73 percent. Almost three in four social purchases were never planned. Only 7 percent of social shoppers knew the exact product they ended up buying before the feed put it in front of them.

The number that matters is not the 73. It is the 7. Search advertising, retail media, and most of ecommerce is built to capture demand that already exists: somebody wants a thing, they go looking for it, you show up. On social, that shopper is a rounding error. 7 percent knew the exact product. Everybody else was doing something else when the purchase happened.

The full breakdown of how unplanned these purchases were:

Read those top two rows together and 72 percent of social purchases start with no product in mind at all. The feed does not answer a question the shopper was already asking. It asks the question.

Most published impulse-buying numbers range from 39 to 84 percent, and almost all of them describe ecommerce in general or come from surveys that predate social checkout. This one is specific to what people actually bought on social, in the six months before we asked.

The practical consequence is that keyword intent is the wrong model here. There is no keyword. The job is not to be findable, it is to be worth stopping for, and then to be trustworthy enough to close in the same session. That is why the trust signals matter so much on a channel that looks, from the outside, like it runs on impulse. The purchase is unplanned. The decision to hand over a card is not.

The full data, including the cross-tabs behind these findings, is in the Social Commerce Intelligence Report 2026. The free executive summary covers the headline findings. The full report (PDF) is available at three license tiers.

ScrollSignal is an independent B2B research firm publishing original consumer research on social commerce. No platform funding. No sponsored findings. n = 1,000 active US social media shoppers, ages 18 to 64, fielded April 2026, stratified to US Census on age, gender, and region.