"We can manage by hand a bit longer."
I hear this sentence in almost every first conversation with leaders of e-commerce brands and performance agencies. It always sounds reasonable. And it's always said at the same moment: right after someone has calculated what automation costs, and before anyone has calculated what the lack of it costs.
The short answer: the cost of manual work in a company doesn't grow linearly. It grows like compound interest. Every month of "a bit longer" adds new channels, new files and new dependencies on a single person. The bill doesn't arrive as an invoice. It arrives the day a report is wrong and the budget is already spent.
This article shows where that accumulation comes from, how to tell you're already paying, and how to get out of the trap without a revolution.
Why This Sentence Sounds So Reasonable
Because today, it's true. With 10-20 sales and marketing channels, someone really can pull data from ads, the store and payments on Friday, paste it into a spreadsheet and prepare a report for Monday. It works. The board gets its numbers. Nobody complains.
But that situation has a hidden side. Because it works, nobody sees the cost. There's no "Excel stitching" line in the budget. There's no invoice for the analyst's Friday afternoons. What isn't visible in the books stops being a topic for the board, and what isn't a topic for the board is never urgent.
From my conversations, in D2C companies and performance agencies, stitching data together consumes an estimated one to two days a week of one person's time, sometimes of several people. These are observed values, not audit results. But even at the cautious end of that range, we're talking about 20-40% of the time of people you hired to think, not to copy and paste.
The Cost of Manual Work That Never Shows Up in the Budget
How much does manual data work cost? Three things at once, and only the first is visible.
- People's time. Hours of analysts and marketers spent retyping instead of analysing. You only see it if someone counts it, and rarely does anyone count.
- Delayed decisions. If this week's report is built next week, every ad budget decision is made on data that is several days old. When monthly spend runs into the hundreds of thousands, that's not a detail.
- Risk of error. Every manual copy is a chance for a mistake: a shifted column, a wrong filter, a stale file. A manual process has no alarm. An error doesn't announce itself. It waits.
I call this the EBITDA Leak: a silent margin leak with no line of its own in the budget, so nobody owns it. You won't find it on an invoice. You can only find it by asking people what they do on Friday afternoons.
The Compound Interest of Manual Work
Here is the core of why this sentence is so expensive.
Manual work doesn't cost the same amount every month. With every step of company growth, its cost rises and stacks on top of the previous one:
- a new ad channel arrives, so there's a new file to stitch,
- a new market arrives, so there's a new currency and new rules,
- a new hire arrives and gets their own copy of the spreadsheet "because it's easier for them,"
- a new board report arrives that someone can no longer verify.
After a year, you don't have "a bit more manual work." You have a web of dependencies in which nobody knows which file is the true one. That is Data Debt: the sum of shortcuts, workarounds and manual patches that once saved time and now generate interest. Like any debt, the longer you delay repaying it, the more it costs.
The most dangerous part is that the interest is invisible. The company grows, revenue grows, so the problem seems to vanish into the background. Only when growth slows does it turn out that margin isn't growing with revenue. A large share of the increase went into servicing your own chaos.
Manual work is not a cheap temporary fix. It's a loan with no repayment schedule.
The Day the Bill Arrives
In practice, I see two scenarios in which "a bit longer" ends. It's almost always the same two.
Scenario one: a report error burns the budget. Someone decides to scale a campaign based on numbers that turn out to be wrong. A few days later it emerges that the channel that got the money wasn't profitable, and the one that was switched off was. Nobody is personally to blame, because the system simply had no safeguards.
Scenario two: the person who "knows how it works" leaves. An analyst who has manually stitched data together for two years gets an offer. Overnight, the company loses not only an employee but also the only user manual for its own numbers. That knowledge lived in their head and in files nobody else understands.
When do companies usually start looking for a solution? Exactly then, after one of these events. It's always the most expensive moment to decide, because the decision is made under pressure, in a hurry, with the bill already issued. Those who start earlier pay less. Not because they're smarter, but because nobody has pushed them yet.
Why Nobody Changes It (Even Though Everyone Knows)
From my conversations, almost nobody defends manual work as a good solution. Everyone knows it's a problem. The reasons nothing happens are different:
- Nobody owns the problem. The analyst won't build their own automation because they have reports to deliver. The board doesn't see the cost, so it doesn't make it a priority. The problem belongs to everyone, which means to no one.
- Change looks like a big project. In decision-makers' heads, automation means six months, a large budget and risk. Manual work is "nothing." The comparison is unfair, because one side has a visible cost and the other a hidden one.
- The pain is spread over time. Nobody pays a lot at once. You pay a little, every week, quietly. People react to shocks, not to erosion.
When does data automation pay off? When the cost of one day of manual work per month becomes comparable to the fee for a solution that eliminates it. In practice, for companies spending tens of thousands a month on advertising, that threshold is crossed much earlier than boards assume. I won't give you a single number here, because it depends on the company. But the number exists, and it can be calculated in an hour.
Three Questions That Take Fifteen Minutes
You don't need an audit to check whether you're already paying. Ask yourself these three questions:
- How many hours a week does your team spend collecting and retyping data instead of analysing it? Ask them directly rather than guessing. The answer usually surprises both sides.
- What happens if the person who assembles your reports leaves tomorrow? If the answer is "there will be a problem," you have a risk that appears in no register.
- How old are the numbers behind your latest budget decision? If the data is several days old and spending runs into the hundreds of thousands, you pay for that delayed picture every month.
If any of the answers worries you, you've just found your EBITDA Leak.
What I Do Differently
My philosophy is simple, and I wrote it into the DataMinq manifesto: code is a cost, a solution is an asset. It's not about building something complicated. It's about making the data your people stitch together today flow on its own into one place, a Single Source of Truth: one shared source from which everyone in the company takes the same numbers.
When that works, something changes that you can't measure in a spreadsheet. The team stops firefighting and returns to analysis. The board stops asking "are these numbers really right?" because it has them daily, effortlessly. Margin grows not because someone cut costs, but because people do what they were hired to do. I call it peace of mind: the data simply works.
You don't have to start with everything. One process is enough, the one that eats the most Friday afternoons. An automated data flow, what in our vocabulary is called a Digital Worker: a digital employee who does what a person does, without mistakes and without holidays. The rest follows, at a pace the company can handle.
The Most Expensive Sentence Has a Cheap Alternative
"We can manage by hand a bit longer" is not a bad decision. It's a decision not taken that looks like a decision. And like every unmade choice, it has a price, only nobody issues the invoice.
If something in this article rings true, you don't need to implement anything right away. You just need to calculate. And if you'd rather calculate with someone who has been doing this for years, I invite you to a 15-minute call. No presentation, no obligations. We'll talk about where your margin most likely leaks and whether it's worth dealing with now or later.