Daily Weather Now Factored As Indicator Of How Consumers Will React To an Ad

Anecdotal evidence suggests that weather influences how we behave: taking a cab rather than walking, ordering take-out rather than going to a restaurant etc.

But now those influences can be factored into prediction models to help determine how consumers are likely to respond to ads. JL


Barry Levine reports in Venture Beat:

Marketers will now be able to evaluate if weather appeared to have an impact on a sale — and, if so, how can they take advantage of (or avoid) that impact in the future.
You may just think it’s raining outside your office window in Manhattan. But to AOL’s ad platform, today’s rain is a key reason why you decided to use an online discount coupon for lunch delivery.
In a blog post today, AOL Convertro general manager Amy Mitchell announced that the company is now adding daily weather to its advertising attribution and prediction models. Convertro is the attribution technology for AOL’s ONE ad platform.
This is a first step in AOL’s effort to bring real world conditions into its assessments, Mitchell told me, with local gas prices or neighborhood geography among the factors being considered as possible additions.
“The goal,” she wrote in the post, “is to create a higher standard for what contributes to a sale.” In other words, marketers will now be able to evaluate if weather appeared to have an impact on a sale — and, if so, how can they take advantage of (or avoid) that impact in the future.
Machine learning in the AOL ad platform considers thousands of variables when a sale is made from a digital ad in the network, so the advertiser can hopefully use similar triggers to generate more sales under similar conditions.
Before, Mitchell said, the weather’s impact was considered in the aggregate, such as an average temperate for a week across a region.

---

---

Now, the weather is considered at ground level in a granularity that Mitchell says is finer than what any other ad attribution service can offer. Many competitive services consider weather, she noted, but with the kind of broad brush that AOL previously used.
Aggregate weather has been considered as a factor for decades, Mitchell pointed out, even since the advent of the marketing mix modeling, or MMM. MMM uses statistical analysis to determine which marketing tactics work best, and then it recommends specific ways to get the most out of the next such assignment.
The updated AOL attribution model collects high and low temperatures plus precipitation every day for each zip code in the U.S., and then it adds 27 variables for such things as a comparison of temperatures over recent days. The data comes from the National Oceanic and Atmospheric Administration.
The model might now predict that for online clothing ads, an email campaign is the best bet on blizzard days in Boston’s South End, since people remain indoors with their computers. But an app-based campaign might be suggested as the better approach on sunny and moderately cold days for the same neighborhood.