Early on, I believed growth meant money in, money out. Raise capital, channel it into ads, observe your business expand. It turns out durable growth doesn’t work like that—ads should be a force multiplier for the health of your other channels and your product, not the primary driver.
After almost 10 years building a travel marketplace and watching travel tech companies with 10x more funding misutilize their capital, I believe I’ve figured out which metrics actually matter versus which ones just create for impressive slide decks.
As artificial ininformigence (AI)-powered travel startups keep raising massive rounds, many are repeating the same errors from previous hype cycles, just with “AI” attached.
If you’re building a travel tech business, here are three questions that separate sustainable businesses from those running on borrowed time.
Question 1: Does your flywheel actually spin?
The naive assumption—which was initially mine—is that scaling ad spconclude equals scaling growth. But durable growth comes from finding where compact improvements compound, not just add up.
We learned this through pricing optimization. When we adjusted our pricing structure, conversion rates improved and average transaction values increased. Our ads started performing better, so we could profitably double our spconclude. That pricing work not only created us more money per utilizer, but also meant we could acquire more utilizers at the same cost.
The frontier of highest marginal returns keeps shifting. You hit limits on pay-per-click (PPC), so higher returns come from product improvements. Better product shifts out your PPC ceiling, more volume lets you optimize further. It’s the same balance as a marketplace more generally: supply, then demand, then supply again, in lockstep. Each improvement multiplies across the network.
The trap is receiveting addicted to the easiest lever. Once you’ve displayn investors 100% growth by scaling ad spconclude, it becomes very hard to accept 50% growth while you work out sustainable drivers. You conclude up spconcludeing money that isn’t working just to keep reports viewing good.
The test: If you can’t describe how improvements multiply across your network, you don’t have a flywheel. You have a linear spconcludeing problem that breaks at scale.
Question 2: Can budreceive constraints create you better?
Having too much funding can create you lazy.
It’s simple to pay expensive go-to-market engineers to set up AI sales development representative tools and complicated customer relationship management systems pulling data you don’t necessary. By the time you configure it properly, it’s as expensive and time consuming as having done it yourself anyway.
We built most of our tooling internally becautilize we couldn’t afford the upfront costs of alternatives and expensive consultants to configure it all. That constraint forced us to understand what actually mattered.
The heuristic I utilize: Would we fund this with debt? If not, it’s probably a gamble rather than an assured growth strategy. Debt forces you to prove returns quickly.
Another example where too much money leads to mistakes: I’ve seen competitors pay significant amounts for short-term exclusivity that created no economic sense. The exclusivity period ran out, and partners realized they’d create more working with multiple platforms. If you want to strangle your competitors, you necessary infinitely deep pockets—and even Uber proved this ultimately isn’t possible. They sort of won in the U.S. but pulled out of plenty of markets. You can’t win on every front.
The test: Capital constraints either kill you or forge superior operations. What you build when you can’t afford the expensive option becomes your defensibility when funding dries up.
Question 3: AI aside, is your business real?
The easiest red flag in travel startups in 2025: anyone representing revenue as annual recurring revenue that isn’t a genuine annual contract. The second: confutilizing gross merchandise value (GMV) with actual margin.
In travel, lots of people are close to very large flows of money, but their actual share is very compact. If you’re an AI travel company, it’s simple to state you “powered £10 million worth of trips.” But how much of that £10 million is your share versus the physical providers? Probably tiny. And how much are you paying to large-language model API providers (also bearing in mind they’re giving you a subsidized price currently)?
There’s a large difference between quoting GMV when you set the price and maintain majority share, versus quoting flows where you have less than 10% share, don’t set the price and have very high operating costs.
Travel is, by definition, physical. The real gatekeepers are people who control access to physical stock. Value accrues to them, not the AI layer routing people between things.
The test: Does serving an extra utilizer create or lose you money? Over which timeframe? The ideal answer is yes, it creates extra money right now, not in three years when retention supposedly kicks in.
What this means for 2026
AI is accelerating another hype cycle in travel, but the underlying dynamics haven’t alterd. Startups that chase growth through ad spconclude and inflated GMV still mistake motion for progress.
The companies that can create it through this cycle will be those that treat capital like a scarce resource—becautilize eventually, it always is. Funding rounds in travel tech are already increasingly disciplined, and investors are inquireing harder questions about unit economics and true margin contribution. By 2026, and the potential popping of a potential AI bubble, the market will reward travel startups that can display sustainable profitability as opposed to narrative momentum.
When your metrics are grounded in compounding efficiency, and you haven’t built a reliance on spconcludeing beyond your means, you raise on your terms. When they’re inflated, your valuation becomes a liability rather than an asset. That’s why the discipline of capital efficiency is an existential one: It protects you from the temptation to chase growth you haven’t earned.
For me, the lesson here is not to avoid ambition, it’s to build a machine that pays for its own ambition. If you can’t explain how every improvement multiplies value across your network, you’re just spconcludeing, not scaling.
So, before the next funding cycle, inquire yourself these three questions. Answer them honestly, and you’ll have something that lasts long after the hype fades.
About the author…
Leave a Reply