We got tired of travel brands guessing
TravelSpy was started by people who spent years inside travel commercial teams watching brilliant marketers make expensive decisions on almost no evidence. The data existed. It was just scattered across forty systems and nobody had joined it up.
The trip graph
Travel is the only major purchase where the customer does months of visible research and the seller retains almost none of it. Someone reads a destination guide, compares four operators, checks the weather, looks at flights twice, asks a forum a question, and then books — and every brand involved saw a fragment.
We built TravelSpy to join those fragments into one structure: the trip graph. It connects what a traveller does, where they came from, what the market is charging, what is happening in the destination and what the economy is doing to their budget. From that graph, prediction stops being guesswork.
The result, consistently, has been a fifty per cent lift in booking conversion for the brands running it — not because the marketing got louder, but because it got correct.
Founded
2023, London. Now distributed across the UK, Spain and Portugal.
Team
62 people, roughly half in engineering and modelling, a third from travel commercial backgrounds.
Coverage
Customers operating in 34 markets, modelling around 41 million sessions a month.
What we believe
Evidence beats instinct
Experience is valuable and frequently wrong. We build tools that let a team test the instinct instead of arguing about it.
Explain or do not ship
A score nobody can interrogate is a liability. Every prediction we make comes with its reasoning attached.
Consent is not a tax
The consent-first constraint made our models better, because it forced us to build on signal rather than on surveillance.
Prove the lift
Holdouts are permanent, not promotional. If we cannot show incrementality, we have not earned the renewal.
Your data is yours
Full export, documented schemas, no lock-in clause. Retention through value, not through friction.
Boring infrastructure
The interesting parts should be the models. Everything underneath should be dull, observable and unremarkable.
How we got here
2023
Founded after a consulting project that accidentally became a product. First operator pilot runs on a spreadsheet and a lot of goodwill.
2024
Platform rebuilt properly. Data source library passes fifty feeds. First measured fifty per cent conversion lift against a randomised holdout.
2025
Journey orchestration ships. ISO 27001 certified. Source library passes one hundred feeds and coverage expands to thirty-four markets.
2026
Private model hosting, EU and US residency regions, and a customer portal our users no longer have to apologise for.
Come and build it with us
We are hiring across modelling, infrastructure, product and revenue.
See what your traffic is actually worth
Book a 30-minute session. We will run your own traffic through the model and show you the segments, the predicted spend bands and the packages you are not selling yet.