The free sample is not an investment

Every new earner is told to do a few jobs free to build a reputation. The advice is well meant, and it mistakes an absence of evidence for an absence of demand.

Free work does not produce the evidence you need, and it produces a signal you do not want.

What the free job actually buys

A buyer who pays nothing has told you nothing. You have not learned that they would have paid, because they did not have to decide.

The review it generates is weaker too. Reviews of free work read as reviews of free work, and buyers discount them without needing to think about it. What persuades the next buyer is that someone chose to spend money and was glad afterwards, and that is exactly the part the free job omits.

There is also a selection problem. The people most eager to take free work are, on average, the people least likely to ever pay for it. So the queue you build is made of the wrong people, and it looks like traction while it is happening.

The signal it sends

Price is read as information about quality in every market where quality is hard to verify in advance, and this is one of those markets.

A floor of zero says something about the ceiling. It is very hard to move from free to a normal rate with the same audience, because you are not raising a price, you are introducing one.

What to do instead

Publish the shape of the work rather than giving away an instance of it. The structure of what a buyer receives, the axes you score on, the length, the turnaround. This answers the only question a free sample answers, at no cost, and it scales to every visitor rather than one.

Make one generic worked example, on material nobody owns, and reuse it forever. That is a sample in the useful sense: it shows the work without transferring it.

Take small paid jobs instead of large free ones. A low price is recoverable; zero is a different category. The cold-start post covers the rest of what actually shortens the first month, and what buyers scan for covers where that effort should land.

The one case where it is fine

Doing a job free for someone who was always going to be a reference - a person with reach, a repeat buyer testing a new format, an established name who will say so publicly - is not a sample. It is paid work where the payment is the reference, and you should treat it with the same seriousness: scope it, deliver it, and be explicit with yourself about what you are being paid in.

Everything else is a discount you have given to a stranger.

Buyers, for what it is worth, do not expect free work and are frequently surprised to be offered it - their account of how a commission normally opens has nothing in it about samples. What has genuinely made cheap generic assessment worthless is automation, not competition between people; the account of what an automated score already does explains why the floor of this market moved and why undercutting it is a race you cannot win. The measurement conventions worth being precise about, if precision is the thing you are selling, are documented here.

If you want the platform-side version of getting started rather than the general one, Rate Cock's judges page has it.