Turning the shortcut teachers already used into a hiring marketplace

Teachers bypassed the hiring system and knocked on doors

Districts hire through applicant tracking systems. In the schools we talked to it was principals doing that hiring, not district HR, and they were reading hundreds of applications by hand to find a few worth an interview.

Teachers were not waiting for that process to work.

"I'd go straight to the school office to drop off my resume and try to chat with the principal." - teaching candidate

Nimble was a K-12 specialist competing against entrenched general-purpose systems we could not out-feature. I was its first designer. My PM and I went looking for the place we could win instead, and found it in outbound recruiting, a capability a primary competitor technically offered and had never made its own.

Teachers built 4,822 profiles before the new product existed

Outbound recruiting needs a pool to recruit from, and we had none. A state department of education opened the door. We built them a centralized job board, live hiring dashboards, and a pilot marketplace where educators built talent profiles employers could browse.

Scope was tight, so I worked with a PM and an engineering manager to fold profiles into the existing platform rather than stand up a separate product. That cost us clean login flows and brand cohesion, and it cut the build down to something we could test.

Those profiles took twelve months, with the state's own job board behind the recruiting.

I sat in both research tracks so the findings landed in one head

Customer success talked to educators daily and almost none of it reached the product team. We built the intake together: a standing cadence, one format for what they captured, personas we revised as the evidence arrived. 49 teacher interviews and 44 principal interviews came through it, and I sat in both tracks so the friction points landed in one head instead of two reports.

That moved the target. Principals carry the recruiting load, and they do it without tooling.

"I'll spend 8 hours sorting through applications only to find a handful to interview. I need to increase my yield." - school principal

The prototype round put ten teachers in front of the flow.

A qualified candidate who isn't interested still burns trust

Principals had been burned by services pushing unqualified leads. They did not want more candidates. They wanted fewer, vetted, and already interested.

"Send me the best I haven't seen then I'll reach out." - school principal

That set the requirement I designed against: dual opt-in, so a principal never sees a candidate who is not interested and never spends trust on a dead lead.

Teachers said one thing but did another

Teachers had told us a principal's leadership philosophy mattered most, so I built discovery around the principal. Ten teachers clicked through the prototype and narrated what they were doing.

Almost every one went straight to open jobs, then reached for filters.

Actions over words. Knowing the principal is what makes a job worth taking and an open role is what makes it worth looking. I restructured the information architecture around job discovery and moved leadership and culture inside the job itself, where it answers a question the teacher has already asked. I led with filters for the same reason. Until we had the data to recommend properly, teachers narrowed their own search rather than being handed suggestions they had already learned to distrust.

The discovery page after the restructure. Jobs come first, with the filters above them.

I split the principal experience across the phone and the desk

The MVP targeted two Missouri metros, St. Louis and Kansas City. Teachers got a self-serve responsive platform. Principals got an email, and a profile waiting behind the tap.

The email had to reach them on a phone, because a good candidate is gone in days. Some principals were making verbal offers before anyone reached the ATS. But nobody evaluates a teacher from a phone between classes. That is desk work.

The same match alert on desktop and phone, walked through with a principal.

So the email carried pre-screened candidates and their contact details, enough to act on without opening anything, and the full profile sat one tap behind. I put both action models in front of ten principals, side by side.

The same match email, two action rows. One routes back to a full candidate list; the other lets them contact or skip without leaving the inbox.

We built that second surface no-code, which kept the engineering spend waiting on evidence the matches were any good.

The mechanic I trusted most is the one I would measure first

The team shipped the MVP in March 2024, a month after I handed off the spec.

Dual opt-in is what earns a principal's trust, and it is also a throughput tax: every match needs two yeses, and the second one is where a marketplace like this breaks. That handshake is what I would instrument before opening a second market. I would rather ship a mechanic whose failure I can see than one I have to trust.