The concrete version of this product is a Tuesday alert saying a driver's licence expired over the weekend on someone rostered for that afternoon. That is founder and CEO Andrew Johnson's own example: if a licence lapses the day after someone is hired, the employer should know that day rather than at the next annual review, which he called the difference between a photograph and a live video feed [11]. Read as plumbing it is modest and useful. Re-run a check against a registry, compare the result with the stored one, raise the difference. The framing around it reaches further, with SiliconANGLE's account describing a platform that keeps tabs on employees for as long as they work at the company and gives employers greater visibility into what they are up to [21]. Those are two products on one invoice, and the one that survives contact with a manager's inbox is the licence.
The customer list says who this was built for. Yardstik's biggest accounts include TaskRabbit, Gopuff and Liveops [16], and at Liveops the platform flagged 10% to 15% of applicants as risks for using fake identities or credentials [17]. Against the company's own stated baseline of one in 30 applicants failing a government identity check [12], or 3.3% [19], Liveops runs three to four and a half times higher [18]. Applicant identity fraud is concentrated in high-volume remote contractor hiring rather than spread evenly across payrolls, so the pre-hire numbers do less work for a 300-person firm with stable headcount, and the post-hire monitoring has to carry that sale on its own.
The figure doing the emotional work in the announcement comes from the Association of Certified Fraud Examiners: the average occupational fraud scheme runs about a year before detection [5]. Refreshing credential and identity records does not obviously shorten that year. What refreshes is registry data, so the person whose licence lapsed or whose driving record changed appears; the person quietly approving invoices to a company they own is in no registry. Johnson's founding case, employees at Branch Microfinance Bank working under stolen Social Security numbers until the real owners started receiving tax bills [20], is an identity problem, and identity is the part of this stack with the cleanest signal.
Yardstik reports 149% revenue growth over the past year, 98% customer retention and a 99.4% satisfaction rating [15]. Retention at that level tells you renewals happen; it does not tell you how many flags became a decision an employer could defend six months later. The volume figure it does publish, roughly 2,900 incidents a year of applicants using a flagged or duplicate Social Security number [13], works out at about eight a day across the whole base [14], and it is applicant-side. There is no published post-hire flag rate, which is the number a buyer needs in order to staff the queue.
Two axes settle this for most teams. First, does the signal map to a requirement of the job that can expire, such as a licence or a driving record, or is it an inference about the person? Second, is there a written action for that signal, with a named owner and a step for telling the worker, before the first alert fires? Job-linked signal plus written action is renewal plumbing, and worth automating. Job-linked with no action is an alert log that proves you knew. Inferential signal with an action attached is automated suspicion. Inferential with no action is a dashboard someone will stop opening. The announcement covers the enrolment default and the aggregated view [7][8] and says nothing about worker notification or dispute handling [22], so that policy is the buyer's to write, and writing it is the real cost of moving from a photograph to a live feed.
