
Every small business has one. The person who's been there longest, who somehow knows the answer to every odd customer question, who the rest of the team quietly defers to whenever something unusual comes up. It's a genuine strength, right up until that person takes two weeks of annual leave, and the business realises how much of its actual operating knowledge was never written down anywhere else.
This isn't a criticism of that person, it's usually a sign they're good at their job. The problem is structural, not personal: when knowledge exists only in someone's head, the business has a single point of failure it doesn't notice on a normal day, because normal days don't test it. It shows up the moment that person is unavailable, whether that's planned leave, sick leave, or them leaving the company entirely.
Ask yourself honestly: if your most experienced customer-facing staff member disappeared for two weeks tomorrow, how many customer questions would go unanswered, or get answered wrong, by whoever's left covering for them?
It rarely announces itself as a crisis. It looks like a colleague messaging the person on leave anyway, "sorry to bother you, quick question," because nobody else knows the answer and the customer is waiting. It looks like a customer getting a noticeably different, less confident answer than they'd have gotten from the usual person. It looks like a booking or a quote taking twice as long because someone has to double-check something the absent staff member would have known instantly.
None of these show up as a formal incident. They show up as friction, delay, and a slightly worse customer experience that nobody quite traces back to its actual cause.
Lean teams are the norm for Singapore SMEs, which is exactly what makes this risk sharper here than in a larger organisation with redundant coverage built in. A ten-person business often runs on two or three people who each individually hold knowledge nobody has formally captured. There's rarely slack in the system for a proper handover process, work just gets covered as best it can, and quality dips quietly during the gap.
It's tempting to think of this as a two-week problem that resolves itself when the person returns. But the same risk exists every time that person is sick, in a meeting, on a call, or simply focused on something else. The "single point of failure" isn't a leave-specific issue, it's a permanent fragility that leave periods just happen to expose more visibly.
There's also a harder version of this problem: what happens if that person leaves the company for good? Businesses that have never had to answer this question are often the ones most exposed to it.
Part of what makes this risk so persistent is that it's invisible on any given normal day. The person in question is usually excellent at their job precisely because they've internalised so much, product details, past customer preferences, the specific way a tricky supplier likes to be handled, the informal workarounds for problems that come up every few months. None of that gets written down because there's never an obvious moment where it feels necessary to. It only becomes visible in the gap, and by then it's already costing the business something.
Owners who've been through this once usually become fairly disciplined about documentation afterward. The trouble is that the first time it happens is also the most expensive, because nobody saw it coming and there's no fallback plan ready to go.
Most businesses do have some plan for when a key person is away, someone else picks up the phone, someone else checks the messages. But there's a meaningful gap between a role being technically covered and being covered well. A colleague filling in can usually keep the lights on: they'll answer the phone, they'll try their best. What they often can't do is answer with the same confidence, accuracy, or speed, because the knowledge genuinely isn't theirs to draw on. Customers can usually tell the difference, even if they're too polite to say so.
This matters because "technically covered" can create a false sense of security. The business owner assumes the gap is handled because someone is nominally responsible, while the actual customer experience during that period quietly degrades.
The instinct when someone identifies this risk is often to launch a formal documentation initiative: a wiki, a shared drive, a structured project with a deadline. These efforts are well-intentioned and frequently stall, because writing everything down all at once is a large, unrewarding task that's easy to deprioritise against the actual work of running the business.
A more realistic approach is smaller and more habitual: every time someone messages the absent staff member with "sorry, quick question," treat that as a signal. Write the answer down immediately afterward, in whatever tool you already use, a shared note, a simple spreadsheet, anything searchable. Do this consistently for a few months and you end up with a genuinely useful knowledge base built entirely from real, asked questions, rather than a guess at what might be useful.
The actual fix is unglamorous: write down what's in that person's head, gradually, starting with the questions asked most often. This isn't a one-time documentation sprint (those usually stall out), it's a habit of capturing an answer the moment you notice only one person knows it. Over a few months, the shape of a real knowledge base emerges, without anyone needing to block out a week for "documentation."
Once that knowledge is written down, even in rough form, it can train a chatbot to answer the repetitive share of it consistently, for customers and, in some setups, for staff covering an absence too. This doesn't replace the judgement your key person brings to genuinely complex situations, that stays with a trained human. But it means the ordinary, asked-daily questions get answered the same way whether your most experienced person is at their desk or on a beach somewhere.
For customer-facing questions, this is exactly what an AI Customer Service chatbot is built for. If the knowledge in question is more internal, product specs, SOPs, policy documents, our Internal Knowledge Base chatbot page explains what that looks like and, honestly, what still needs to be scoped carefully before you build one.
There's a second, less obvious cost to undocumented knowledge: it makes hiring and onboarding slower and riskier than it needs to be. A new staff member joining a business where critical knowledge lives only in one person's head has no real way to ramp up quickly, they're entirely dependent on that person's availability and patience to learn the job. Businesses that have documented their key answers, even imperfectly, give new hires something concrete to learn from on day one, which shortens the time before they're genuinely useful and reduces how much ongoing supervision they need.
This becomes especially relevant if you're planning to grow. A business that scales by hiring more people who each individually need months of informal mentoring to become competent scales much more slowly, and more expensively, than one where a documented base of knowledge lets new hires get productive faster.
Next time your key person is away, even for a day, keep a simple log: every question that came up where someone said "I'd normally ask [name] this." At the end of the week, look at the list. That list is your actual documentation backlog, prioritised by real frequency, not guesswork. It's also the clearest argument you'll ever get for why this is worth fixing before it becomes an emergency instead of an inconvenience.
You don't need to solve this all at once, and you shouldn't try to. A handful of written-down answers, added consistently over a few months, does more for your business's resilience than a single ambitious documentation sprint that never quite gets finished. Start with this week's list, and treat next week's leave, sick day, or unusually busy shift as another opportunity to add to it.
The businesses that handle this well aren't the ones with no key-person risk at all, that's rare for a small team. They're the ones that noticed the risk early, started writing things down before an emergency forced the issue, and kept at it steadily rather than treating it as a problem solved once and never revisited.
Book a free demo and we'll talk through what's realistic to automate, and what should stay with your team.
Book a free demo