Methodology.
Research design, scope, modelling approach and limitations for the economy, upskilling, and retail and CPG chapters.
What this series measures
The headline numbers in the series answer different questions and are built in different ways. They are shown side by side and are never added together.
How much activity in the Saudi economy is linked to businesses using the platform. SAR 5.7 Bn of it is linked to retail and CPG.
Work across those businesses, their suppliers, and the spending of the people they employ. About 28,000 are linked to retail and CPG.
How many hours a year adult users spend on content that teaches or explains a skill.
How the learning hours are measured
TikTok records time spent on content it classifies as learning and reports it to Redseer in aggregate. The learning categories are business and finance, education, home and life skills, humanities, law, science, technology, and workplace. Time on learning content recorded from 24 June 2025 to 24 June 2026 came to 363.7 Mn hours. Removing the 18.9 Mn hours attributable to users aged under 16, estimated from the share of 13 to 15 year-olds in the population aged 13 and over, gives 345 Mn hours. The hours cover all adult users, so they are not divided by the 13.3 Mn upskillers to give an average. The measure counts time spent rather than training delivered.
Counts of people, such as the 13.3 Mn upskillers and the 9 Mn shoppers, are survey results applied to a population base. Their builds are set out under National scaling.
Surveys and samples
Every survey finding in the series comes from one of these. Each percentage is a share of a defined group of respondents.
How the samples were built
The two consumer surveys ran online through a managed panel in February 2026. Both applied the same two rules: aged 16 or over, and living in a household with a combined monthly income of SAR 3,000 or more. Every retail respondent had also bought in at least one of five categories in the past 12 months (fashion, beauty and personal care, grocery, electronics, and home and living) and answered questions on each category they bought in. Fieldwork ran against demographic quotas, and results are not weighted.
The business survey ran online in July 2026. Marketing, sales and management staff answered for 452 Saudi businesses: 33 micro, 86 small, 196 medium and 137 large.
Each questionnaire was tested before launch, and each consumer survey ran a 50-person pilot that is excluded from the counts. During fielding, answers were checked for people rushing, giving the same answer to every question, or contradicting themselves. Those responses were removed.
How precise the findings are
These are opt-in online panels, not probability samples of the population, so a standard margin of error does not strictly apply. Findings on a full sample of 452 to 575 are indicative to within a few percentage points. Findings on smaller groups carry much wider uncertainty, and differences of a few points between subgroups should not be read as real.
Scaling survey findings to national estimates
When a chapter gives a number of people rather than a percentage, that number comes from applying a survey result to a population base.
The build starts from the 2025 population aged 13 and over, since 13 is the platform’s minimum age. The population comes from the IMF World Economic Outlook, split by age using GASTAT Population Estimates 2024. The survey share of internet users who used TikTok in the past six months is applied to it. Users aged 13 to 15 are then removed, because the survey and the chapter cover people aged 16 and over. Last, the 56% of surveyed TikTok users who encountered upskilling content in the past 12 months is applied, giving 13.3 Mn.
The build starts from the same 2025 population, limited to people aged 16 and over. Non-Saudi domestic workers and non-Saudi workers in elementary occupations are then removed, using GASTAT register-based labour market statistics for the first quarter of 2026, because most fall below the survey’s income screen. The 54% of internet users surveyed who discover products on TikTok is applied to the remaining base, giving about 9 Mn.
Numbers built this way are indicative. They carry the error of the survey and the error of the population estimate together, and should be read as an approximate scale rather than an exact count.
Economic impact framework
The model starts with what businesses spend on advertising, turns that into the sales it produced, then follows those sales through the economy. It covers seventeen sectors and includes large companies and small businesses together.
The starting data
The volume of advertising activity by businesses on the platform, a measure of the sales that activity produced, and the sector mapping that places each business in the national accounts. TikTok supplies these inputs at a category level, separately for small and large advertisers, and they are cross-checked against the business survey, in which respondents set out their marketing spend by channel alongside order values and sales volumes. The exact commercial values are not published.
What businesses get for free is counted too
Businesses use tools that cost them nothing: publishing content, creating it, and measuring whether it worked. Because no money changes hands, none of that appears in the national accounts. The business survey asked firms what they would pay each month to keep those tools if they stopped being free, and the value stated by small and medium businesses is converted into the additional sales it represents. Those sales are added to the advertising-driven revenue before anything flows through the economy, so the value of the free tools is included in the totals rather than reported separately.
Sales then travel through the economy by three routes
Value created by the businesses themselves, after taking out what they buy from others.
Value created by their suppliers as those businesses order more.
Value created when the people employed along that chain spend their wages.
Scope and exclusions
Sales moved from somewhere else
The figures are gross. If a sale moved from another shop rather than being new, the model does not take it out.
What the money would have done elsewhere
The model does not ask what those businesses would have earned by spending the same budget on another channel.
Taxes and subsidies
These are not calculated. Gross value added is used in place of GDP contribution throughout.
How to read the estimates
Survey answers record what someone remembers doing and what they credit it to. That is their own judgement, not a measured result. The rules below apply everywhere in the series.
People are describing themselves
Answers are self-reported judgement, not a measured outcome.
Link, not cause
Someone saying content influenced a purchase does not prove the purchase would not have happened anyway.
Jobs are supported, not created
The figure counts work linked to the activity measured. It is not a count of new jobs.
No other platform is named
Where TikTok is compared with one other platform, that platform is the one ranked second on the measure shown, labelled Platform 2 or the next platform.
Evidence classes used across the series
TikTok commissioned this series. Redseer designed the research, selected the methods, and is responsible for the estimates and the interpretation in every chapter.
TikTok supplied platform inputs at a category level, including advertising activity, return on advertising spend and time spent on learning content, and this page sets out how each is used.