Salesforce moved the number everyone still quotes at you

Salesforce publishes a research report called State of Sales. Its sixth edition says sales reps spend about 30 percent of their time actually selling. The seventh edition says 40 percent, in a report with 4,050 respondents fielded August to September 2025 (Salesforce, State of Sales 7th edition, https://www.salesforce.com/en-us/wp-content/uploads/sites/4/documents/reports/sales/salesforce-state-of-sales-report-2026.pdf).

Same publisher. Same report series. Same question. Ten points.

The 30 percent version is the one that got into the culture. It is the slide that opens a hundred pitch decks, the line in the LinkedIn post, the premise of an entire category of software. The 40 percent version has been sitting on Salesforce's own domain and almost nobody has updated the slide.

Nothing about the working day changed

The interesting part is not that the number moved. It is why.

The sixth edition was fielded 8 March to 18 April 2024 with 5,500 respondents; the seventh was fielded August to September 2025 with 4,050, both figures taken from the editions' own methodology notes. Different year, different panel, different sample size. That alone would move a number by a few points.

But the larger mover is definitional: the seventh edition counts prospecting as nonselling time. That is a survey-design choice, and it is a big one. If prospecting is selling, you get one number. If prospecting is overhead, you get another. A rep whose Tuesday is identical in both years shows up on different sides of the line depending on which edition you read.

And underneath both editions is the same instrument: people typing an estimate of how they spend their week into a survey form, on a paid third-party research panel. Nobody observed a calendar. Nobody instrumented a CRM. It is a self-report about time, which is the category of self-report humans are worst at.

That does not make it worthless. It makes it a survey result, which is a different kind of object from a measurement, and it should be quoted as one. "Reps say they spend about 40 percent of their time selling, when the survey defines prospecting as nonselling" is a sentence you can defend. "Reps only sell 30 percent of the time" is a sentence about the physical world that nobody measured.

The same shape, twice more

Once you know what to look for, the pattern repeats across the most-quoted numbers in B2B sales.

Gartner published a figure that buyers spend about 17 percent of their time meeting with suppliers. It appears in the 2021 and 2022 Internet Archive captures of gartner.com/en/sales/insights/b2b-buying-journey, and it is absent from the January 2024 capture of the same URL. We are relaying that check rather than re-running it: it was verified by a research stream working on our sourcing file, and the live page returns 403 to non-browser clients, so you may find it awkward to confirm yourself. The figure kept circulating for years after the page stopped carrying it. The companion "six to 10 decision makers" number went the same way, and citers routinely drop its scope, which was specific to a complex B2B solution.

Then there is the most-quoted number in the category, from CEB research published around 2012: that B2B buyers are 57 percent of the way through the purchase before they contact a supplier. The CEB deck's own footnote reports n=1,399 from the "MLC 2012 Customer Purchase Decision-Making survey", while the report prose describes more than 1,500 contacts across 22 large B2B organizations. Those are both CEB's numbers, in the same document, and they do not agree. That check is also relayed rather than re-run here. The survey instrument has never been produced by anyone in the fourteen years since. A trace of the related 70 percent variant (https://www.jvminc.com/Clients/JVP/57percent.pdf) found it began as a single member's anecdote posted on a CEB blog before the survey was fielded at all.

LinkedIn's own sales blog now calls the 57 percent a myth (https://www.linkedin.com/business/sales/blog/b2b-sales/this-popular-stat-is-wasting-your-time--the-57--engagement-myth), which is a strange sentence to write about a statistic still being used to justify content budgets.

A number that moves while the advice stays put was never doing the work

Here is the tell that generalises beyond any one figure.

The same assertion about buyers being deep into the journey before contacting sales has been published as 57 percent, then roughly 70 percent (Forrester, 2019), then roughly 80 percent (Gartner, 2024), per a critique at https://www.inflexion-point.com/blog/the-b2b-buying-decision-process-challenging-the-57-myth. On those three published figures, the number rose by 23 points across a decade. The recommendation drawn from it did not change at all.

That is diagnostic. If a figure can move 23 points without disturbing the conclusion it supposedly supports, the figure was never load-bearing. It was decoration on an argument that was being made anyway. The same is true of the Salesforce selling-time number: nobody who quoted 30 percent has revised their strategy now that the publisher says 40.

Which raises the honest question. If the advice does not depend on the number, why is the number in the deck? Usually because a figure makes an assertion feel like a finding.

Three checks before a statistic goes in your deck

None of these take long.

Find the edition. Numbers from serial publications have version numbers. State of Sales is on its seventh. Anything you quote from a recurring industry report should carry the edition and the field dates, because those are the two things most likely to have moved since you first saw it.

Read the footnote, not the headline. Sample size, field window, panel source, and the definition of the thing being counted. The Salesforce figure moves ten points largely on the definition of prospecting. The CEB figure has two incompatible sample descriptions in the same document. Both of those are visible to anyone who opens the source.

Ask what the number would have to be for your advice to change. If the answer is "nothing would change", cut the number. You are not using it as evidence, and carrying it exposes you to being wrong about something you did not need.

We ran the third check on ourselves

Our sourcing review flagged the 28 to 30 percent figure as still in wide circulation, and noted that if it appeared anywhere in our own material it would be a checkable error on a site whose whole argument is that its claims survive checking. That impression of wide circulation is ours and we did not enumerate the pages carrying it, so treat it as an impression rather than a count.

So we searched. At the deploy commit, on the branch production actually runs, a search for "28 percent", "30 percent", "28%" and "30%" across every marketing page returns four hits, and all four are CSS gradient positions. A positive control on the same search form returns matches for "499" across five pricing files, so the search works and the absence is real rather than a broken pattern.

One limit, stated because it matters more than the result. Our marketing copy is served from a CMS at request time, with the string compiled into each component acting as a fallback. So a repository grep tells you what the default says, not what a visitor sees. The command that would settle it is a fetch of the live page. The clean grep is evidence, not proof, and we would rather say that than let a green check stand in for a check we did not run.

That distinction is the whole point of this article. A number is only as good as the document behind it, and a check is only as good as the thing it actually read.