. That is ~2,200 visits a month per team. A manager doing two full field days a week observes perhaps 100–120 of them. Coverage: roughly 5%. The other 95% of the team’s customer-facing time is invisible — its quality is inferred from the sales number, weeks later, with every confounder attached.

The observer effect

The visits a manager does see are not typical visits. Reps prepare differently for an accompanied day, pick friendlier customers where they can, and perform the standard rather than inhabit it. Doctors and buyers also behave differently with two visitors in the room. The double visit measures best observed behaviour; the business runs on typical unobserved behaviour. The gap between the two is precisely where revenue leaks.

The consistency problem

Ask two managers to score the same recorded visit against the same checklist and the scores routinely diverge — a spread of ±2 points on a 14-point scale is a normal calibration-session finding. Each manager carries their own severity, their own pet criteria, their own relationship with the rep. The consequence: scores are not comparable across territories, so a “strong” team under a lenient manager and a “weak” team under a strict one may be the same team.

Side by side

What the double visit still does better

An honest comparison cuts both ways. A manager in the room reads things a transcript cannot: body language, the state of the relationship, the politics of an account. A live coaching conversation on the drive between visits changes behaviour in a way a dashboard does not. And for new reps, an accompanied day is onboarding, not audit.

What changes operationally

• Span of control grows. When routine review is automated, one first-line manager can support 15–20 reps instead of 10 without losing the coaching relationship.
• Calibration becomes real. Managers and trainers calibrate against the same recordings during rollout — disagreements get resolved once, in the checklist, not repeatedly in the field.
• The rep’s experience changes. Feedback arrives from a neutral system minutes after the visit, with a reference example from a teammate — which lands very differently from a verdict delivered by a supervisor.

How this works in a pharma context — visit models, share of message and compliance — is on the NUMU Pharma FF Intelligence page; the equivalent for promoters and field sales is NUMU Field. Both start with a 30-day pilot on one team, where AI scores are calibrated against your own trainers’ scores on the same visits.