The Hidden Cost of Asking the Person Next to You

Picture a Tuesday morning on a busy operations floor. Someone gets a customer email about a fee they don’t recognise. They turn to the colleague next to them. “Quick question, do you know what this charge is?” The colleague isn’t sure either, so they ping the team lead in Teams. The team lead replies thirty minutes later with their best guess. The original person sends an answer to the customer based on what the team lead said.

This happens dozens of times that morning, across the floor. Nobody counts. Nobody invoices for it. It looks free.

It isn’t.

Most of my career has been inside regulated financial services. Six of those years running operations in a contact centre, watching this exact pattern play out daily. The single biggest piece of operational waste I saw wasn’t bad systems or wrong staffing models. It was the cost of people asking each other instead of asking a system that had the answer.

The pattern that looks free

When somebody shoulder-taps the colleague next to them, no line item appears anywhere. No timesheet. No alert. No invoice. The work happens off the books.

That’s why the cost is invisible. It doesn’t show up in your operations dashboards. It doesn’t show up in your reporting. It doesn’t show up in any annual review of operational efficiency. It just happens, all day, across every team that handles knowledge work.

But it IS time. On both sides. The asker loses thirty seconds writing the question. The answerer loses anywhere from two to twenty minutes giving the answer, depending on whether they know it cold or have to check. Then the asker waits, context-switches, comes back. The answerer drops what they were doing. The thirty-second question costs four minutes of focus across two people.

Even at ten questions per person per day, that’s over seven thousand hours of off-the-books work a year across just fifty people. None of it shows up on a dashboard.

What it costs operationally

Two people interrupted. Focus broken on both sides. Context-switching tax on whatever the answerer was working on before. The asker’s flow is also gone, even if they don’t notice.

Beyond the time, the answer itself has a quality problem. Whoever you ask gives you their understanding of the policy, the procedure, the fee, the rule. It’s their interpretation, not the authoritative version. Even if the colleague is correct, the answer carries no record. Nobody can audit it later. If they were wrong, you find out when the customer complains, three weeks later.

And here’s the other thing nobody mentions. The colleague being asked is rarely the right person. They’re just the closest person. They might be three months into the job. They might be twenty years in and out of date on the latest policy update. Either way, they’re not the policy owner. They’re a guess wearing a name badge.

What it costs in quality

Same question, asked five times this week, by five different people, to five different colleagues. Five slightly different answers go out the door.

Customers compare notes on social media. Or escalate to a manager who has to work out which version was right.

This is how knowledge quietly diverges across a team. Not through training failures. Through a thousand quick questions that each got a slightly imperfect answer, until the team carries five competing mental models of how the business actually works.

What it costs in regulated work

I’ve written about this before in more depth, so I won’t repeat it. The short version: in regulated industries, an unattributed answer is a problem regardless of who delivered it. If a customer was given the wrong fee structure because a colleague guessed, that’s the same regulatory issue as if an AI bot guessed. Somebody has to explain how the answer got produced. “A colleague told me at the coffee machine” doesn’t pass.

The shoulder tap is the friendliest version of inferred truth. It’s also auditable nowhere.

What teams usually try

Most teams have already tried to fix this. There’s a Confluence space somewhere. Or a SharePoint site. Or an internal wiki nobody updates. The intent was good. Centralised knowledge, searchable, available to everyone.

The reason it didn’t work is usually one of three things.

The content is stale. Nobody owns it. Articles from 2022 sit next to articles from last week with no signal which is current.

The content is scattered. Three intranet platforms, two different wikis, four shared drives. Nobody knows which one has the latest version of anything.

The search is bad. Even when the right answer is in there, finding it takes longer than asking the person next to you.

So the wiki gets ignored. People go back to asking. The shoulder tap is faster. The wiki dies a quiet death.

The actual fix

The fix isn’t another wiki. It’s a substrate the asker can self-serve from.

One approved version of each answer. An owner. A review date. Search that actually finds what’s in there. AI that can answer questions grounded in the article, not in inferred guesswork.

When that exists, the asker types the question, gets the answer, moves on. The colleague never gets interrupted. The customer gets the same answer your most experienced staff member would give. The audit trail is in place.

The shoulder tap stops being the first move. It becomes the last resort, for genuinely novel cases that aren’t in the substrate yet. Which is what a shoulder tap is actually for.

The honest admission

You can’t kill the shoulder tap entirely. People will always check with the colleague they trust. Humans like asking humans. That’s fine and it’s not the goal.

The goal is to make eighty percent of the questions answer themselves before the asker even thinks about turning to their neighbour. Make the canonical answer faster to find than the colleague’s desk. Most people, most of the time, choose the faster path. You don’t need to mandate it.

The remaining twenty percent, the questions that genuinely need a human, then get the focus they deserve. The shoulder tap isn’t gone. It’s earned.

Where this is going

Tribal knowledge sounds romantic. It’s the comforting story businesses tell about how they actually run. “We just ask each other.” “The team knows.”

In practice it’s a tax. An invisible drag on every team’s productivity. A quiet generator of quality risk. A nightmare for anyone trying to onboard the next hire.

The companies that figure this out aren’t the ones with the biggest training programmes. They’re the ones that quietly built the substrate underneath, made the canonical answer easier to find than the human, and watched the shoulder taps fall to the questions that actually needed a person.

Make the answer easier to find than the colleague’s desk. The rest follows.

The KnowledgeScout Team