Numbers don’t lie, but they never tell the whole truth – what a dating app and the hospitality industry have in common

Imagine a recruitment software provider advertising with the following claim:

“With our AI tool, your hiring quality will increase by 40%.”

Sounds convincing. But: 40% of what? Measured how? Over what period? And what exactly does ‘hiring quality’ mean in a company that currently has no structured onboarding process, whose leadership culture is unevaluated, and whose staff turnover has been over 70% for years?

This is precisely where the train of thought I’d like to share with you today begins – and it doesn’t start in the hospitality industry, but in a lecture theatre.

A lecture excerpt that opens your eyes

In an excerpt from a lecture by cognitive scientist and risk researcher Prof. Gerd Gigerenzer, he uses a seemingly simple example to demonstrate how statistics can deceive us without lying.

The example: The dating app Parship advertises that a single person falls in love via its platform every eleven minutes. Gigerenzer does the maths: With a million users and an assumed match rate of 5% per year, this results in a chance of around 5% within twelve months – meaning you would have to pay for around ten years to have a statistically 50:50 chance of finding your ideal partner.

The figure ‘every eleven minutes’ is not wrong. It is simply cleverly chosen. It creates a feeling, not information.

Gigerenzer’s central argument is this: in a world full of data, algorithms and AI, the crucial human skill is not trust in numbers – but the ability to interpret them. And in uncertain, complex, emotional situations, we humans are often superior to algorithms – because we understand context that no model can capture.

(Source: Lecture excerpt “Humans are superior to AI” by Prof. Gerd Gigerenzer, available on Facebook: https://www.facebook.com/share/v/1NPvkRZH42/?mibextid=wwXIfr)

What does this have to do with recruitment in the hospitality industry?

A great deal. More than it seems at first glance.

In the first three articles in this series, I described how automated systems filter out qualified candidates before a human has even seen them (#1), why keyword-based recruitment in the hospitality industry fails structurally and what makes competency-based recruitment better (#2), and which metrics really matter in recruitment – and how to interpret them correctly (#3).

The common thread running through these three posts was one and the same problem: figures and algorithms are used as if they reflect reality. Yet they always represent only a snapshot of reality – filtered through assumptions, configurations and perspectives that are rarely disclosed.

And this is precisely what Gigerenzer warns against.

Three figures we believe too often in the hospitality industry

“Our time-to-hire has halved.”

Sounds like a success. But as described in the third part of this series: a shorter time-to-hire can be a sign of an efficient process – or it can mean that candidates were hired under time pressure and without sufficient vetting. If you don’t also measure the retention rate after 90 days and performance after six months, you’re only getting half the picture.

“Our ATS system filtered 300 applications – only 12 made it through.”

That could mean efficiency. But it can also mean that 288 candidates were rejected because they wrote “house management” instead of “hotel manager” – or because they came from a different career background and were unable to translate their hands-on skills into ATS-compliant keywords. In the second article in this series, I demonstrated that the system does not assess suitability. It assesses wording.

“Our offer acceptance rate is 80%.”

Fine – but who accepted those 80%? And who turned them down? What do those who declined have in common? Why did they ultimately decide otherwise? As outlined in Article #3: exit interviews with candidates who have turned down an offer often provide valuable insights that no single metric can deliver on its own.

The real problem: we trust algorithms more than our own judgement
In his lecture, Gigerenzer describes a paradoxical trend: the more data we have, the more we rely on it – and the less we trust our own judgement. Yet in certain situations, we humans are clearly superior to algorithms:

• in unstructured, context-rich situations
• when making emotional and interpersonal judgements
• when experiential and implicit knowledge are required
• when gut instinct is actually distilled experience

Does that sound familiar? That is precisely the domain of the hospitality industry.

The skills that define an excellent service staff member, that keep a management team together during the peak season, that make a guest want to come back time and again – they arise from lived experience, from situational judgement, from human sensitivity. No ATS system in the world can measure them. No metrics dashboard can capture them.

What does this mean in concrete terms – for you and your business?

It’s not about rejecting AI or metrics. It’s about putting them into context.

Gigerenzer calls this statistical thinking – the ability to place a number in its context, understand connections and form your own judgement. Not against the data, but with it.

For recruitment in the hospitality industry, this means:

• No metric can replace a face-to-face conversation. Potential, personality and cultural fit don’t emerge on paper – they become apparent through interaction.
• No ATS system is worthless – but no ATS system is all-knowing. It filters based on what you teach it. If you don’t configure it based on competencies, you create blind spots. 
• Do not view any metric in isolation. A falling time-to-hire, a rising offer acceptance rate, a good source of hire – they only tell a story when read together.
• AI supports – but does not decide. The second wave of digitalisation opens up possibilities. It does not replace human judgement, which is fuelled by experience, context and empathy.

From practice: When processes scare off talent – a personal example

Theory is one thing. But sometimes your own experience provides the clearest picture.

I applied for a total of nine positions at a private university – some freelance, some permanent: proofreading, editing for various departments, lecturing on topics that I not only know in theory, but also strategically plan, implement, support and guide on a daily basis in multiple roles. I was qualified for every single one – not just on paper, but proven through real-world experience.

Nine applications. Nine rejections.

I hadn’t just applied via the website’s application portal; I’d also sent my documents directly to the recruitment team by email – knowing full well that direct contact can sometimes make all the difference, as I described myself in post #1 of this series. The reply came straight away: they’d deleted everything. I was asked to apply exclusively via the website. Data protection.

I did so. The rejections followed.

Last week, members of that very same recruitment team got in touch with me – joined by external recruitment partners who had since been brought in because the positions had remained unfilled for months. Would I be interested in working with them?

No. I’m not anymore.

And this is not a question of wounded pride or stubbornness. It is a question of context – very much in the spirit of Gigerenzer. Anyone in recruitment who is unable to question their own processes demonstrates that they lack structural self-reflection. And experience shows that those who lack structural self-reflection do not do so in other areas either. That is a piece of information. And I place it in context.

What happened to me is something that freelance lecturers and colleagues in the industry describe time and again in similar terms. And time and again in relation to the same private university. The pattern is familiar: you apply, you’re rejected, and months later you’re contacted – not because you’ve suddenly improved, but because there are no other options left. Or because your LinkedIn profile has been found, and you’re clearly identified as an expert in this or that field. By people. Not by AI.

This reminds me of Gigerenzer’s Parship example – and of a feeling familiar to many who use dating apps: you show interest, get rejected, and are then expected to fill the gap when no one better has been found. Anyone with self-respect is too good for this approach. This applies just as much to people looking for a partner as it does to professionals in the job market. And it doesn’t even have to come to rejection; ghosting is quite enough – in both personal and professional contexts.

The real message here is not a personal one. It is systemic: recruitment processes that first weed out talent and then try to win them back when the company’s own pipeline is empty are not a sign of efficiency. They are a sign that the company is not interpreting its own figures correctly – and is failing to see the people behind them. They are a sign that AI has been fed the wrong parameters and that, for whatever reason, one relies exclusively on this very AI rather than on common sense.

The difference between information and insight

Parship matches a single person every eleven minutes. That’s true. And yet it says almost nothing about one’s own chances.

Similarly: an ATS system narrows 300 applications down to 12. That’s true. And yet it still says almost nothing about whether the 12 remaining were the best – or simply the best-written.

Information and insight are not the same thing. The difference lies in how we interpret them. And it is precisely that – interpreting, questioning, judging – that makes us, as humans, irreplaceable.

In the hospitality industry, which has always thrived on genuine encounters, this should actually be familiar to us.

💬 My question to you: Have you had similar experiences – as a candidate, as a freelance lecturer, as an industry expert? Or are you familiar with the reverse phenomenon: talents you recognised too late? I look forward to hearing your perspective in the comments.