“Requests increased by 50%” sounds precise, but it leaves several questions unanswered. Fifty percent of what starting number? During which period? Does “requests” mean submitted forms, unique customers, or messages received? A correct calculation can still produce an unhelpful sentence when the surrounding explanation is missing.
Clear numerical writing gives readers enough context to understand the measurement and the comparison. It also preserves the limits of the evidence. The quantities below are invented for illustration, not results from Nadouri Digital, a customer, or a research study.
Identify what the number counts
Before polishing the wording, write a plain description of the measurement. “Forty requests” might count forty form submissions, even if one person submitted several. “Forty customers” makes a different claim. Confirm the unit before choosing a noun that sounds more impressive.
For a hypothetical help-desk report, “The team received 40 messages about account access in October” is more specific than “Forty users had account problems.” The first statement describes the records. The second assumes each message represents a different person and that every reported problem had the same character.
Keep definitions stable across the article. If the source counts completed forms, do not later call them orders, sales, or customers unless those relationships are established. A change in vocabulary can quietly turn an operational count into a business outcome.
Give a percentage its denominator
A percentage describes a part in relation to a whole. Identify that whole close to the number. “Thirty percent selected email” is incomplete if readers cannot tell who was asked, who answered, or whether respondents could select several options.
In an invented feedback exercise, 18 of 60 respondents selected email as their preferred contact method. A clear sentence is: “Eighteen of the 60 respondents, or 30%, selected email.” It does not say that 30% of all customers prefer email. Respondents and the full customer population are different groups.
Check whether the denominator changes between statements. “Thirty percent of respondents preferred email, and half of customers wanted faster replies” may combine two sources or two populations. Name each group and avoid presenting the results as directly comparable unless the underlying data support that comparison.
Distinguish percent change from percentage points
Percentage points describe the arithmetic difference between two percentages. Eurostat’s definition of percentage points makes this distinction explicit. If a completion rate rises from 20% to 30%, it rises by 10 percentage points.
The relative increase is 50%, because the increase of 10 is half the original rate of 20. Both calculations describe the same change from different perspectives. Calling it “a 10% increase” would describe a different result.
For a general business audience, state the starting and ending values before introducing the derived figure: “The completion rate rose from 20% to 30%, an increase of 10 percentage points.” Add the relative increase only if it helps answer the reader’s question. Stacking both measures into every sentence can add complexity without improving understanding.
For ordinary counts, Eurostat’s percentage-change formula compares the later value with the earlier value. Requests rising from 80 to 120 represent 40 additional requests and a 50% increase. Name the baseline so readers can judge the scale.
If the starting count is zero, the usual percentage-change calculation is undefined because it requires division by zero. Report the change directly: “The pilot received 12 submissions, compared with none in the previous period.” Do not label that movement a 100% increase.
Make the comparison period visible
A comparison needs a defined interval. “Up this month” could mean compared with last month, the same month last year, or the first week of the current month. Those comparisons answer different questions.
Write: “The team received 120 requests in October, compared with 80 in September.” If the data cover only part of October, state the cutoff. Do not compare a complete month with an unfinished month as though their coverage were equal.
Period length also matters. A count for five working days is not directly equivalent to a count for four. You may need a rate, such as requests per working day, if that matches the question. Explain the adjustment and retain the original counts where readers need them to understand the calculation.
Check whether totals and rates tell different stories
A larger count does not automatically mean a larger share. Imagine that 30 of 200 registrations were incomplete in one period and 45 of 300 were incomplete in the next. The incomplete-registration count rose, but the share remained 15% in both periods.
A useful description depends on the reader’s task. A staffing planner may care about the additional 15 cases. An editor evaluating form completion may care that the proportion did not change. Giving both the count and the rate can prevent one question from being mistaken for the other.
Be equally careful when combining percentages. If two groups have different sizes, a simple average of their percentages may not describe the combined group. Return to the underlying counts when possible. If those counts are unavailable, do not manufacture a combined result from the visible percentages alone.
Match precision to the source
A calculator can produce more decimal places than the evidence justifies. “About 40 people” should not become “40.000 participants,” and an approximate source should not become an exact claim merely because it appears in a table.
Choose a rounding approach that preserves the distinction readers need, then apply it consistently. Keep the unrounded values for calculations when available and round the final reported result. Repeated rounding during intermediate steps can change the answer.
Check the wording after rounding. A value rounded to 50% does not necessarily support “exactly half.” Small categories may round to zero even when some observations exist. If that distinction matters, give the count or use a more suitable level of precision rather than allowing the rounded value to imply absence.
Keep uncertainty beside the claim
Estimates, provisional counts, and incomplete records need labels readers will actually see. The Office for Statistics Regulation’s guidance on presenting uncertainty emphasizes making estimates identifiable and placing useful explanations where people encounter the figures.
Be specific about the limitation. “The count includes submissions received by noon and may increase as remaining forms are processed” is more useful than “Interpret with caution.” The explanation tells the reader what is missing and how that affects the number.
Do not supply a confidence interval or margin of error unless the source provides one or a qualified analysis establishes it. An editor can clarify a reported uncertainty measure, but should not create statistical assurances to make a sentence look complete.
Review the sentence without its spreadsheet
Before publishing, read each important number as a reader who cannot see the working file. Can they identify the unit, group, period, comparison, and any material limitation? Can they distinguish a measured result from an invented demonstration? The guide to using examples without unsupported claims is useful for that last check.
Verify the arithmetic separately from the wording. Then compare headlines, captions, and summaries with the qualified statement in the body. Short formats should retain any detail that changes the meaning. The strongest numerical sentence is the one readers can interpret correctly, not the one that displays the largest percentage.
Photo: Aaron Lefler / Unsplash. Illustrative stock photograph.
