Every floor has a theory about the weather. We joined a random pull of payments to the hourly weather at every zip code and checked. Almost nothing moved, and the one thing that did isn't payments.
Every collections floor has a theory about the weather. Rainy Friday nights are good. Snow days are dead. Nobody pays when it is 95 degrees. We had the data to check, so we checked.
We took a random pull of 660,594 payments that consumers set up through Resolvah portals. Every one of those has a timestamp and a zip code. The National Weather Service has an airport weather station within about 14 miles of nearly every zip code in the country, reporting every hour. We joined the two: for each payment, the weather at that consumer's location in the hour they paid.
Then we did the part that matters. Raw counts would mislead you, because it rains more in Seattle in November and people in Seattle pay differently in November for reasons that have nothing to do with rain. So every payment is compared with what that same place, same weekday, same hour and same month normally produces. A rainy Tuesday evening in Minneapolis in December is only ever compared with other Tuesday evenings in Minneapolis in December. What is left after that is the weather's own effect, and only the weather's.
Here is the whole story in one line: no kind of weather moved payments more than about 5%.
We tested the rainy Friday night theory specifically. Rainy Friday evenings ran 2% below dry ones. Not a strategy.
We also tested whether it is the weather being unusual, not the weather itself. A 45°F day is a cold snap in Miami and a thaw in Minneapolis, so we compared each day with its own local normal for that month. Days 15°F or more below normal: flat. Days 15°F or more above normal: flat.
If you have been holding a campaign for the forecast, you can stop. If a campaign is ready, send it.
The weather does not change whether people pay. It changes whether they answer.
The people who were going to pay still paid. The casual replies are what disappeared. That gives you a rule for snow days: email it, do not text it, and send the text the day after.
One more detail from the snow data: the mix of payments shifts. Pay-in-full drops 7% in snowy hours while total setups barely move, which means the people who do commit in a snowstorm are a little more stretched than usual. Worth knowing before you push a settlement that day.
Averages hide the extremes, so we looked at the big events one at a time, comparing each day in the affected states with the same weekdays in the surrounding weeks.
Cold delays the money. It comes back within a week, close to dollar for dollar. Hurricanes lose it, at least within the two weeks we could see. That is two different playbooks. After a freeze, wait; the payments are on their way. After a hurricane, the accounts in those zip codes need an easier offer, not another reminder.
Here is the perspective that makes all of this a footnote. The same dataset, measured the same way, shows the end of the month running about 36% above normal and the 4th through the 9th running up to 24% below. Friday runs 42% above the average day; Sunday runs 65% below. The strongest weather effect we found, snow, is a 5% dip. The entire weather story fits inside one week of the calendar.
Schedule to the calendar first, then to the hour of day, and treat the forecast as a footnote.
We published the calendar findings as their own report: resolvah.com/reports/payment-calendar.
A random pull of 660,594 payment setups and 3.4 million engagement events from the Resolvah platform, with account numbers hashed and customer identities withheld. Weather is the hourly observation from the nearest airport station to each consumer's zip code (1,691 stations, median 14 miles away), from the Iowa Environmental Mesonet ASOS archive. "Usual" is what that station, weekday, hour and calendar month normally produces, so seasons, geography, the weekly cycle and the daily cycle are removed before comparing. Ranges are 95% intervals from resampling weather stations; anything whose range crosses usual we report as no measurable effect. The full report, with every chart and the numbers behind it, is at resolvah.com/reports/weather-and-payments.
This is how we think — and how Resolvah works. See it for yourself.
Thirty minutes, a real operator, and the actual product. No script.