Your support person spends the first two hours of every day answering questions they've answered a thousand times. What are your business hours. Where's my order. How do I reset my password. None of it is hard. All of it is necessary. And none of it is what you hired that person to do.
That's opportunity cost, and it's the most expensive kind of cost precisely because no invoice ever shows up for it. The money you spend is obvious. The value you never captured is invisible, which is why it goes unquestioned for years.
The cost that never sends a bill
Every hour has one owner. Spend it one way and you can't spend it another. When your best support person burns the morning on repetitive lookups, the bill isn't their salary for those hours, it's everything that didn't happen instead.
The follow-up call to the customer about to churn. The help article that would've prevented forty future tickets. The upsell conversation with the account that was clearly ready. Those don't show up as losses because they were never on a spreadsheet. They just quietly didn't occur, week after week.
That's what makes opportunity cost sneaky. A late shipment gets noticed and fixed. A morning of a skilled person doing work a machine could do looks like a normal Tuesday.
It also survives for years precisely because nobody chooses it. No one wakes up and decides to spend their best hours on password resets. It accretes one reasonable "I'll just handle this" at a time, until it's simply how the week runs and nobody remembers deciding it should.
Where the hours actually go
Look at a typical support week and it usually splits into two very different piles.
| Type of work | Rough share of volume | Needs a human? |
|---|---|---|
| Repeated simple questions (hours, shipping, order status, resets) | The large majority | No |
| Genuine problems, complaints, edge cases, judgment calls | A smaller slice | Yes |
The uncomfortable part is that the big pile, the repetitive one, tends to eat the most time, not because each question is hard but because there are so many of them. And it's the small pile, the one that needs real thought, that gets rushed because there's no time left.
You've got it backwards, and so does almost everyone. The work only a person can do is being squeezed by the work a person shouldn't be doing.
A small shop, two versions of the same week
Picture Maple & Co, a fictional home-goods store with one support person, Dana. In the manual version of her week, Dana handles roughly the same questions on repeat: order status, return instructions, "is this in stock," store hours. Say that's the first couple hours of every day, gone before she starts anything meaningful. By afternoon she's tired, and the tricky refund dispute that needed care gets a quick, curt reply.
Now the other version. A chatbot trained on Maple & Co's shipping, returns, and product pages handles the repeat questions the moment they come in, at any hour, including the nights and weekends Dana isn't working. Dana's morning opens up. She spends it calling three customers whose orders had issues, writing a returns guide that cuts future tickets, and giving the refund dispute the attention that keeps a good customer.
Same person, same hours, wildly different output. The difference wasn't effort. It was where the effort got spent. In SpideyChat you'd get Dana's version by pointing a bot at the pages that already answer those questions, so the repetitive wave resolves itself and the human work rises to the top.
Putting a rough number on it
You don't need a precise figure to make the case, just an honest one. Pick your busiest support person and estimate how much of their day goes to repetitive questions. Say it's the first two hours. Over a five-day week, that's ten hours, roughly a quarter of their time, spent on work a trained bot could largely absorb.
Now ask the more useful question: what's the most valuable thing that person could do with a reclaimed quarter of their week? Not "answer questions faster," but the specific high-value work sitting undone. Retention calls. A help article that prevents future tickets. Real attention on the hard cases. That undone work is the actual cost, and it's usually worth far more than the hourly wage the time represents.
Frame it as a range, not a hard claim. "We're losing something like ten hours a week of our best person's time, and here's what they'd otherwise be doing" is a sentence that gets a decision made. It's honest about being an estimate and clear about what's at stake.
What you get back
The reclaimed time doesn't vanish into "efficiency." It becomes specific, valuable things your team never had room for:
- Retention work. Actually reaching out to customers before they leave, instead of after.
- Root-cause fixes. Time to notice that forty tickets share one confusing product page, and fix the page.
- Better answers on hard tickets. The complex cases get a rested, focused person instead of a frazzled one.
- Growth conversations. Upsells and expansions that only happen when someone has bandwidth to have them.
- Fewer late nights. After-hours questions handled without anyone staying late or checking their phone at dinner.
Notice none of these are "answer simple questions faster." That's the input. These are the returns, and they're where the real money is.
The returns also compound, which the repetitive work never does. A help article written this month keeps deflecting tickets next month and the month after. A customer talked out of leaving keeps paying. Repetitive answering buys you nothing tomorrow; you just do it all again. That's the difference between spending time and investing it.
Deciding what to keep human
Automating support isn't about removing people; it's about aiming them. A quick sorting exercise tells you what to hand off and what to protect:
- List the questions your team answers most. Be honest about how many are near-identical.
- Mark the ones with a stable, correct answer that lives in your existing content. Those are automation candidates.
- Mark the ones that need judgment, empathy, or a decision. Those stay human, always.
- Automate the first group and route the second straight to your team, with context attached so nobody re-asks what's known.
The goal is a clean division of labor. Machines take the repetition. People take the problems. Both do what they're good at, and neither is stuck doing the other's job.
You don't need a study to feel this one. Watch where your team's morning goes for a week and ask a simple question of each hour: could a well-trained bot have handled that, or did it genuinely need a person? The answer to that question, repeated across the week, is the size of the opportunity you've been paying for without noticing. Reclaiming even part of it is one of the highest-return moves a small team can make.