The Human Layer Smarter Pricing Software Still Needs

1. You spent over four years in leveraged finance at Deutsche Bank before moving into hospitality revenue, first at Wynwood House and now at PriceAdvize. What from structuring and analysing debt deals actually carried over into pricing vacation rental listings, and what turned out to be completely useless outside banking?

More than I expected, actually. The biggest thing that carried over was the way you approach a problem. In leveraged finance, you’re constantly looking at a business through numbers, trying to understand what really drives performance, where the downside is, and which assumptions matter most. Revenue management is surprisingly similar. You have a huge amount of data, but the challenge isn’t accessing it – it’s figuring out which variables actually matter and translating them into a decision.

The other thing banking taught me was to think in terms of opportunity cost. Every night that goes unsold in hospitality is perishable inventory. You can’t sell yesterday’s room tomorrow. So every pricing decision is essentially a risk/reward decision: how much revenue am I willing to potentially leave on the table in exchange for a higher probability of occupancy?

What was completely useless? Probably 80% of the technical finance knowledge. I haven’t built a debt waterfall or calculated an LBO return since leaving banking. And hospitality is much more operational and behavioural than finance. You can build a beautiful model and still be completely wrong because a competitor changed strategy, an event moved dates, or guests simply behaved differently than you expected. That forces you to be much more comfortable with imperfect information.

2. PriceAdvize’s model is built around a guarantee: increase revenue within 90 days or the operator doesn’t pay. With a finance background, how do you actually price that risk internally, and has there been a portfolio where the numbers didn’t move the way you expected?

We don’t really price the guarantee like an insurer would price risk. We manage it through selection.

Before taking on a portfolio, we try to establish whether there is actually enough revenue leakage for us to create value. We look at pricing configuration, occupancy and ADR patterns, booking windows, seasonality, market positioning and, importantly, how much control we’re actually going to have. If a portfolio is already extremely well managed, or if the operator isn’t willing to implement the changes we recommend, then offering a performance guarantee doesn’t make much sense.

The 90-day period is also deliberate. Revenue management is noisy over short periods. A week or even a month can be distorted by events, cancellations, seasonality or just normal variance. Over roughly 90 days, you have enough booking cycles to start separating the impact of your decisions from noise.

And yes, we’ve had portfolios where performance didn’t move as quickly as we expected. That’s inevitable. Sometimes the issue isn’t pricing at all – it might be the product, reviews, listing quality, distribution or simply weak underlying demand. That’s actually an important discipline for us. If we can’t demonstrate that we’ve created value, we shouldn’t expect the client to pay for a promise we didn’t deliver.

3. Dynamic pricing tools like PriceLabs and Beyond keep shipping more sophisticated AI-driven features, positioning themselves as increasingly self-sufficient. As a fractional revenue partner selling hands-on human expertise on top of exactly these tools, where do you see the real white space left for what you do, as the software itself gets smarter?

I actually think better pricing software makes our opportunity bigger, not smaller.

The misconception is that revenue management is primarily about calculating the right price for a particular night. Software is already very good at that, and it will continue getting better. The harder problem is everything around the algorithm: how you configure it, how you segment a portfolio, what strategy you want it to execute, when you should override it, and how you diagnose why a portfolio is underperforming.

PriceLabs can process far more market data than a human revenue manager ever could. But give the same PriceLabs account to ten different operators and you’ll get ten very different outcomes. The difference is increasingly not access to technology; it’s how intelligently that technology is configured and managed.

That’s also where we’re evolving PriceAdvize. We’re building our own AI layer on top of tools like PriceLabs rather than trying to compete with their pricing engine. The idea is to continuously analyse performance and configuration at the portfolio, group and listing level, identify revenue gaps, and tell the revenue manager where intervention is actually required.

So I think the role changes. The revenue manager becomes less of a person manually changing prices and more of someone designing, supervising and auditing an increasingly autonomous system. That’s the white space we’re interested in.

4. Before PriceAdvize, you ran revenue for a single boutique hospitality brand at Wynwood House in Mexico City. What pattern did you see across that one portfolio that convinced you the same approach could scale as a service across dozens of different operators, rather than staying a specialty tied to one brand?

At Wynwood House, we grew to more than 1,000 listings across Latin America, and what struck me was how repetitive the underlying revenue problems became.

The properties were different, the markets were different and the guest profiles could be different, but the same mistakes kept appearing: pricing rules that hadn’t been revisited in months, minimum stays that were suppressing demand, gaps between weekday and weekend strategy, poor handling of orphan nights, discounts stacking in ways nobody intended, or portfolios being managed with essentially the same strategy despite very different booking patterns.

That was the insight behind PriceAdvize. The value wasn’t that I had developed some unique intuition for one particular brand or one particular market. We had developed a process for diagnosing revenue leakage and deciding what to do about it.

Once you realise that, the question becomes whether you can systematise that process enough to apply it across operators without losing the contextual judgement that makes revenue management work. That’s essentially what we’ve spent the last few years doing.

And now AI makes that much more interesting, because a lot of the analytical work that historically limited how many listings a good revenue manager could properly supervise can increasingly be automated. That potentially lets us scale the methodology much further without turning revenue management into a generic, one-size-fits-all service.