How hiring processes create the scarcity they measure

That's a strange thing to claim. Recruiters pull real data from applicant tracking systems. They track pipeline conversion rates, measure where candidates come from, and compare how different sources perform. The numbers are accurate. But where the numbers come from is a question most teams never think to ask.

In the 1960s, American cities built wider highways to reduce congestion. More lanes, more capacity, shorter commutes. Except that's not what happened. Every new lane attracted drivers who had previously taken other routes or stayed home. Within a few years, congestion returned to roughly where it started. The highway planners pointed to the traffic counts and said we need even more lanes. They were measuring a problem their own solution had produced.

Transportation planners now call it "induced demand". The road creates the traffic.

Recruiters do the same thing. Karl Weick, who spent his career studying how organizations construct their own realities, called it "enactment" in his 1995 book Sensemaking in Organizations. People don't just observe their environment and react to what they find. They produce part of it through their own choices, then treat what they've produced as an independent fact about the world.

In recruiting, you see it most clearly in sourcing. A recruiter who runs a search exclusively through LinkedIn and one job board produces a pipeline that reflects those channels. When the pipeline comes back thin, they report that the market is extremely competitive. And in a narrow sense, they're right. The pipeline really is thin. But a different recruiter working the same role through professional communities, academic networks, and employee referrals might find something different. The scarcity was real enough to shape decisions. It was also, in large part, created by the decisions that preceded it.

Job ad language works the same way. Twelve mandatory requirements filter out people who take requirements at face value (a skew that research links to gender and cultural background), and the recruiter concludes the remaining pool lacks diversity without seeing that the ad shaped who showed up.

The tricky part about enactment is that it's invisible to the person doing it. The feedback loop between "I chose these channels" and "the market looks like this" is long enough that cause and effect feel like separate events.

Whatever pool survives the sourcing funnel then runs into a second force. In 1968, two psychologists told a group of elementary school teachers that certain randomly selected students would show surprising intellectual gains over the coming year. By year's end, those students had actually gained more. The teachers, believing the prediction, had given them warmer feedback, harder problems, and more patience. The students responded to better treatment with better performance. The teachers then pointed to the results as proof that their original assessment had been correct. Robert Rosenthal and Lenore Jacobson called this the self-fulfilling prophecy.

The same dynamic runs in every unstructured interview. When a hiring manager sees a strong resume before the conversation (familiar company, right title, internal referral), they walk into the room already leaning forward. They ask follow-ups that let the candidate show their best thinking. Ambiguous answers get generous readings. The candidate, picking up on the warmth, relaxes and performs well. The interviewer leaves thinking she was excellent and genuinely believes they observed this rather than co-produced it.

You've probably seen the inverse. A candidate with a weaker signal on paper gets a cooler reception. The interviewer asks tougher questions, leaves less silence. The candidate reads the room instantly, the way anyone reads a room where they sense they're not quite welcome. They tighten up. Their answers get shorter. In the debrief, the interviewer reports what they saw. Solid, but not quite there. They're partly describing their own behavior reflected back at them.

Each one alone skews things in a predictable direction. Together they form a closed loop, and this is the part I find most interesting.

Enactment narrows the pool at the top. Sourcing choices and ad language determine who enters the pipeline. The self-fulfilling prophecy then works on whoever makes it through, confirming expectations the narrowed pool already established. A candidate sourced from an unfamiliar channel arrives carrying a weaker signal. The interviewer treats them accordingly. And the candidate responds in kind. The debrief produces a data point that says this channel yields weaker candidates.

That data point feeds back into the next sourcing cycle. The channel gets deprioritized. The sourcing narrows further. The next round comes from the same familiar sources, carrying the same familiar signals, receiving the same warm reception. The pipeline data now shows a clear pattern. Candidates from the usual channels outperform candidates from everywhere else. The pattern is real. The interpretation (that the usual channels find better people) is where the error lives. The usual channels find people who get treated better.

The wall gets taller with each pass. And every brick comes with a data point that makes it look structural.

Successful hires from familiar channels get credited to the channel, not to the favorable conditions the process created for them. And you can't prove the loop exists using the data it generates. Pipeline metrics, interview scores, source-of-hire analysis. All produced by the process you're trying to evaluate. "Why would we invest in channels that our data shows produce weaker candidates?" is a perfectly rational question if you don't see the loop.

This is where most experiments die, and understanding why might save you from drawing the wrong conclusion. A recruiter who diversifies sourcing but works within a team that still runs unstructured interviews will see their new-channel candidates underperform. The prophecy is still running on the interview side. They'll eventually conclude, with supporting data, that the new channels don't work. The system produces confirming evidence for whatever you're already doing.

Both ends have to move at once. Start upstream. When a search is producing poor results, change one sourcing variable and observe whether the "market" changes. If it does, the original constraint was at least partly self-created.

The interview end probably matters more, because it's where the loop closes. There's a study from the early 90s where researchers told one group of interviewers that their job was to get to know the person, and told another group that their job was to form the most accurate assessment possible. The prophecy effect disappeared in the accuracy group. One sentence in a pre-interview brief can make this shift. "Your task today is to form the most accurate assessment you can of this candidate's capabilities." That sounds too simple for how deep the problem runs, but the research on it is surprisingly clear. It changes what the interviewer is optimizing for. Instead of looking for signals that confirm a prior impression, they start testing it.

Some talent shortages are genuinely structural, and it's worth not pushing this too far. When an entire industry faces a skills gap driven by demographic shifts or training pipelines that can't keep pace with demand, no amount of channel diversification will fix the problem. The recruiter didn't create that shortage. Most real constraints are probably some mix of what's genuinely out there and what your process is creating. These models can't tell you the exact ratio. They give you a way to estimate it.

The highway engineers eventually learned to ask a different question. Instead of "how much traffic is there," they started asking "how much of this traffic did we create." Some cities tried the experiment. They narrowed roads instead of widening them, invested in alternatives, and watched congestion fall. The demand had been induced all along.

Pipeline data is sitting in your ATS right now. Some of it describes the market. The rest describes your process. The only way to tell the difference is to change your inputs and see what moves.

The Recruiting Lattice takes mental models from fields like behavioral science, sociology, and decision theory and turns them into practical tools for talent acquisition.