Data quality and the GIGO principle: when not to trust the number

Thursday morning, the monthly commercial review. Esther, Hotel Peaqplus City’s marketing manager, points at a number on the report: “Our corporate (company, contracted) segment is underperforming. The average rate there is only 95 EUR — barely more than leisure. Maybe we should push a corporate promotion to fire it up.” It sounds logical: the number is low, so let’s do something about it.

Adam, the general manager, pauses for a moment. Over recent weeks he’d seen with his own eyes that the company guests take the most expensive rooms, at full rate. “Strange. My sense is that corporate is the strongest thing we have. This 95 doesn’t add up. Let’s look at what’s behind it before we do anything.”

This lesson is about the most important piece of data-driven self-defence: that a number is worth only as much as the data behind it. The leader’s job isn’t to believe everything on the report — but not to believe nothing either. The right stance is healthy scepticism: knowing when a number is reliable, when it isn’t, and what to do when it isn’t.

GIGO: garbage in, garbage out

An old, merciless saying from computing: GIGO — garbage in, garbage out. However clever the system, however polished the dashboard, if it gets bad data it gives a bad result — only in a confident, well-formatted, convincing package. The wrong number doesn’t look wrong. It looks exactly like the right one: it has its decimal place, the currency is there, the font is nice. That’s what makes it dangerous.

So it’s not enough for the leader to read the number — they have to be able to doubt it at the right moment. Not out of paranoia, but because a single bad data point can turn a whole decision the wrong way, as Esther’s corporate promotion nearly did.

The average trap

The most common and most insidious distortion is the average trap. An average compresses many different things into one number — and if the “many different things” are really two completely different groups, the average tells the truth about neither. This is exactly what happened with corporate’s 95 EUR. Let’s unpack it.

It turns out that an airline crew allotment (a pre-committed room block) had been accidentally booked into the corporate segment — a deeply discounted, static-rate block that has nothing to do with the genuine company business; it just landed in the wrong segment when it was entered. Let’s see what this does to the average on one month of Hotel Peaqplus City’s data:

The report shows 95 EUR — perfectly correct mathematically: 28,500 / 300 = 95. Except this average is about not a single real guest. The genuine company business’s average rate is not 95, but:

25,000 / 200 = 125 EUR

And the crew block is 35 EUR. Between the two there is nothing — there isn’t a single guest who paid 95. The 95 is an invented midpoint between two sets that have nothing to do with each other. If Esther launches a corporate promotion off this 95, then with a rate-cutting promotion she damages the genuinely strong, 125-euro segment — over a problem that doesn’t even exist. The fault isn’t in the corporate business but in the bookkeeping: an item assigned to the wrong segment.

The other typical distortions

The average trap is the most common, but not the only case where the number lies. Four others worth a leader recognising.

Bad segmentation. If the walk-in (a guest who arrives without a prior booking) ends up among corporate, or an OTA booking (online travel agency) counts as direct, then every segment-based decision stands on a distorted footing. The number checks out, it just doesn’t measure what we think.

Incomplete or late-entered PMS data. If bookings enter the PMS (property management system — the hotel system, here Sabeeapp) not in real time but with a lag, then today’s pace can look false: it looks as if we’re behind, when it’s only the data entry that’s late. A forecast built from such incomplete pickup data (the bookings picked up) errs confidently — the number is precise, its foundation full of holes.

Too small a sample. “The new Junior Suite package converts — three bookings came in on it!” You can’t read a trend from three bookings. On a small sample it’s easy to mistake random noise for a real pattern, and to make a lasting truth out of one lucky week.

A misleading comparison. “Revenue is +20% versus last year!” — but this time last year half the house was closed for a refurbishment. The number is true, the comparison is false. The question is always: what are we measuring against, and are the two even comparable?

Healthy scepticism — but not data denial

Here comes the lesson’s delicate balance. The danger runs both ways. One extreme is the naive believer, who accepts everything on the report and launches a promotion off the 95. The other is the cynical denier, who waves it away after the first strange number: “the system’s wrong anyway, forget it, let’s decide on gut feel.” This is just as bad — worse, in fact, because it slides back to the gut-driven way of working discussed in lesson 1, only now in the disguise of “the data’s unreliable anyway.”

Neither is the right stance. The data-driven leader doesn’t throw away the number, but has the bad data fixed. When a number doesn’t match business reality, that’s no reason to reject the number — it’s a signal that something behind the data is wrong, and that’s what has to be found. The goal isn’t distrust, but traceable trust: I trust the number because I know where it comes from, and I can check it.

A simple reflex helps you decide when to doubt: if a number contradicts what you see with your own eyes, don’t decide on it until you understand why. Esther looked at the 95; Adam looked at the reality he’d lived. The discrepancy didn’t mean one of them was wrong — it meant there was a loose thread in the data. In such a case the good leader doesn’t choose between the number and the gut, but looks into it.

Back to the commercial review

Adam doesn’t sweep Esther’s suggestion aside, nor does he accept it. He calls in Daniel, the revenue manager: “Let’s look at what’s sitting in the corporate segment.” Five minutes, and there’s the crew block that landed in the wrong place. They move it out into its own segment — and corporate’s true average rate jumps to where it belongs: 125 EUR.

This didn’t make the corporate business better by a single cent; it was 125 EUR all along, the report just lied about it. But the decision changed radically. Esther doesn’t launch a needless — indeed harmful — corporate promotion on a non-existent problem. In fact, she now sees that corporate is one of the house’s strongest segments, and that the thinking should run exactly the other way: not to push it down but to defend and grow it. One fixed data point turned a bad decision into a good one — and it didn’t kill the faith in data but strengthened it, because now they know that the number, when clean, tells the truth.

This is the GIGO principle’s lesson for leaders. A bad decision is born not from whether we have data or not — but from whether we trust a number without knowing what’s behind it. The good leader doesn’t believe blindly, nor deny blindly: they ask, they look into it, and they have the bad data fixed — they don’t throw away the good decision over it.

The daily, practical discipline of clean segmentation and data hygiene is shown from the revenue manager Daniel’s perspective in the RM Academy lessons Segments and markets and Segmentation in depth.

Key takeaways

• GIGO — garbage in, garbage out: bad data becomes a bad number, only in a confident, well-formatted package. The wrong number doesn’t look wrong — that’s what makes it dangerous.
• The average trap is the most common distortion: behind a 95 EUR “corporate average” there can stand a strong, 125-euro genuine business and a wrongly booked 35-euro crew block — and not a single real guest paid the 95.
• Recognise the typical distorters: bad segmentation, late-entered PMS data, too small a sample, a misleading comparison. A number can be mathematically correct and business-false at the same time.
• The right stance is healthy scepticism, not data denial. The cynical “the system’s wrong anyway” takes you back to the gut-driven way of working.
• Reflex: if the number contradicts what you see, look into it before you decide. The good leader doesn’t throw away the number but has the bad data fixed.