Digital consumer behavior: similar phone use, different computer habits

The survey asks about social networks and, separately, online shopping, travel reservations or other consumer services. We call the second item “consumer activity” throughout this article. Both sit within the questionnaire’s six-month activity framing. A yes answer can cover occasional use; it does not measure how frequently someone performs the activity.^[1]

Those two answers produce four groups. The largest, 62.9%, reported both activities. A further 13.4% reported social-network use without the consumer activity, while 13.8% reported the consumer activity without social-network use. The remaining 9.9% reported neither. Together, the two groups reporting just one of these activities account for 27.1% of eligible adult Internet users.

The labels describe answers to two particular questions. Someone in the social-without-consumer group might buy in a physical store, have made an online purchase outside the survey window, or interpret the broad consumer question differently. Someone in the reverse group is not necessarily unreachable through advertising. The survey does not identify their marketing exposure or acquisition channel.

This also prevents an easy but misleading label: these are not “social shoppers” and “nonshoppers.” The consumer item includes travel reservations and other services, and it does not require a completed purchase. Keeping that definition intact matters whenever the figures appear in a campaign brief or presentation.

Phone use was high in both single-activity groups: 91.3% among people reporting social use without consumer activity, and 92.6% in the reverse group. These point estimates are descriptively close. We did not test whether phone use was statistically equivalent.

Computer use tells a different story. The corresponding estimates were 55.1% and 78.9%, a gap of 23.8 percentage points. Here, computer means a desktop or laptop. Phone means an Internet-capable cell phone, including a smartphone. The questions describe general use at least occasionally, rather than ownership or use for a specific shopping task.^[1]

The full table keeps the other two groups visible. Computer use reached 81.1% among people reporting both activities and 49.2% among those reporting neither. That spread argues against turning the profiles into rigid personas. Even in the social-without-consumer group, more than half reported computer use. No group can be reduced to a single screen.

Tablet or e-book-reader use adds another part of the picture. Its estimates were 25.8% and 37.0% in the two single-activity groups. The questionnaire combines those devices, so the figures cannot isolate tablet users or establish that a tablet substitutes for a laptop. We retain the source’s combined category.

Looking at devices together makes the distinction more concrete. An estimated 39.5% of the social-without-consumer group used an Internet-capable phone without reporting desktop or laptop use. The corresponding estimate was 18.0% in the reverse group. The denominator is everyone in each activity group, not just its phone users.

“Phone without computer” describes this particular combination of reported use. It does not mean the person owns only a phone, cannot obtain a computer, or never uses another device. Tablets, e-book readers and other devices remain possible. Nor does it show whether the person used a phone for the consumer activity. For a team planning a review, the useful distinction is between a familiar device and an assumed alternative.

The difference in computer use remains substantial in our prespecified age check. We divided adults into four age bands: 18–29, 30–44, 45–64 and 65 or older. We then recalculated computer use in the two single-activity groups as if each had the age distribution of all eligible adult Internet users in the sample.^[2]

The original gap was 23.8 percentage points, with a pointwise 95% confidence interval of 21.9–25.8 points. After this adjustment, the gap was 24.6 points, with an interval of 22.5–26.7. Using this common age mix therefore did not remove the contrast.

That is a narrower result than saying age does not matter. People can differ within each broad age band, and the calculation does not adjust for income, education, work or other circumstances. It does not show that one activity causes computer use, or vice versa. It shows that the observed contrast survives this particular change in the groups’ age composition.

For a marketing team, the implication is to avoid replacing one assumption with another. Knowing that an audience is young or old does not, by itself, settle which devices it normally uses or how it will approach a particular resource. The source supports checking those details, rather than assigning them from an age label.

Return to the fictional workshop. Its creator wants people to understand the offer, choose a session and use the workbook. These are separate tasks. A participant might read the offer comfortably on a phone yet choose a computer for the workbook. Another might use a phone throughout. Those are possibilities to investigate; the national survey does not establish either sequence.

The team can start with one concrete task: “Find the session that fits your schedule and explain what you would receive.” Observe the device the person chooses. Record whether they can find the information and what, if anything, makes the task difficult. A survey profile should not supply the answer in advance.

Next, ask the person to open a sample workbook and complete one representative exercise. Record the device actually used and the devices they normally use separately. If they change devices, ask what prompted the change. If they stay on the same device and complete the task easily, that is useful evidence too. The aim is to learn about the resource, not to manufacture a device problem.

Suppose, purely as a worked example, a participant understands the offer on a phone but asks to use a laptop for an exercise requiring several documents. That observation could justify checking the exercise’s instructions and file handling. It would not prove that every phone user needs a different workbook. Conversely, smooth completion on a phone would give the team a reason to keep that implementation while checking other relevant situations.

The national percentages provide context, not recruitment quotas or a forecast of this creator’s customers. They cannot determine a device-specific advertising budget, predict sales or establish how much a redesign would help. A team that already reviews real tasks across varied device habits may need no operational change. The useful addition is to keep “uses a phone,” “also uses a computer” and “chooses this device for this task” as separate facts.

Use the following prompts for one offer, booking page or resource. They are an application of the findings, not a validated assessment instrument. Keep the record focused on a task that matters to the customer and a decision the team can actually make.

Task and success: Name the page or resource, then describe what the person should be able to do. For the workshop example: choose a suitable session and explain what is included. Avoid using “check the mobile experience” as the entire task.

Audience evidence: State what you know from your own customers and what remains an assumption. Record devices normally used separately from the device chosen for the review. Do not infer ownership, purchasing ability or customer value from either answer.

Observation and follow-up: Describe what happened in the task, including successful completion. Identify any change worth investigating, who will examine it and what evidence would support keeping or revising the current design. A small review can reveal something to inspect; it cannot supply a reliable conversion forecast.

This brief can accompany an existing campaign plan. It requires no new tracking arrangement and does not replace the business’s actual audience evidence with national averages.

Public survey files can contain values supplied through the producer’s editing and imputation procedures. The Census allocation flags identify these values. Our main analysis retains the released answers; we also planned checks that set aside particular allocated answers before examining the results.^[3]

The first check kept 26,963 adults whose two activity answers were marked not allocated. Their unadjusted computer-use gap was 27.3 percentage points, compared with 23.8 in the main analysis. Giving the groups a common age mix produced a 27.8-point gap, compared with 24.6 in the main analysis.

A further check required enough unallocated computer information to establish use or nonuse. It retained 26,959 adults and again produced gaps of 27.3 and 27.8 points after rounding. The complete estimates and intervals appear below. The direction and substantial size remained, while the exact estimates changed.

These checks do not prove that imputation has no effect. They also do not create a corrected population estimate: removing records changes the analyzed group, and we retained the original survey weights. The full released sample remains primary. Proxy reporting, residual nonresponse and the way people interpret broad activity questions can still affect the joint pattern.

We used the November 2023 CPS Computer and Internet Use Supplement, sponsored by NTIA and conducted by the U.S. Census Bureau. This article concerns civilian noninstitutionalized U.S. Internet users aged 18 or older in that survey. It is historical evidence, not an estimate of behavior in 2026 or in other countries.^[1][6]

The activities were asked for one randomly selected Internet user aged 15 or older in each participating household. We restricted the analysis to adults with a positive primary-respondent weight and known answers to both activity questions, leaving 35,879 records. Answers can be supplied by a household proxy. The results should therefore be described as reported use, not necessarily personal testimony from every person represented.^[4]

Every percentage uses the primary-respondent survey weights. Profile percentages describe the eligible adult Internet-user population. Device percentages describe the people within the named activity profile. The tables also show unweighted sample counts, which indicate the number of records supporting an estimate rather than the number of people in the country.

We used all 160 supplied primary-respondent replicate weights to estimate sampling uncertainty. Ratios, differences and age standardization were recalculated within every replicate. The intervals are approximate pointwise 95% confidence intervals: each concerns its own estimate. They are not simultaneous coverage for every table cell and do not measure all error from nonresponse, proxy answers or imputation.^[4]

Computer configurations have a wider research history. A 2025 study using the 2021 and 2023 surveys examined connections between computer and home-Internet configurations and activities such as job searches, government services and health information. Our contribution is narrower: comparing device use within the four social-network and consumer-activity profiles, with a defined age comparison and allocation checks. It adds a descriptive audience benchmark rather than a new theory of why people use particular devices.^[7]

The question, device definitions, main contrast, age bands, sensitivity checks and publication criteria were frozen internally before we calculated these joint results. Published background statistics had already informed the research, so this was not a blind study or a public preregistration. Two implementations independently reproduced the estimates. The download includes the method, aggregate outputs, code and instructions for obtaining the exact source files.^[5]