How Often Do People Pick Up Their Phones? Why There Is No Single Average

There is no defensible universal average for phone pickups per day. The circulating figures count six different things, rest on samples as small as 23 people, and two depend on a platform metric with no published definition. Two are citable with conditions attached: 85.76 Apple Screen Time pickups a day, and a median of 51 among US teenagers.

The commonly circulated figures reviewed for this article are each either counting something other than pickups, or drawn from a metric whose owner has not published a definition anyone can check, or resting on a sample small enough to fit in a seminar room.

The limits of that answer matter as much as the answer. It does not mean nobody has counted. Several research groups have logged real phone use on real devices and published what they found, and those numbers are usable if you carry their conditions with them. What it means is narrower: the published counts cannot be averaged together, cannot be compared with each other, and cannot be applied to a population they were not drawn from.

Five things stand between the published counts and a universal average.

Construct incompatibility. Pickups, unlocks, glances, sessions, episodes and touches are six different measurements. The figures in circulation mix all six.

Vendor definitions that are not published. The single best-documented pickup figure available comes from Apple’s Screen Time. A documented search of the research databases and of manufacturer documentation, dated 4 September 2026, did not locate a published definition from Apple or Google of what counts as a pickup or an unlock.

Sample sizes. The two most-quoted device-logged figures rest on 23 people and 199 people.

Dispersion. In the 23-person study the standard deviation was 55.23 against a mean of 84.68, roughly two-thirds of the mean. In a 203-teenager logging study the pickup range ran from 2 to 498 a day.

Variation by age, country and year. The available figures come from a UK sample in 2015, a UK and US sample in 2020, two US panels in 2016, and a US teenage sample in 2022. Even within a single source, Deloitte’s 2016 US survey reported 47 checks a day overall and 82 a day for 18 to 24 year olds.

Six things get counted and only three of them have a published definition

The taxonomy below comes from Parry and Toth’s methodological primer on extracting usage measures from Android event logs (Parry & Toth 2025). Treat it as one research group’s proposal rather than a field standard: it is the clearest published attempt at these distinctions located in the documented retrieval, and it has not been adopted as a convention by the platforms that produce the numbers.

Measure What it counts Where the definition comes from What it is not

Glance A check of the phone without unlocking it Parry & Toth 2025 Not a session. Nothing is opened. Session One unlock-to-lock period Parry & Toth 2025 Not an app use. A session can contain many. Episode A single app use within a session Parry & Toth 2025 Not a session and not a pickup. Unlock A locked-to-unlocked transition on the device No published definition from Apple or Google located in the documented retrieval, 4 September 2026 Not a glance. Glances leave the phone locked.

Pickup The device-attention event reported by a platform screen-time dashboard No published definition from Apple or Google located in the documented retrieval, 4 September 2026 Not an unlock count, because the rule that converts device events into a pickup is not published.

Touch A physical contact with the screen surface dscout 2016, a vendor study with no methods appendix Not a check and not a pickup. One check can contain dozens of touches.

Two of these are the ones people actually want when they search for an average, and they are exactly the two with no published definition behind them. That is a statement about what the documented retrieval located in public documentation, and nothing more. What it costs the reader is concrete: without a published definition, nobody outside the company can check what the metric counts, compare it across platforms, or rely on it to stay constant between software versions.

The two logged figures that appear to agree do not agree

84.68 uses per day (Andrews et al. 2015). Twenty-three UK participants were analysed out of twenty-nine recruited, logged on Android for fourteen days. The measure was device-active periods, which includes screen-off activity such as calls and music. It is not a pickup count. The same participants estimated 37.20 uses a day. The standard deviation was 55.23, and the correlation between what people estimated and what their phones logged was r(21) = .11, p = .610, which is to say no relationship at all (Andrews et al. 2015). The point estimate is weak evidence: no direct replication of the 84.68 figure was located in the documented retrieval. The estimate-versus-log gap is the strong finding in that paper, not the number.

85.76 pickups per day (Shaw et al. 2020). One hundred and ninety-nine iPhone users in the UK and US, 2020, with the figure read from Apple’s Screen Time. This is the best-documented pickup figure available, and its central term is a metric whose definition Apple has not published. No direct replication was located in the documented retrieval.

Those two numbers sit 1.08 apart, and that near-agreement is where most articles on this subject go wrong. It is a coincidence of two different constructs. One counts device-active periods on Android including screen-off audio; the other counts whatever Apple’s unpublished rule treats as a pickup on iPhone. They cannot corroborate each other, because they are not measuring the same event. Two instruments that measure different quantities landing on similar values is a fact about arithmetic, not about phones.

The biggest number in circulation is a vendor figure with no methods appendix

2,617 touches per day. This comes from a 2016 study by the research platform dscout, of 94 Android users over five days, of whom roughly two thirds completed. Touches are physical screen contacts: taps and swipes and keystrokes. They are not checks, and a single check can produce many of them. The study has no methods appendix, no DOI, and no replication was located in the documented retrieval (dscout 2016). It is quoted more often than any peer-reviewed figure on this subject, usually without the word touches.

The 47-a-day figure is people’s guess about themselves

Americans check their phones 47 times a day is a self-reported survey estimate from the US edition of Deloitte’s Global Mobile Consumer Survey, which reported that “during the day, we look at our phones approximately 47 times and that number rises to 82 for 18- to 24-year-olds”. The survey covered 2,000 US respondents aged 18 to 74, split across six age groups, and was fielded online by an independent research firm (Deloitte 2016).

Two things about that figure are worth carrying. Deloitte reported it as flat: its own later description says the number had remained nearly constant for approximately three years at around 47 times a day. A figure that does not move across three years of rapidly changing devices is one to question rather than to trust. Deloitte also appears to have discontinued the metric: no daily check-count figure appears in the 2019 global report or in the firm’s later connected-consumer reporting, so there is no current edition of it to cite.

Retrospective frequency estimates are also the specific measure known not to track logged behaviour. A preregistered systematic review and meta-analysis of 106 effect sizes found that self-reported phone use does not correspond to logged use (Parry et al. 2021). The pooled sample size for that synthesis could not be verified for this article. Andrews’ 23 participants are a small illustration of the same thing: they guessed 37.20 and logged 84.68, and the two were uncorrelated.

So the 47 figure is real, in the sense that people really said it. It is evidence about what people believe they do. It is not evidence about what they do.

The one recent objectively logged pickup figure is from teenagers

A 2023 Common Sense Media report logged 203 US teenagers aged 11 to 17 on Android over a nine-day window, with data collected between August and November 2022. The median was 237 notifications received and 51 pickups a day, with a median of 46 notifications actually seen. The pickup range ran from 2 to 498 (Radesky et al. 2023).

This is objective logging, which makes it stronger than any survey figure here. It is also a non-probability sample of Android-owning American teenagers, which means it does not describe adults, non-US teenagers, iPhone users, or any general population. The 496-point range is the part worth keeping. Whatever the median is, it describes almost nobody in the sample.

What the retrieval did not find is the second finding

A documented search across eight query formulations, dated 4 September 2026, did not locate a qualifying post-2020 large-sample peer-reviewed study reporting a mean pickups-per-day figure. The best available peer-reviewed pickup number is still the one from 199 iPhone users in 2020.

That absence shows up on the results page too. The SERP check recorded for this query on 4 September 2026 returned a first page made up of forum and newsletter posts, a retailer blog, a comparison site and general-interest coverage, and none of the reviewed results traced its figures back to the studies that produced them. Within the documented retrieval there is nothing newer and larger for them, or for this article, to cite.

How to cite a number without inheriting the problem

If you need a figure and you want it to survive scrutiny, three rules cover it.

Name the construct before the number. “85.76 Apple Screen Time pickups per day” is citable. “85.76 pickups per day” invites the reader to compare it with a touch count.

Carry N, country, year and method every time. A figure from 199 iPhone users in 2020 is a different object from a figure from 94 Android users in 2016, and the difference is not decoration.

Never merge across the six measures in the table above, and never average two figures because they look close. The closeness of 84.68 and 85.76 is the clearest example in this literature of why that fails.

If you want a single number to quote, the defensible options are 85.76 Apple Screen Time pickups a day among 199 UK and US iPhone users in 2020 (Shaw et al. 2020), or a median of 51 pickups a day among 203 US Android-owning teenagers aged 11 to 17 in 2022 (Radesky et al. 2023). Both are honest. Neither is an average for people in general, and the documented retrieval located no such average to quote.

A reader who accepts that there is no average usually wants the next question instead, which is what actually drives the count. That is a question about cues and starting positions rather than about totals, and it is answered in why the defaults are doing the work.

How figures like these get used, and what the wider attention research does and does not settle, is set out in the digital distraction and focus guide.

HOS checks the figures that other publications pass along: what was searched, what was found, and what turned out to have no locatable source at all. The Weekly System has not launched yet. You can join the launch list.