PredictHQ

Here is a fact about demand forecasting that is both obvious and, until recently, largely ignored by the software that does the forecasting: the real world happens on top of your business, and it moves the numbers. A marathon reroutes your delivery drivers. A stadium concert empties your suburban store and floods the downtown one. A school holiday quietly relocates an entire week of demand three miles east. Most models never see any of this. They know history, and history alone, which is a bit like navigating by looking only in the rearview mirror.

PredictHQ exists because two people ran directly into that blind spot and it cost them money. Campbell Brown and Mike Ballantyne were running a global travel company, and they kept getting blindsided by demand surges they couldn't explain - until they started tracking the real-world events sitting underneath them. That tracking, they say, unlocked $1.2 million in value in its first year. This is the useful kind of origin story, because it isn't about a whiteboard epiphany. It's about a spreadsheet that suddenly started paying for itself.

So in 2015, Brown and Ballantyne teamed up with Robert Kern and founded PredictHQ in Auckland, New Zealand - a country better known for exporting sheep and Lord of the Rings than enterprise data infrastructure. The idea was to take the thing that had worked inside their own company and turn it into a product anyone could buy: a single, verified source of truth for the events that drive demand.

Sit with that number, because it reframes the whole exercise. If events explain the majority of why demand moves, then a company that models everything except events is spending enormous effort on the minority of the problem. That's the pitch, and it's a good one: not "we have more data," but "we have the data you were structurally ignoring."

The unglamorous moat

The clever part isn't collecting event data. Anyone can scrape a concert calendar. The clever - and tedious - part is what PredictHQ does next: verifying that the event is real, deduplicating the same show listed on six different sites, estimating how many people will actually show up, and ranking how much any given event is likely to matter. This is deeply unsexy work, and that is precisely why it functions as a moat. Nobody wants to do it, which means somebody can get paid to do it once, well, for everybody.

The result is delivered the way modern enterprises actually want data: through an API, or dropped into the data warehouse they already run. You don't rebuild your stack to use PredictHQ. You make a call, or you subscribe on Snowflake, and verified event context shows up next to your own numbers.

PredictHQ's customer list reads like a directory of businesses that live and die by getting demand right: ride-hailing, pizza delivery, grocery, hotels, airlines, retail pharmacies. When your business is fundamentally about staffing, stocking and pricing against a crowd you can't yet see, verified event data isn't a nice-to-have. It's the crowd, described in advance.

The sectors sort into a few clean buckets - retail and quick-service restaurants managing inventory and staff, transportation and logistics planning capacity, travel and accommodation pricing rooms and seats, and financial services hunting for signal. The common thread: every one of them is trying to answer the same question a day, a week or a quarter early. How many people, and when?

The AI turn

The most recent chapter is a repositioning that, on inspection, isn't much of a pivot at all. As companies race to put AI models into production, they keep hitting the same wall PredictHQ's founders hit a decade ago: a model that only knows the past makes confident, wrong decisions when the present looks different. A forecasting system that knows a hurricane is inbound, or that a stadium holds a sold-out show on Thursday, simply makes a better call. PredictHQ now frames itself as the layer that feeds production AI that verified real-world context. Same data, newer buyer.

In 2026, the company put a headline number on it - partnering with Expedia Group on joint forecasts projecting more than $8.1 billion in traveler spend across North American host cities during the summer. It's a tidy demonstration of the thesis: big events are forecastable months out, and the spend, the crowds and the surge are all knowable if someone has bothered to structure the data.