Business statistics are useful only when their scope and status are clear. An estimate of internet use, an employer expectation and an economic forecast are different kinds of evidence. This article compares dated indicators from three primary sources and shows how to use them in planning without presenting a future scenario as an observed result or a global total as your addressable market.
Label each indicator before comparing it
A statistical estimate approximates a quantity using available data and a stated method. It need not concern the future. An expectation describes what someone anticipates; a forecast uses evidence and assumptions to describe a possible future outcome. A projection shows what would follow under specified assumptions. An observed result records what happened within a defined period and scope, although the record may later be revised.
These distinctions explain why the following figures answer different questions. Internet use describes a population, job projections describe possible labour-market changes, and economic growth describes changes in output. None directly measures demand for a particular company.
| Source and period | Indicator | Status |
|---|---|---|
| ITU, 2025 | About 6 billion people online, or 74% of the world population | Statistical estimate |
| World Economic Forum, 2025–2030 | 170 million jobs created and 92 million displaced; net increase 78 million | Projection based on employer expectations and analysis |
| World Bank, June 2026 release | Global growth of 2.5% in 2026 | Forecast for the year |
The sources and periods are deliberately retained. These figures are not interchangeable, and their publication does not make the projected outcomes certain. Check the current release before using any forecast in a new decision.
Internet use is not a customer count
The ITU measure concerns people using the internet. It does not count online shoppers, paying subscribers or customers who can access your delivery area. Do not multiply a global user total by an assumed purchase rate and call the result a validated sales opportunity.
Hypothetical example for illustration: for an online service, narrow the population by geography, language, need and ability to buy. Then test actual interest and conversion through an appropriate small trial. Record the distinction between the broad digital environment and the evidence supporting the service’s reachable customer base.
Job projections are not an AI-only impact estimate
The report draws on more than 1,000 employers across 55 economies. It extrapolates their expectations; it is not a census of jobs already created or displaced. The World Economic Forum report examines several forces affecting work, including technological, demographic, economic and environmental changes. Its job figures should not be described as jobs created or removed by AI alone. A net increase also does not mean the same people will move directly into the newly created roles.
Use the report to ask which tasks and skills in your organisation may change. Review actual roles, service requirements and learning needs. Do not turn a global projection into an automatic staffing reduction target. A workforce decision requires local evidence and appropriate consultation.
A growth forecast is not a company target
The World Bank figure concerns growth in real global output, adjusted for price changes. It is a dated forecast, not a final observation for 2026. Forecasts depend on assumptions and can change as conditions change. Do not use a global growth rate as a company sales target without examining the country, sector and customer segment involved.
Use alternative scenarios to test exposure to delayed demand, supplier costs or financing conditions. Record which assumption would change an investment or hiring decision. The purpose is to prepare for uncertainty, not attach false precision to a single number.
Remove unsupported statistics rather than refresh their year
An old prediction for a technology market does not become a current market value when its target year arrives. A mobile-search share is not a voice-search share. A consumer survey about stated preferences is not proof of purchasing behaviour. Check the original question, sample, geography, period and measurement before repeating a headline.
If the primary source cannot be located or the definition is unclear, omit the number and explain the evidence gap. Preserve a dated source record and document revisions. Never combine percentages from different populations to create an apparent market trend.
Test the assumption between a global figure and a sale
Hypothetical example for illustration: consider a provider deciding whether to offer online training in one market. The internet-use estimate establishes a broad context, but the decision still depends on reaching suitable buyers, solving a recognised need and delivering at an acceptable cost. Write those assumptions separately. Do not hide them inside a calculation that begins with billions of internet users and ends with a precise revenue forecast.
In this fictional trial, suppose 200 eligible people receive the same offer during a defined 30-day period. Of these people, 30 request details and 10 of those 30 buy within that period. Count each person once. The observed purchase rate among those invited is 5%, while the rate among people requesting details is approximately 33.3%. The calculations are 10 ÷ 200 × 100 = 5% and 10 ÷ 30 × 100 ≈ 33.3%. The denominator is the group against which the buyers are compared. Both calculations can be correct because their denominators differ. Neither is the purchase rate of the whole market. Record how participants were selected, whether an incentive was offered and whether the price and delivery conditions were realistic. Do not switch denominators without explanation to make the result appear stronger.
Examine the buyers as well as the number of purchases. Did they belong to the intended segment? Did they complete the service, request refunds or require more support than planned? A low-price trial may test interest without establishing willingness to pay the planned commercial price. Keep these limitations visible when estimating the next stage, and avoid describing expressions of interest as completed sales. Separate the trial’s actual results from the team’s expectations about reaching more participants.
Decide what evidence would justify another limited trial and what would stop further spending. Assign an owner, a spending limit and a review date. These are proposed management checks, not conditions stated by the statistical sources. The global indicator remains in the context section; the trial results and unresolved assumptions support the specific business decision. Preserve that separation when the presentation is shortened for senior management. Removing an explanation must not turn an assumption into a fact.
Build a planning note that separates fact and inference
Write the published indicator, its period and status first. Follow with the team’s interpretation, the decision it could affect and the additional evidence needed. Give the review an owner and revisit it when a new release or a material operating change occurs.
ATMACSS strategic planning and KPIs can help turn external evidence into defined assumptions and measurable actions. The practical starting point is one decision and a source that actually measures the issue, rather than a long collection of impressive but unrelated statistics.