A Case Study With No Stated Duration Has No Rate

You have two case studies open and you are trying to decide which result is stronger. One says the campaign ran from March to July. The other gives you a large percentage and nothing anywhere about how long it took. You cannot rank them. Not because the second number is dishonest, but because it is missing the input that would let you size it.

The number without a clock

A percentage, a lift, or a total is only half a claim. The other half is the period it happened over. A case study reporting a 40 percent increase in qualified leads sounds complete and is not: “increase” describes a change between two moments and never says how far apart they were. Forty percent in a month and 40 percent in three years are the same number attached to entirely different work.

Duration is not a detail for the methodology note. It is the denominator. Without it the figure stays a raw total, and raw totals cannot be compared unless the periods behind them match, which you cannot check. Calling a number fast, slow, or ordinary means converting it to a rate first. It is the distinction that lets cumulative totals rise while the rate falls. Here the rate is not falling. It does not exist yet, because nobody supplied the time.

What a stated duration lets you do that a raw total does not

Here is a worked example, entirely hypothetical. It describes no real agency, no real brand, and no case study published anywhere.

Picture a case study reporting a 40 percent increase in monthly sessions from social, and run it twice. In the first version the campaign ran three weeks, so 40 percent of the starting base arrived across 21 days, roughly 13 percent of that base per week. In the second, the same 40 percent accumulated over three quarters, about 39 weeks, or roughly 1 percent per week. Same headline, thirteen times slower.

Compounding rather than dividing moves the figures, roughly 62 percent a month against roughly 3.8 percent a month, and the conclusion not at all. One is a sprint, one is a drift, and neither reading is available to you until the duration is stated. Both versions are invented for illustration and are not anyone’s reported result.

Where a duration is normally disclosed, and where it quietly goes missing

Timing usually appears in one of a few places:

• An explicit start and end date, often in a summary box or the opening paragraph.
• A phrase in the copy: “over eight weeks”, “across the first quarter of the engagement”.
• A chart x-axis labelled with dates or week numbers.
• A dated screenshot with the date range selector visible rather than cropped out.
• A methodology note or footnote, where timing lives when the headline is written for skimming.

The trap is the chart with an unlabelled x-axis. A climbing curve feels like evidence and says nothing about pace, because you cannot tell whether it covers ten days or ten months. That is a duration omission in a different costume, easy to miss because a chart looks like disclosure. It helps to know how to read a results chart before the caption.

This is not the same failure as a disclosed window that does not match the headline

A related problem gets confused with this one. Sometimes a case study does state a window, and it sits awkwardly against the number paired with it, covering a different stretch than the headline describes. That is a disclosed but misleading duration, and you can work on it, because the window is on the page. You can ask why that slice, or note that the period was a choice. In the same family, the attribution window is a choice, and a disclosed choice is one you can argue with.

A missing duration gives you nothing to argue with. No window to interrogate, no slice to question. The first is a claim you can push on, the second one you cannot get hold of.

What to do when a case study never says how long it ran

Work through this before giving up on the number:

1. Reread the page, captions and footnotes included. Timing is often stated once, quietly, far from the headline.
2. Check charts and screenshots for axis labels or a visible date range.
3. Look for a methodology note, a PDF version, or an award entry for the same work, which often carries detail the summary drops.
4. Ask. Emailing the agency or the brand for the campaign window is an ordinary editorial question, not an accusation, and reluctance to answer is itself information.

If none of that produces a window, take the fallback: treat the number as an anecdote rather than a comparable result. It may be true. It simply cannot be placed on a scale with anything else, and pretending otherwise is how a number with no denominator ends up quoted in a deck as a benchmark.

A missing duration is not a red flag, it is a missing input

Back to the two case studies. The one that named its window can be turned into a rate, set beside other results and ranked. The untimed one cannot be ranked at all. Not low, not suspicious, simply not ranked, because ranking needs a rate and there is none. Set it aside. If the duration surfaces later, the number becomes usable that day.

Read more results, and send us yours

Every entry here is another agency’s published work, credited to the agency that reported it. Browse the case study library and count how many name their campaign window, or submit your own.

FAQ

Does every case study need to state an exact start and end date?

No. It needs enough for a duration to be recovered. “Over eight weeks” does the job, and so does a dated screenshot or a labelled time axis. The requirement is that some window is findable, not that a calendar date appears in print.

Is a case study with no stated duration automatically fake?

No, and it is worth resisting that jump. Duration goes missing for dull reasons more often than dishonest ones. The useful conclusion is narrower: the number cannot be sized, so set it aside as unratable rather than treating it as evidence in either direction.

Sources

• Cumulative Totals Can Rise While the Rate Falls, fetched 2026-09-06.
• How to Read a Results Chart Before You Read the Caption, fetched 2026-09-06.
• The attribution window is a choice, and it is rarely disclosed, fetched 2026-09-06.