Most technical hiring still relies on a conversation and a feeling. The conversation may be good. The feeling may be right. But neither is evidence.
Bennett & Fletcher has been running online technical assessments since 2003. The tools have changed; the principle has not. Assessment is not about adding more interviews. It is about deciding what you need to know, asking the same questions in the same way, and writing down what you heard before you compare candidates.
Start with a scorecard
A role scorecard is not a job description. It sets out the few things that actually predict success in the role: technical depth, scope, ambiguity, communication and motivation. Each dimension has a small number of observable indicators.
For an engineering manager, for example, we might look for evidence of:
- running a team through a difficult delivery;
- making a technical trade-off with product and commercial constraints;
- developing someone into a more senior role;
- communicating a decision to people who disagreed.
The scorecard is agreed with the hiring manager before the search starts. If the hiring panel cannot agree on what good looks like, the search will not fix that.
Use structured interviews
A structured interview asks the same core questions of every candidate and scores the answers against the same rubric. It feels less conversational than an unstructured interview. It is also much more reliable.
We use a mix of past-behaviour questions and scenario questions. Past behaviour gives us evidence. Scenarios tell us how someone thinks when the answer is not in their memory. Neither is enough on its own.
Include a work sample
For technical roles, a work sample is often the most useful signal. It can be a code review, a system design discussion, a data model, a product critique or a written response to a real problem from the client.
The point is not to recreate a day of work. It is to see how someone reasons, what they prioritise and how they explain a trade-off. A good work sample is short, relevant and scored against the same rubric as the interview.
Calibrate before you compare
Calibration is the step most teams skip. Before a shortlist is presented, the interviewers compare their scores and discuss the gaps. Where two people disagree, they go back to the evidence.
This does two things. It reduces the influence of the loudest voice in the room. It also produces a written rationale that the hiring manager can use with the rest of the business.
Watch for the false positive
The most expensive hiring mistake is not a candidate who interviews badly. It is a candidate who interviews well and cannot do the job. Charisma, confidence and a strong CV can all mask a lack of depth.
That is why we separate “interview performance” from “evidence against the scorecard”. A candidate can be impressive and still not meet the bar. The scorecard gives us permission to say so.
Make the process fair
A structured process is also a fairer process. It gives candidates a clear view of what is being assessed, a consistent experience and useful feedback. It reduces the role of background, accent and similarity bias.
We tell candidates what the process will involve, who they will meet and what the salary band is. We give feedback after every interview. That is not a favour; it is part of doing the work properly.
The test
A good assessment process should let you answer three questions:
- What did this person actually do?
- How does that compare with the scorecard?
- What is the risk if we hire them?
If you cannot answer those questions in writing, you do not have an assessment. You have an impression.