eMarketer's time-spent figures are a forecast, not a platform count

The test

EMARKETER, publishing at the time under its former parent brand Insider Intelligence, issues recurring forecasts of daily time spent on social platforms among U.S. users. Its 15 February 2023 release, 'Digital Video Time to Surpass TV Time in 2023', forecast that U.S. adults would spend more daily time with digital video than with linear TV for the first time that year, and broke social video out separately, forecasting that 'average daily time spent with videos on social networks among US adults will climb 9.3% to 45.2 minutes' in 2023, naming TikTok as 'a key driver.'

What the evidence says

The release states two distinct TikTok numbers and explicitly defines the difference between them in a footnote: 'Time Spent Among Adults' is 'average daily time spent on a specific social platform among all US adults; counts non-users,' while 'Time Spent Among Adult Users' measures only people who use the platform and 'does not count non-users.' Under the first definition, the release forecasts TikTok's total time spent among all US adults would climb 14.2% to 17.4 minutes a day in 2023; under the second, it forecasts that adult TikTok users specifically would average roughly 56 minutes a day, ahead of an estimated 48 minutes for adult YouTube users. Every one of these numbers is a forecast, not a figure either platform disclosed.

The sample and the variance

eMarketer's own methodology page, current as of 16 September 2026, states that its estimates are 'based on our proprietary analysis' and combine 'quantitative and qualitative data curated from public companies, government agencies, research and media firms, and interviews with expert executives,' processed through a stated cycle of defining, collecting, normalizing, analyzing, reviewing, contextualizing and re-evaluating each forecast at least twice a year. The same page claims 'less than 5% forecast variance against company-reported earnings' as a general track record, not a guarantee for any single social-platform time-spent number, and the platforms named do not confirm or deny these figures.

What to try next

A reader citing eMarketer's time-spent forecasts should specify which of the two metrics, among all adults or among adult users only, is in play, since the same platform can carry two very different minutes-per-day figures in the same release. This is an editorial habit: label the number a modelled forecast rather than a measured result, and check the current release before repeating a figure tied to a past forecast year like 2023.

• Does the cited eMarketer figure describe time spent among all adults, including non-users, or among adult users only?
• Is the number a forecast for a stated future or current year, or has it been carried forward past the year eMarketer forecast it for?
• What platforms did that specific release cover, since eMarketer's tracked list of social platforms has changed over time?

Two adjacent metrics with the same platform name attached can describe very different daily habits; eMarketer's own definitions are the only way to tell which one a cited number is.