Portfolios show the best work a firm has ever produced, tidied for presentation. References are chosen. A small paid test on your own project shows what actually arrives on a Tuesday morning, which is the only thing you need to know.
Pay for it
Free samples sound appealing and give you the wrong signal. A supplier doing unpaid work assigns their best available person and treats it as sales, which is not the experience you will get at volume. Paying a modest fee buys you an honest look at the normal process, and it means you can give proper feedback without feeling awkward.
Choose real work, not a showcase
Pick something representative rather than something impressive: one sheet or a small package from a live project, ideally with the usual imperfect inputs. If your real inputs are a marked up PDF and a phone call, test with that. Testing with an unusually clean brief tells you nothing about how they handle your actual working conditions.
Brief it exactly as you normally would
Resist the urge to over explain. If you would normally send a redline and a two line email, do that. You are testing whether they can work from your habits, not whether they can follow an unusually thorough brief you will never write again.
What to judge when it comes back
- Questions asked before starting. Silence is a warning. Good drafters query ambiguity early.
- Drawing discipline. Layers, line weights, text styles, whether the title block is complete.
- Accuracy. Do dimensions close? Do plans, sections and elevations agree with each other?
- Annotation judgement. Anyone can trace geometry. Knowing what to label and what to leave out is the skill.
- Adherence to your standards rather than substitution of theirs.
- Turnaround honesty. Did it arrive when they said it would?
Then test the correction round
This matters as much as the first delivery. Mark up the set properly and send it back. How they handle corrections tells you how the relationship will feel for the next two years: whether they fix the specific issue or the whole category, whether the second version introduces new errors, whether they push back when your markup is wrong.
Give the feedback you would give internally
Be specific and unsparing. A partner worth keeping will absorb it and adjust. One who becomes defensive on a test project will be worse under deadline pressure.
Decide on the second delivery, not the first
Almost anyone can produce one good sheet. Consistency across a correction round and a second package is what predicts the next two years. Budget for two small tests before committing volume, and treat the fee as the cheapest due diligence available.