A short piece of fiction, written by API Evangelist. The reader is imagined; the post, and every number and quotation from it, is real: "Forty-Four Percent Of Job Postings Said Workday And The Word Was Ours," Kin Lane, November 11, 2026, https://apievangelist.com/2026/11/11/forty-four-percent-of-job-postings-said-workday-and-the-word-was-ours/
Marisol ran data for a staffing firm's research desk, and she had a slide due Friday that said Workday was in 44 percent of job postings. She'd pulled the number from a vendor dashboard two weeks earlier and liked it because it was round and surprising.
Then she read the post, and it began by saying that most of the wrong numbers were wrong "because of us." She read that line twice. Her own team never said anything like it.
The mechanism was almost funny. The harvester wrote a small header into every posting, and for many employers that header read "Source: Workday CXS API". The matcher then counted the word as if an employer had written it. The author's summary of the damage was plain: "We were measuring which companies we happened to pull through Workday." Once the four header lines were stripped, the share fell from 44% to 3.9%.
Marisol felt her stomach drop a little. Her slide used a figure that looked a lot like that first one. She had no idea whether her vendor's pipeline stripped anything, and she had never asked.
Her first reaction was irritation, and it was aimed at the post. It was tidy in a way she distrusted. A team that finds its own error and then announces that the fix made things more reliable is also a team asking to be trusted more, and she noticed herself resisting that. She wrote in her notes that a confession can be a marketing move too. But she couldn't make the objection hold up. The post named the bad numbers. It gave counts for teams, office and project. It admitted that a maritime term had made it into one of their tables at 45 companies before anyone noticed it was air-insulated switchgear. People who wanted to look good would have left that out.
What bothered her more was the part about precision. The post said a clean match doesn't make a count believable, because "Use Cases" matched in 507 companies and was not a technology anyone was hiring for. She had always treated a clean match as a good one. She thought of three charts in her last quarterly deck and felt slightly sick.
She also disagreed with something. The post says it would rather publish its pipeline's failures than have readers find them. She thought that was admirable, and also easy to say when your readers can't check. Nobody she worked with would have the patience to read a post like this one. They'd want the number.
She still didn't take the slide to her manager as a problem. It was a Wednesday afternoon, and she was fairly sure she was overreacting.
What she did was small. She opened the vendor's methodology page, found nothing about headers, and wrote a two-line email to her account contact asking whether the tool removed its own source text before matching terms. Then she moved the Workday figure off the slide's headline and into a footnote that said the count was unverified.