What External Data Can Do for a Small Business

External data is information about the world around your business that you didn't create, such as the weather, the App State calendar, a home football game, leaf season traffic, county tax receipts and public records. On its own, it describes the world. Joined to your own sales, bookings or job history, it can tell you which of those outside forces actually moves your numbers, so next Saturday's staffing or next month's order stops being a guess.

A forecast tells you what may happen. Public data tells you what happened. Neither one tells you how it affected you. That last part lives in your records.

What counts as external data?

Anything published by someone else that touches your customers. The National Weather Service runs a free public forecast API, which it describes as open data "free to use for any purpose." The North Carolina Department of Revenue posts sales and use tax collections by county and by type of business. The National Park Service publishes visitor counts for the Blue Ridge Parkway. App State publishes when students arrive, when they leave and when parents come up for graduation.

The difference from your own records is ownership. Your point-of-sale history, your books and your calendar belong to you, and what that internal data can tell an AI model is its own subject. External data belongs to everyone, which is exactly why it's worth so little until you join it to something only you have.

What are drivers and indexes?

Two words help here. A driver is a real thing in the world that moves a business's numbers, like rain, the semester, gas prices, a road closure on 421. An index is one number built to track a driver. You already know how indexes work from the heat index or the price of gas on the sign at the corner. Many readings go in, and one number comes out that you can watch week to week.

Nobody publishes an index for what moves your Saturday. The weather service publishes temperature and rain chance. It doesn't publish "how busy the salon gets when the forecast turns," because that number depends on your salon. It can't be bought, so if you want it, it has to be built from public sources and your own history together.

A business may need exactly one of these. A roofing crew's whole season can hinge on rainfall, and the school calendar means nothing to it. An accounting office near campus might care about the calendar and not the weather at all. Building a pile of indexes because they're available is how you end up paying for a dashboard full of numbers nobody acts on.

How do you join public signals to your history?

The join is lining up your records and the outside data on the same calendar, week by week or day by day, and asking which outside movement shows up in your numbers. That sounds abstract until you look at how much the outside world swings on its own.

Take short-term rental occupancy across the High Country. Across nine markets, listings-weighted, occupancy ran from a low of 28.5% in April 2026 to a high of 42.8% in December 2025, with summer climbing back to 41.5% by July. That's market data from a commercial short-term-rental data provider, via the Pigasus High Country panel. A swing that size is happening to every lodging business up here, and no single owner's records would show it as a market. Each one would just see a slow April.

Now put that next to one lodge's bookings. If its April is down about as much as the market's, the market explains it, and there's no fixing it with a discount. If the lodge fell twice as far, something about that lodge is the story. The outside number turns your bad month into a question you can actually answer.

Weather works the same way for a retail shop on King Street. The register says Saturday was slow, and the forecast says Saturday rained. Lay three years of Saturdays against rainfall, home games and leaf season, and you can see which one actually explains the slow days, and whether rain costs the shop a little or a lot. The school calendar does it for a restaurant that loses half its lunch crowd the week after commencement, and for any owner who staffs from the student labor pool and watches it leave in May.

I built the High Country panel partly to see how far public and licensed data could go. Most of the effort went into finding out which sources opened a door at all, which ones published a spreadsheet once a quarter and called it data, and which ones changed their format without telling anybody. That last group is why I check a source's track record before I build anything that depends on it.

Where the public sources are

These are the drivers that come up most often for owner-run businesses up here, and where the numbers live.

Treat the table as a map of where to look, and pick one row. Getting the data into one place is mostly plumbing, and the plain-English version of APIs, connectors and MCP covers how one program asks another for it. Why connected systems are worth more than separate ones is the case for data integration in a small business.

What should you look at first?

Start with the driver you already complain about. If you've said "it always dies when it rains" or "the town empties out in May," you've named your driver, and the first job is checking whether your records agree with you. Sometimes they won't. Owners are often right about the direction and wrong about the size, and the size is what decides whether to cut a shift or order less.

A spreadsheet can handle the first pass. Where AI models start to help is when you have a few years of history and more than one driver at once, and you want to ask in plain English which one explains last October. That's Level 2 or 3 in how to use AI in your business.

When the driver is clear and the history is clean, the number can be built to look forward, so next week's forecast becomes next week's staffing estimate. That's the work behind Pigasus intelligence.