Cumulative Totals Can Rise While the Rate Falls
A case study opens with ten million impressions since the campaign began, and you feel two things at once: this was big, and this is working. Only the first is supported by the number. A lifetime total is a scale claim wearing the clothes of a momentum claim. Separating the two takes one question.
The number that cannot go down
A cumulative total is a running sum. Each period is added to everything before it, and since impressions, followers and downloads cannot be negative, the sum can only rise. “Up and to the right” is a property of addition, not a finding about the campaign.
The chart version of this is covered already: if a plotted line is in front of you, read a results chart before the caption. Here there is no chart at all. The total sits in a sentence, “10 million impressions since the campaign began,” and you read it as proof the campaign is working now. There is no curve to inspect for flattening, so what is left is why cumulative is the default, how long “since launch” really is, and what a run of lifetime figures from one source tells you.
A hypothetical: the same campaign, two charts
The numbers below are hypothetical, invented for this article to show the arithmetic. They are not a real reported result and belong to no agency, brand or campaign. Picture a four month campaign where monthly impressions fall by about 18 percent each month after the first.
| Month | Impressions that month | Change on prior month | Cumulative since launch |
|---|---|---|---|
| 1 | 3,300,000 | n/a | 3,300,000 |
| 2 | 2,700,000 | down about 18 percent | 6,000,000 |
| 3 | 2,200,000 | down about 18 percent | 8,200,000 |
| 4 | 1,800,000 | down about 18 percent | 10,000,000 |
Chart one plots the last column: it climbs every month and lands on ten million. Chart two plots the second column: it falls every month and ends a bit over half where it started. Both are accurate descriptions of the same campaign, built from the identical four rows. The only difference is which question each answers: how much has this produced in total, or how much is it producing now against before. A case study showing only the first answers a question you were not asking, and feels like it answered the one you were.
Why case studies default to the cumulative version
Resist the assumption that this is always a trick. The cumulative total is the path of least resistance: one dashboard pull at the moment of writing, rather than a monthly series someone has to assemble and check. It never looks bad, so it never needs a paragraph explaining a dip with an ordinary cause like a paused budget. It is the habit, not necessarily the decision.
And a large total is not meaningless. It establishes that the work reached real scale and ran long enough to accumulate something. The failure is not the number, it is what a reader infers when no rate appears near it.
The one question that separates the two claims
Ask this: what did this look like in the most recent comparable period, on its own, not added to everything before it?
A satisfying answer is specific and bounded: one period’s figure, say “last month delivered 2.1 million impressions,” or a rate against a comparison, “the most recent 30 days were down 12 percent on the previous 30.” It may be worse than the total implies, and that is fine. A source that volunteers a soft recent month next to a strong lifetime figure is showing you it can tell the two claims apart, which is the competence you are trying to detect.
The red flag is not one case study that leads with a total. It is a pattern: every result from the same source given as a lifetime figure and never broken into periods. A consistent absence of any rate is a decision about what you get to see. The neighbouring trap is why the best metric on a dashboard is rarely the real story.
Where else this shows up
The same shape recurs well outside social reporting:
- Software reporting: “total users” against “active users.” Total users counts everyone who ever registered, including everyone who left; active users in a defined window is the rate. A product losing people every month still adds to its total users figure.
- App distribution: “total downloads” against “weekly downloads.” Lifetime downloads include installs deleted the same afternoon three years ago. Weekly downloads tell you whether anyone is arriving now.
A short checklist for reading a lifetime number
- Does a rate or recent-period figure appear alongside the total? If not, treat the total as a claim about scale only.
- How long does “since launch” cover? Ten million over six weeks and over four years are different campaigns, and a total with no stated duration is unreadable.
- Could the total be rising mainly because the measurement window keeps getting longer? Every month that passes adds to it by default.
- Is the curve flattening while the axis scaling keeps it looking steep? For the next distortion to check, see how a percentage change can be expressed to mislead.
Try it on a published result
Browse the case study library and check, for each result, whether the headline figure comes with a stated period or stands alone as a lifetime number. Every figure there belongs to the agency that reported it, so read each as that agency’s claim about its own work. If you have published a case study, you can submit your own.
FAQ
Does a rising cumulative total mean the campaign is accelerating?
No. A cumulative total rises whenever each period contributes any positive number, including a contribution smaller than the one before it. Rising and accelerating are different claims, and only a rate or period-over-period figure supports the second.
Is it dishonest for a case study to report a lifetime total?
Not inherently. A lifetime total is a true number and it says something real about scale and duration. The problem sits in reader inference rather than in the reporting, so the fix is asking for the rate alongside it rather than assuming the total was chosen to mislead.
What is the fastest way to check if a total is masking a decline?
Ask for the most recent single period’s figure on its own, not the running sum. If the source cannot or will not supply it, treat the total as a scale claim only and not a momentum claim.
Sources
- Social Case Studies, case study library index, fetched 2 September 2026.
