November 12, 2026
During the 2026 US midterms, online users turned to news outlets, discussion platforms, and predictions markets in different ways. RealityMine(R) data reveals sharp election-night spikes, more frequent app switching, and increased advertising activity, but not necessarily new users or sustained engagement.
Going into the 2026 US midterms, one survey about mobile behavior during the election season reported that, among US consumers, 30% said they’d rely on social platforms for election coverage, and 28% on news websites/apps. What’s more, more than half (56%) said they were actively trying to manage their exposure to social-media and political content online.
Our data shows what actually happened on people’s phones during the midterms. While those on our US panel of around 63,000 daily users didn’t necessarily adopt new phone habits during the midterms, they used existing platforms in different ways at specific moments.
To investigate this behavior, we grouped apps and websites into five categories relevant to the election cycle:
We then compared usage leading up to the midterms and on election day itself. Here’s what we found.
Note that all times below refer to Eastern Standard Time (EST).
News outlet use was flat through the lead-up week, then more than doubled on election day itself and stayed about double normal the next day, before returning to normal by Nov 5. Within election day, usage was about double the usual rate from about 12pm and accelerated from 5pm. At 8pm it jumped to about seven times a typical Tuesday at the same hour.
News aggregators showed a more modest spike. Reach stayed close to normal until late evening and picked up from around 11pm. By midnight 2.5 times as many people were using them as on a typical night.
Unsurprisingly, results and forecast sites saw the most dramatic jump, but from a very small base (these platforms aren’t used much at all when it’s not an election). Their reach was more than 30 times normal on election day and the day after, surging from around 8pm on election night.
The spike was short-lived. News outlet use was back to normal by Nov 5, and results and forecast sites were back to their pre-election level by Nov 6. Brands planning around similar events should keep in mind that this is a small window of visibility rather than a sustained engagement.
Unlike news outlets, real-time discussion platforms (e.g. X, Reddit, Threads) didn’t pull a bigger audience on election day. But what did change was checking behavior: people opened these platforms 22% more often but stayed for less time overall than usual.
Another change? Movement between news and discussion apps occurred far more than normal. For every 100 people using these platforms on election night, there were about five switches in each direction between news and discussion within a 10-minute window, compared with about two on a normal night, roughly three times the usual cross-traffic.
The hopping between news and discussion is a classic example of “information foraging,” where people jump between different sources looking for information during uncertain moments.
Understand how consumers move between apps and websites, and uncover the digital behaviors that shape their decisions.
Election-related ads on social (TikTok, Facebook, Instagram, and X) climbed steadily throughout October, topping out at 1.4% of all ads on October 31. That may sound small, but it means that 1 in 7 Android panelists saw an election ad on a typical day, up from 1 in 17 earlier in the month.
Most of that volume was what you’d expect: 89 of every 100 election-related ads in the final week came from campaigns, PACs and voting groups. But two of the categories we studied went in on midterms messaging too – prediction markets went from almost none in the baseline weeks to 4% in the run-up and 14% in the final week; Kalshi’s “Who will win the House?” ad was the single most-seen election ad among the brands we tracked.
A quarter of news-outlet social ads promoted election-night coverage in that final week (e.g., ABC News: “With so much at stake on Election Night...”; CNN: “Stay on top of every election with CNN”), though it should be said that these types of platforms don’t advertise on social nearly as much as prediction markets. News shifted messaging more strongly, but prediction markets brought more ad volume.
For prediction markets, reach was at normal levels until the day itself, then ran above normal for most of the day (20–70% in most hours), peaking at 8pm EST alongside news platforms. But politics seems to have taken over sports-betting for at least one night: Similar to real-time discussion, usage clustered around news more than usual, with 1 in 6 of their visits starting within 10 minutes of using a news outlet. (On a typical Tuesday night, that number is 1 in 2,000.)
What that bump wasn’t: a wave of new consumers. 61% of election-day prediction market visitors were pre-existing users of the platforms, which is slightly higher than the norm of 54%. So, the people showing up on election day skewed more toward existing users rather than less. Social advertising didn’t convert any better either: about 1 in 450 views of a Kalshi ad led to a visit within 30 minutes, and that rate was identical whether the ad was election-themed or not. A spike in visibility is not the same thing as a spike in acquisition, and without tracking the conversion, there’s no way to determine which one happened.
A brand’s own data can tell you impressions went up on election night. It can’t tell you that a sixth of Kalshi’s prediction-market traffic arrived within ten minutes of someone leaving a news app, or that the ad spend brought attention without moving a single new customer to convert. Those are the questions that actually decide whether a campaign around a moment like this is worth the spend. And they’re only answerable when you can see what the same person did across apps, sites and ad exposure.
RealityMine® helps you uncover what consumers actually do across the digital landscape, revealing insights that can inform smarter business decisions.
Featured
