Every client-facing team runs the same quiet experiment, whether they'd call it that or not. You hire someone capable, give them a few weeks of onboarding, then point them at real customers and hope the gaps are small ones. The customer becomes the training ground. It usually works. The bill arrives in the times it doesn't, and in how long "usually" takes to show up.
Nobody designed it this way on purpose. It's just the path of least resistance, and because it mostly works, it never quite rises to the level of a problem anyone owns. Which is exactly what makes it expensive.
The cost is real, it just hides well
It hides because it almost never appears as a line item. There's no invoice for the deal that cooled after a shaky first call, the renewal that quietly didn't happen, or the client who decided your new account manager wasn't quite ready and started, very politely, taking other meetings. The damage is real. It's just diffuse, delayed, and lands months later in a column no one connects back to onboarding.
That's the trap with this kind of cost. If it arrived as a single visible number, you'd act on it tomorrow. Because it arrives as a slow leak (a slightly higher churn rate here, a slightly longer sales cycle there, a relationship that never warmed up) it reads as the weather rather than a decision you can change.
Take one account and watch it play out. A new hire takes their first awkward kickoff call; nothing blows up, but the client comes away a little unsure they're in safe hands. Three months later they're slower to return emails. At renewal they want a discount, or a quick chat with someone more senior, or simply to "explore options." None of that gets logged as an onboarding failure. It gets logged, if at all, as a tough renewal. The first shaky call and the cool renewal are the same event, separated by enough time that nobody files them together.
And then there's the cost you can feel
The other half isn't on the customer's side at all. It's the time your best people spend covering for the system. The senior closer pulled into a junior's call to rescue it. The team lead re-explaining the same objection-handling, person by person, hire after hire. The manager whose week quietly becomes a help desk.
Expertise spent on repetition is the most expensive training budget you have, and it's the one that never shows up as a budget. Every hour your most capable people spend shadowing and firefighting is an hour they're not doing the work only they can do. You're paying senior salaries to deliver inconsistent, ad-hoc onboarding, and calling it mentoring.
The question isn't whether new hires will make mistakes. It's who's in the room when they do.
Why it persists
Not because anyone thinks it's ideal, but because the obvious alternatives are weak. Slides and e-learning teach the theory cleanly and leave the actual hard part (doing it live, with a real person reacting) completely untouched. Shadowing is better, because at least it's real, but it's passive: watching someone handle a call is not the same as handling one. It varies wildly between managers and offices, so two hires from the same intake get two different educations. And it burns the senior time we just talked about.
None of these gives the new hire the thing they actually need, which is unglamorous and specific: reps at the real thing, before the real thing has consequences. Not more information about the job. Attempts at the job, somewhere a mistake costs nothing.
Move the mistakes off the customer
The fix isn't more content or more shadowing. It's a safe place to be bad first. If a new hire can rehearse the difficult renewal, the tense escalation, the multi-stakeholder call where everyone wants something different, and get honest feedback after each attempt, then the version the customer eventually meets is the rehearsed one, not the first draft.
That single change moves three things at once:
- Ramp gets shorter. People arrive at live accounts already past the fumbling stage, because the fumbling happened in rehearsal where it was free.
- Quality gets consistent. Everyone practises against the same scenarios to the same bar, instead of inheriting whatever their particular manager happened to model that week.
- Your seniors get their time back. The repetitive coaching happens in the rehearsal, so your experienced people are pulled in for the genuinely hard cases, not the hundredth run-through of the basics.
You also get to stop guessing
There's a quieter benefit. When new hires practise in realistic scenarios, you can finally see who's ready before they're live, instead of finding out the expensive way on a real account. That's a deeper problem than it sounds, and we pull it apart in completion isn't readiness: a finished onboarding checklist tells you someone showed up to the training, not that they can do the job.
This is the problem we built Velenta around: dropping client-facing new hires into realistic role-simulations of the conversations they're about to have, so the costly early reps happen with an AI counterpart instead of a paying customer. The mistakes still happen, because mistakes are how anyone learns this. They just stop happening on your book of business.
If your new hires are still learning on your customers, you don't have a training gap. You have a training cost. You're just paying it in client goodwill, where it's hardest to see and most expensive to recover.