AI Interviewing and Disposition Data

Happy Friday, Job Board Doctor friends! This Doc is beaming this morning after an excellent set of sessions provided by the dynamic duo of Louise Triance and Louise Grant, the ladies leading Job Boards Connect at their Unplugged event this week in London.

My takeaway: There are a lot of niche job boards doing really important and good work. They are bringing energy, ideas and dedication to their communities to the table.

I was also totally (seriously) humbled with not one, but two DJAX awards in the evening gala, the Community Champion Award and the Outstanding Impact in a Niche Market for an Established Platform, which was awarded to Disability Solutions.

For me, the highlight was the DJAX Community Champion Award. I was nominated along with Steven Rothberg, Patrick Toet, and Martin Lenz.

I am not one to get sappy about an award, but this one coming from my peers in the industry means a lot more than I will express eloquently here. I am grateful.

Now let’s get to it.

Greenhouse Acquires Ezra AI Labs

On the candidate side, Greenhouse’s 2026 Candidate AI Interview Report says 63% of US job seekers have now been on the receiving end of an AI interview, up thirteen percentage points in six months. Of those candidates, 38% have already withdrawn from a hiring process because it included an AI interview, and another 12% say they will.

Seventy percent were never told upfront that AI would be evaluating them. Twenty-one percent only discovered AI was in the room once the interview started. Of the candidates who completed an AI interview, 51% never heard back at all. Not a rejection, not an advance, just silence.

Worth pausing on who is publishing this report. Greenhouse released the candidate experience data on May 1. Five days later, on May 6, they announced they had signed a definitive agreement to acquire Ezra AI Labs, a voice-AI interviewing platform, with the explicit pitch that it would bring “conversational AI to the hiring process.”

CEO Daniel Chait’s own quote in the report was that “most AI in hiring today is making a bad system worse: more applications, less signal, and less transparency.” Greenhouse is now in the business of selling you their version of the thing they just told you isn’t working.

When the vendor publishing damning data on AI interviewing is the vendor about to sell you a new AI interviewing product, the industry should probably take a second look at we are doing.

ZIP gives away the disposition data game

On the employer side, the AI matching pitch keeps getting louder. The argument runs that once a job board plugs into your ATS and feeds disposition  and status data back into its matching engine, the platform learns which candidates actually advanced, refines its targeting, and starts delivering fewer, better applications.

Less noise. Better “fit”. 

In case you didn’t know it, and honestly I did not until last week, I have some news. ZipRecruiter has been collecting ATS disposition data through integrations like one with Greenhouse for years. The Greenhouse documentation is explicit: candidate stage and application status flow back into ZipRecruiter to feed its Smart Matching engine.

ZipRecruiter has had access to the exact data Indeed sales is telling you will turn it into the ONLY platform you need.

Well….ZipRecruiter’s Q1 2026 8-K, filed with the SEC on May 7, just told us what years of that data flow actually produced.

Revenue: Down 2% year over year. Net loss: $4.7 million. Quarterly Paid Employers: Flat year over year. Revenue per Paid Employer: Down 2% year over year.

If years of disposition data feeding a matching engine were the magic beans Indeed is selling, assumedly ZipRecruiter would have bent the curve on at least one of those lines.

They have not. Paid employer count is flat. The marketplace did not consolidate around them, and ZIP has not become the platform every TA leader is throwing budget in to.

Here is where ZIP gave the game away: In their quarterly earnings report, the product win they credited to the AI matching engine and “Be Seen First” feature in Q1 was a 37% increase in applications.

If matching were actually getting smarter in the way the end-to-end pitch demands, the brag metric would be the inverse.

In theory, because it is still theory, better matching should produce fewer, more qualified applications per requisition. That is the entire economic case for handing your disposition data to a job board.

The TA leader sitting in the Indeed pitch meeting is already drowning in application volume. The pitch claims disposition data fixes that, but it doesn’t.

While ZipRecruiter has been demonstrating in market that the disposition-data thesis does not pencil out, Indeed has been running a different play on the employer side. Employer spend per Indeed job is rising. The argument employers are getting for that rising spend is “we’ll be your end-to-end hiring solution if you connect your ATS.” The argument is being made by the company that drove their hiring costs down for years and is now driving them back up, with extra services bundled in.

Enter LinkedIn Apply Connect & RSC

Now, LinkedIn enters the ATS integration and disposition data room with Apply Connect and Recruiter System Connect (RSC).

Is there anyone in this industry who can come up with an original product name?!?!

LinkedIn’s tool is being sold as unified applicant management as you can see below.

For me at least, this is a developing story from LinkedIn so expect more in the coming weeks and tell me what you know so I can share with the our community.

Recruit Holdings Earnings Out Today

This morning, May 15th, Recruit Holdings released record-high financial results for FY2025, which ended on March 31, 2026.

For me, this quote says it.

Junichi Arai, Executive Vice President and Chief Financial Officer, Recruit Holdings

“As Deko mentioned earlier, over the medium term, it is well within our reach to not only maintain double-digit annual revenue growth, but to achieve 20% or greater when hiring demand recovers. I also believe it is well within our reach, and at that time, we expect our EBITDA+S margin to exceed 50%. In the U.S. in FY 2025, amid stagnant hiring demand and while the total number of U.S. job postings on Indeed declined by approximately 7% year-over-year, monetization development resulted in revenue of $5.31 billion, an 8.8% year-over-year increase on a dollar basis with a U.S. ARPJ growth rate of 17%.” (earnings call transcript via investing.com)

Although, I am sure I will have more to say about it next week. 🙂

Let’s Put it All Together

Better signal makes matching better. Nobody is arguing otherwise. The question that should be live for every TA buyer is whether disposition is the right signal in the first place.

ZipRecruiter has now spent years feeding stage and status data into its matching engine and the result, by their own disclosure, is more applications per posting and flat employer growth.

That is the opposite of what better signal is supposed to produce. Disposition data may be telling these matching engines who got hired. It is clearly not telling them anything useful about who should have.

Hand your data over if you want. It will not fix your hiring.

Applications are up. ZipRecruiter just printed 37%. The candidate side has been weaponized by AI-assisted apply tools and AI-generated resumes optimized for ATS filters, and recruiter sunder 50 reqs is buried.

So when the marketplace sales reps call and offer to make your hiring smarter if you’ll just connect your ATS and route disposition data back to them, the question to put back is one sentence long: how did that work out for ZipRecruiter?

They have been running the experiment for years. The results are on file with the SEC.

Tell me what you think.

Until Next Time,

Julie “The Doc” Sowash

[Want to get Job Board Doctor posts via email? Subscribe here.]

[Got a tip, document or intel you want to share with the Doc? Tell me. Tip so hot you need it to be encrypted? Use Signal.]