Introduction | Most teams use profiles as sieves—scoring, ranking, and deciding who gets in and who gets out. But there is a more valuable use: giving you a better set of questions.
1. What goes wrong most easily when you use a profile as a sieve
An 800-person consumer goods company set a profile for sales roles three years ago: bachelor's degree from a top university, FMCG background, extroverted personality. Three years later, the system is still screening people by these criteria—while several of the best-performing salespeople graduated from non-top universities, started e-commerce businesses, and scored as introverted on personality assessments.
The problem isn't that the original profile was "wrong"; it's that it became fixed as a sieve.
A sieve only answers "does this look like the profile?" not "why?" It turns past success into a constraint on the future; once you start using it to eliminate people, you no longer even have a chance to ask follow-up questions.
2. A different use: let the profile generate questions
The same data, asked about differently, becomes completely different in use. Public information actually contains many "questions you can ask":
• Collaborators have long been concentrated in a few institutions → "How did your collaboration start? Would you consider new partners?"
• Research topics have shifted markedly over the past three years → "Why change direction? How does it connect to your earlier work?"
• Citations are concentrated in two early outputs → "Which part of your work do you most want people to remember?"
A profile gives you clues, not conclusions. The value of clues is that before you meet, you already know what to ask—far more efficient than figuring out how to make conversation after you meet.
3. Three preferences when looking at data
The same output record tells a different story depending on how you read it.
1. Timeline matters more than quantity: sustained output over five years and ten papers concentrated in one year describe two completely different ways of working.
2. Structure matters more than total volume: where they publish, with whom, and in which subfield says more about their position than the total count.
3. Change matters more than the current state: the current direction often predicts the next step better than the labels of the past decade.
Once the order is right, looking at the specific numbers will make your judgment much more stable.
4. Why public information is enough
Finding collaborators, assembling review panels, inviting advisors—most of the information you need in these situations is already public: who is doing what, with whom they collaborate, and where their output is concentrated. All of this can be found and verified.
Public information has another hidden benefit: both sides are looking at the same facts. So the first conversation can start with "I saw that you've worked in a certain direction" rather than with self-introductions.
To gather these clues into a presentable document, you don't have to manually go through a dozen websites. MuchSci's Talent Profile Assistant is the tool for this step: enter a name or a direction, and it outputs a sourced analysis report covering scientific output, representative work, collaboration networks, and field positioning.
| Clues in public information | Questions you can ask |
| Collaborators concentrated in a few institutions | How did your collaboration come about? Would you consider new partners? |
| Research topic shift in the past three years | Why the shift? How does it connect to your earlier work? |
| Citations concentrated in early work | Which part of your work do you value more, and where do you want to go next? |
| Title and output don't match | Where do you mainly spend your time? |
| Output concentrated in conferences rather than journals | Which type of peer recognition do you value more? |
5. When not to use a profile
• Cannot replace conversation: data knows output, not the person.
• Cannot produce evaluative conclusions: high output doesn't mean they fit your team or direction.
• Cannot replace what the person themselves says: judgments involving personal intentions and motivations can only be answered by them.
6. A profile is only the first step
Seeing it within the process makes it clearer: public information generates a profile, the profile translates into a question list, the question list goes into the conversation, and the results of the conversation update your judgment. Of these four steps, only the first is suited to a tool.
MuchSci's Talent Profile Assistant can make the first step solid—clearly marking sources and gathering clues so you don't spend time browsing websites. But after the second step, it's still up to you: what to ask, what to hear, and how to judge—no one can do these for you.
A useful profile ultimately doesn't come down to "how many points this person gets," but to three actions: investigate thoroughly first, ask to understand second, and only then make a judgment. Reverse the order, and the profile degenerates into a label.
Next time you get a talent report, don't rush to look at the conclusions—first see how many questions it enables you to ask.