When the Summary and the Body Give Two Different Numbers

When the Summary and the Body Give Two Different Numbers

You are reading a case study and the summary card says one number. A few paragraphs later, the body text restates what looks like the same result, and the figure is different. Not a different metric wearing the same name, the actual number attached to the actual same metric has changed between the top of the page and the middle of it.

The two numbers are in the same document

This is a narrower problem than the more familiar one where a case study swaps metrics mid read, engagement rate in the headline and reach by the third paragraph. That is a label switch, and the label is doing the misleading. Here the label never moves. The summary card says “engagement rate” and the body says “engagement rate,” and both are talking about the same account over what appears to be the same campaign. The only thing that changed is the figure sitting next to that unchanged label.

There are three ordinary explanations, and none of them require bad faith. The first is rounding: a summary card built for scanability often rounds to a clean number, while the body text, written by someone pulling from a spreadsheet, keeps the decimal. The second is measurement moment: a summary card is frequently written last, after the campaign closed, while a body paragraph might have been drafted from a mid campaign dashboard pull and never updated. The third is period boundary: the body sometimes includes a few trailing days that the summary’s stated window excludes, or the reverse, a partial week that nudges the number in one direction without anyone flagging the change.

A hypothetical case study, read twice

Picture a hypothetical case study for an unnamed home goods brand. Every figure below is invented for illustration and does not describe a real reported result.

The summary card at the top of this hypothetical page reads: “Engagement rate rose to 6.8 percent across the campaign.” Clean, rounded, the kind of number that belongs on a card meant to be skimmed.

Three paragraphs into the body of the same hypothetical case study, a sentence reads: “By the close of the promotional window, engagement rate reached 6.4 percent, before ticking up slightly in the days following the final post.” The label is identical, engagement rate, for the identical account and campaign. The figure is not. Read only the card and you would carry away 6.8. Read only that one sentence in the body and you would carry away 6.4.

Nothing here is presented as a real reported result. It is a hypothetical worked example, built only to show what the mismatch looks like when you are staring at it, and the actual explanation, in this invented case, is the third ordinary cause: the card’s 6.8 folds in a few extra days after the promotional window closed, while the body’s 6.4 is the number as measured at the window’s actual end.

Why this check is faster than the others in this library

Most of the honesty checks worth running on a case study need something the case study itself does not provide: an attribution window check needs the platform’s default lookback, a baseline check needs what the account looked like before the campaign, a benchmark check needs a credible outside figure that actually applies to this account’s size and category.

This check needs none of that. If the same label carries two different numbers on the same page, that is a fact available from the page alone, no documentation, no second source, no external lookup. The article already open in front of you is the entire dataset, and it is also the one check readers skip: nobody expects a document to disagree with itself.

Three questions to ask when the two numbers do not match

  • Is one figure rounded and the other exact, and if so, does the rounding land in the direction that makes the headline look better? A card that rounds 6.4 up to 6.8 is doing something different from one that rounds 6.42 to 6.4.
  • Were the two figures measured at different moments? A number stated as of campaign end is a different claim from one stated as of the day the case study was published, even if both describe the same metric on the same account.
  • Does one figure cover a longer or shorter period than the other? Check specifically whether the body includes a partial trailing period, days after the campaign’s stated end date, that the summary’s stated window does not.

When the mismatch is worth flagging, and when it is not

A gap of a few tenths of a point that traces back to ordinary rounding is not a red flag. Numbers get rounded for a card meant to be read in two seconds, and rounding in either direction is unremarkable on its own.

What earns more scrutiny is a pattern: a summary figure that is consistently the more favourable of the two, across multiple case studies from the same source, stops looking like noise and starts looking like a habit.

None of this claims any specific case study you have read does this. This piece describes a check, not an instance, so you know which questions to ask when you find a same document mismatch.

Check the case study library, or add to it

You can browse the case study library and run this same check yourself against a real summary card and body pair. If your agency has a published result worth adding to the shelf, you can submit your own.

FAQ

Is a mismatched number always a sign the case study is misleading?

No. Rounding and different measurement moments are ordinary explanations and usually benign. The point of this check is to notice the gap and account for it, not to assume the mismatch means bad faith.

How is this different from a case study that changes the metric name between the headline and the body?

They are separate errors. A metric name change, engagement rate in the headline and reach in the body, swaps what is being measured. This check covers something narrower: the label stays constant throughout, engagement rate stays engagement rate, and only the figure attached to that unchanged label differs.

Where should I look first when checking this?

Start with the summary card or the pull quote figure, then find the nearest place in the running body text or a chart caption where that same metric is restated, and compare the two numbers directly.

Similar Posts

Leave a Reply

Your email address will not be published. Required fields are marked *