An interview should help predict how someone will perform after joining. But many hiring processes gradually test something else: how well a candidate can perform inside the interview format itself.
Standard questions, familiar cases, rehearsed answers, and now AI preparation make interviews increasingly predictable. Predictability is easy to scale, but it weakens the signal.
Interviewing and working are different skills
Interviews reward fast answers, confident delivery, and the ability to make a strong impression in a short period of time.
Real work often values different things: asking questions, admitting uncertainty, revisiting decisions when new data appears, collaborating, and sustaining effort until a result is delivered.
Someone can be an average interviewee and a strong employee. The opposite is also true.
The more questions come from the internet, the less you learn
If candidates know the questions in advance, they prepare answers rather than thinking.
This does not mean interviews should become trick contests. It is better to start from real experience and keep going deeper until memorized phrasing stops being useful.
A good answer can start with “I don't know”
Some interviewers treat a pause or clarifying question as weakness. In complex work, however, the ability to stop and understand the situation can be more valuable than instant confidence.
This matters even more for leaders and experts, where confidently making the wrong decision can be expensive.
Test behavior when context changes
Do not only ask, “What would you do?” After the first answer, add a new fact and see whether the decision changes.
A strong candidate does not need to defend the first answer forever. The ability to revise a view calmly when the inputs change is often a positive signal.
Take-home assignments are not magic either
A home exercise can be completed with colleagues, AI, or simply with far more time than the real job would allow.
The final document therefore matters less than the conversation around it: how the candidate reached the solution, which options were rejected, where uncertainty appeared, and what would change under different constraints.
Compare candidates against criteria, not impressions
An unstructured conversation can easily become a contest of personal chemistry.
For each role, define a few signals that actually matter: decision quality, depth of experience, communication, conflict handling, learning ability, or whatever is relevant. Evaluate those after the interview.
The interview is not dead. It just needs its purpose back
The goal of an interview is not to prove that a candidate can answer your questions. The goal is to collect enough evidence to understand how that person is likely to work in the real context of the company.
The less the process resembles an exam with known questions and the more it resembles an honest discussion of real decisions, the better chance you have of seeing the person rather than the prepared version.