In short
- Sort on relevance first, then read the three that fit properly.
- A question about your brief is the strongest single signal.
- Compare breakdowns, not totals.
- Interview for judgement, not for a list of tools.
Twenty proposals is not twenty options. Most will be near-misses on scope, a few will be templates, and two or three will be from people who could do the job. The task is to reach those three quickly and then read them slowly, which is the opposite of how most shortlists get built.
First pass: relevance only
Give each proposal about thirty seconds and ask one question: has this person done something like this? Not something impressive. Something like this. A developer with three marketing sites in your stack is a better bet for a marketing site than one with a famous app in another language.
- Discard anything that could have been sent to any job.
- Discard anything that misread the scope, unless they say why they read it differently.
- Keep anything that names a specific part of your brief.
- Keep anything whose examples resemble your problem, even if the work is modest.
Second pass: read for signals
Now read the survivors properly. Four things predict how the work will go.
| Signal | Strong | Weak |
|---|---|---|
| Questions | Asks about the gap you knew was in the brief | No questions, or questions the brief already answers |
| Examples | Two or three close to your situation, with their part explained | A portfolio link and nothing else |
| Estimate | Split into parts, with what each assumes | A single number, or a range wide enough to cover anything |
| Plan | Says what happens first and what they need from you | Promises everything, describes nothing |
Weight the question hardest of the four. Every brief has a gap, and somebody who finds the right one has already done a piece of the thinking your project needs.
What not to read into
- Reply speed. It measures who was at a screen, not who is good.
- Proposal length. A long proposal is often a long template.
- Confidence. "No problem, easy" on a brief with real unknowns is a warning, not reassurance.
- Perfect English. You are hiring for a skill. Judge clarity, which is different.
- Round-number underbids. A quote well under the others usually means a different scope was priced.
Compare prices without comparing totals
Two quotes are rarely for the same work. Before comparing, write down what each includes: revisions, testing, deployment, content, handover. Then compare like for like, and treat the difference as information about how each person understood the job.
The conversation
Two shortlisted people is enough. The call is for judgement, not for a tools checklist, so ask about decisions rather than experience.
- What would you want to know before you start that I have not told you?
- Which part of this is most likely to take longer than expected, and why?
- Show me something that went wrong on a project and what you did about it.
- How do you prefer to get feedback, and how often?
- What do you need from me in the first week?
Shortlisting and messaging commit you to nothing. Marking someone shortlisted or interviewing also tells them they are in contention, which good freelancers act on, so use it for the people you are serious about.
Decide, then make it small
Pick the person whose thinking matched the problem, and structure the start so that being wrong is survivable: a first milestone with a real output you can judge inside a week or two. Breaking a project into milestones covers how to split the rest.
Then reply to everyone else. A short decline costs a sentence, and the person you turn down this month is often the person you want next quarter.
How many proposals should I read properly?
Three to five. Sort for relevance first and read those closely. Reading twenty superficially produces a shortlist built on presentation rather than fit.
Should I hire the cheapest proposal?
Only after you have checked it is for the same work. A low number usually means a smaller scope was priced, and the gap tends to reappear later as extra milestones.
Is a new account with no reviews a bad sign?
No, it is an unknown. Everyone starts somewhere, and new freelancers often do careful work to build a record. Size the first commitment accordingly, and look at whether they can bring verified stats from other platforms.
What if nobody is right?
Close the job and rewrite the brief rather than settling. Proposals that all miss usually mean the post was ambiguous about the same thing.