An analyst hands in their resignation on Monday. Friday marks the end of the month. And at the company, nobody — literally nobody — knows how to generate the sales report that the next quarter's marketing budget decision depends on.
This isn't a hypothetical scenario. It's routine in e-commerce companies and performance agencies growing faster than their data foundations can keep up. And the thesis of this piece is simple: when one person leaving can paralyze a company, the person was never the problem. The problem is that the company never had a system — it had a person pretending to be one.
The myth most leadership teams believe
Most founders and marketing directors think about this problem the same way: "we got unlucky, we had a great analyst, now we just need to find another one just as good." It's a comfortable explanation, because it requires changing nothing except a job posting.
The trouble is, that explanation is false. A great analyst isn't the risk. The risk is a situation where the entire logic of reporting — which sources to combine, how to calculate contribution margin, which campaigns to exclude from shared costs — exists only in one person's head and in files named "report_FINAL_v3_final.xlsx."
This phenomenon has a name in engineering: the Bus Factor. It's the number of people who would need to disappear from a company for critical operational knowledge to disappear with them. If that number is one, the company doesn't have a data system. It has a hostage situation.
Why this happens specifically in e-commerce and performance marketing
In companies at the 10–70 person stage, with marketing budgets in the hundreds of thousands of PLN per month, data grows faster than processes do. Channels multiply — Meta, Google, marketplaces, CRM, warehouse systems — and each one has its own export format and its own definition of "conversion."
In that environment, the natural response is to hire one capable person to "handle the Excel side of things." And they genuinely do handle it. They build an increasingly complex spreadsheet that stitches these sources together manually or through semi-automated macros. The company gains convenience. It loses visibility.
Nobody documents the logic, because nobody has time for it — the pressure is on results, not process. A year later, that spreadsheet is the only place where the truth about campaign profitability actually lives. And only one person can read it.
Data Debt is the accumulated cost of makeshift data management shortcuts — it builds up quietly until, at one single moment, it has to be repaid in full, usually at the worst possible time.
What this actually costs — let's be honest about it
I'm deliberately avoiding hard numbers here, because every company operates at a different scale. But the cost mechanism repeats itself, and it's worth breaking down:
- The cost of decision paralysis. If leadership can't see channel profitability for a week or two, the marketing budget doesn't stop flowing — it just flows without steering. That's the most expensive kind of inactivity: money spent with no feedback loop.
- The cost of rebuilding knowledge. A new person — internal or from an agency — needs weeks to reconstruct their predecessor's logic, if they can fully reconstruct it at all. Some assumptions are simply lost the moment that person walks out the door.
- The cost of an error that isn't visible right away. Reconstructed logic is rarely identical to the original. A few percentage points' difference in how contribution margin is calculated can lead leadership through an entire quarter of decisions to scale campaigns that are actually losing money.
This isn't an IT risk. It's an operational risk with a direct line to EBITDA — because marketing budget allocation decisions are being made based on data that nobody fully understands anymore.
Can this be predicted before someone hands in their notice?
Yes — and this is exactly the moment to ask the question directly, instead of waiting for the crisis to hit.
How do you recognize that a company has a Bus Factor problem in its data? Three warning signs show up almost every time: (1) only one person can explain where the numbers in a key management report actually come from; (2) the logic for combining data from different sources lives in files, not in a documented, repeatable process; (3) the phrase "that person is on vacation" quietly triggers unease about reporting across the team.
If any of these sound familiar, it's not a question of "if" — it's a question of "when" the company will pay for it, in a specific amount of money and a specific amount of downtime.
What to do differently — shifting responsibility from a person to a system
The solution isn't "hire a second analyst just in case" — that just spreads the same risk across two people instead of one. The solution is moving critical logic out of a person's head and into a system that survives every staff change.
In practice, that means three things:
- One source of truth (Single Source of Truth). All channels flow automatically into one central place, with clearly documented logic for calculating key metrics. Not in someone's head. Not in a file. In a system.
- Automated connections between data sources. What today is manual export-stitching should happen on its own, in the background, without human involvement — and therefore without the risk of disappearing along with whoever built it.
- Documentation as a standard, not a luxury. Business logic — how margin is calculated, what's excluded from shared costs — has to be written down in a way any successor can understand, not just the person who built it.
This isn't a year-long project. It's usually a few to a dozen or so weeks of work on one specific reporting area — the one that today rests entirely on one person. The result: leadership stops depending on one person's vacation calendar, and budget decisions stop being hostage to someone's memory.