On Monday morning, a new engineering manager messages Renata in Slack to ask whether unused PTO carries over and when it resets. Renata runs people operations at a 220-person company. The handbook page says up to 40 hours carry over. It does not say that anything carried over has to be used by March 31, because that rule went out in an email to managers a year ago. The information, otherwise, lives only in her head. She types the full answer. By Friday she will have typed it four more times, each time a little differently.
Experts get interrupted for the same questions because the part of the answer that matters most lives only in their heads, which makes asking them the most reliable way to get it. Each answer then disappears into a private message or public channel, so the next person has nothing to find and asks again.
The usual advice treats this as a boundary problem: set office hours, send a link, say no more often. Those fixes address the interruptions but leave the cause alone.
Monday to Friday: Five Askers, One Question
Monday's answer is the complete one: the 40-hour cap and the March 31 deadline. On Tuesday, an engineer planning a December trip gets two sentences, the cap and "use it early in the year." On Wednesday, a finance contractor pings her in the middle of a payroll run, the one task where a mistake reaches everyone's paycheck. She answers fully, then rechecks the last three steps of the payroll batch because she cannot be sure where she stopped. On Thursday, a sales rep who has already read the handbook page gets in response the cap and nothing else. On Friday, a team lead asks the same question, receives the same short version, and tells a new hire in a 1:1 that hours roll over "through spring." In April, the new hire loses 12 hours and comes to Renata.
Renata did nothing wrong at any point. Each answer was correct for the person asking at that moment. She updates the handbook twice a year, but a page cannot hold every deadline and exception, and the March 31 rule, which fell between an email and a page, was between updates. Five answers in five direct messages are five versions of one policy, and only the first was complete. Some weeks the same pings go unanswered instead, which is the other half of why Slack messages go unanswered.
Why Colleagues Keep Asking
None of the five askers was careless. The handbook page gave them part of the answer and no way to know what was missing, a Slack search returned old threads with conflicting numbers, and Renata always answered in minutes. Asking the expert was the lowest-risk option, and a few experiences like that teach a team to skip the page, which is one of the key reasons why nobody uses documentation. None of them knew the question had been asked four other times that week alone, because every earlier answer sat in a message visible briefly and then lost. The expert reluctantly and involuntarily becomes the search engine for the rest of the employees.
What Each Interruption Costs the Expert
The first cost is the drift that nobody tracks. A policy communicated as five slightly different messages produces the April problem: a new hire who loses 12 hours, a conversation Renata would rather not have, and a correction that takes longer than what all the five answers took. The usual label for this is a subject matter expert bottleneck, which only describes where the work piles up. However, it does not describe where the answer is.
The second cost is time. UC Irvine research finds that on average it takes 23 minutes to regain full focus after an interruption. Five conversations of about ten minutes each once follow-up questions are counted, plus 23 minutes of refocus per interruption, comes to close to three hours of Renata's week. No timesheet records it; hence the cost goes unnoticed.
The third cost sits in Wednesday's ping. Michigan State University research found that practice made people faster at a procedural task, and that after interruptions they resumed less accurately, picking up at the wrong step more often. Payroll is a procedural task with many steps, each depending on the last. So, mistakes are more likely to creep in after interruptions.
What Changes When the First Answer Is Saved
Run the same week again with one change. On Monday, the new engineering manager saves Renata's complete answer in three clicks, because it solved their own problem and they want it findable.
Pravodha keeps that thread attributed to Renata, searchable by topic, and marked as useful by the colleagues who found it valuable. On Tuesday, before sending a message, the engineer searches "PTO carryover" and finds Monday's answer, with the cap, the deadline, and Renata's name. The search works because the answer is recent and written by the person who owns the policy, which is what the handbook page could not offer. The finance contractor and the sales rep find it the same way, Renata's payroll run goes uninterrupted, and on Friday the team lead quotes the March 31 deadline to the new hire.
This works when someone saves the first answer, and a question nobody has asked before still goes to Renata, which is the right use of her time. A saved answer also ages with the policy: when the carryover rule changes, the old answer is wrong until its owner corrects it, as with any source. The aim is that the second, their, or fifth person asking a question finds the answer before interrupting the expert.
What to Do Before Next Week
An expert can reduce repeat questions without waiting for new tooling.
- Keep a tally for one week. Mark every question you have answered before; a handful of questions often account for most of the repeats.
- Answer in a channel, not a direct message. A reply in a shared channel can be found by the next person, but a direct message cannot.
- Ask the person who asked the question to post the answer where their team will see it. They have the most reason to do so, because it solved their problem.
- Treat office hours as a supplement. They protect focus time, but a question that waits until Thursday still gets answered in a message nobody else can find.
Documentation is the wrong ask for an expert in Renata's position; we’ve discussed this at length in our earlier blog post: why experienced employees skip documentation. Capture at the moment of answering requires far less effort, as covered in how to capture knowledge from Slack without adding work for experts.
Renata's week had five askers and one question. The ideal week worth aiming for has one answer and four people who found it before they asked.
Pravodha is built around the capture model. It integrates with Slack, Microsoft Teams, and your email to preserve the institutional knowledge your team generates every day: attributing it to the people who created it, surfacing peer validation, and making it permanently searchable without adding any burden on the experts.