The question quality signal is the most reliable one available. Somebody asking about why something is done a particular way, what happens in an unusual case, or how their work connects to somebody else's is building a model of the business. Somebody asking only what to do next, repeatedly, for tasks already explained, is not, and that pattern rarely reverses.
Rework is the second and it is measurable rather than impressionistic. Work that comes back needing correction in the same way twice indicates something is not landing, and the useful response is to establish whether the instruction was clear before concluding anything about the person. Frequently the first correction was ambiguous and the second was inevitable.
Give it a fair period, since a first hire in a small business faces a genuinely difficult start. No established process, limited documentation, an owner who is busy, and no colleague to ask. Somebody productive at week four in that environment is exceptional rather than typical, and expecting it produces unfair assessments.
Watch what happens without instruction, which is the signal that matters most in a business of two. A person who identifies something that needs doing and does it is operating at the level a small business requires. One who waits for direction on everything is not necessarily poor at the work, and they require management capacity you do not have.
Look at whether the relationship is improving or repeating. A person making different mistakes each month is learning. One making the same mistake repeatedly, after it was addressed clearly, is telling you something about fit rather than about effort. That distinction is usually apparent by week six.
Ask them how it is going and listen to the specifics, since somebody struggling frequently knows and says so if invited. A person who names what they find difficult and asks for something specific is easier to help than one who says everything is fine while the work suggests otherwise.
Be honest about your own contribution before deciding, because a substantial share of failed first hires are management failures. Unclear expectations, no documented process, feedback given late or not at all, and a role that turned out to be different from the description. Those are fixable and they are yours to fix.
Then act on the conclusion rather than deferring it, which is the part that costs businesses the most. A hire that is not working rarely improves on its own past the two or three month point, and the delay in addressing it consumes capacity while the person themselves is usually aware. Ending it clearly and fairly is better for both parties than a slow deterioration.
Write down what good looks like before they start, since assessing against an unstated standard produces a judgement you cannot explain and they cannot act on. Three or four specific things somebody should be able to do by month three is enough, and it makes both a positive and a negative conclusion defensible.
Give feedback early and specifically rather than saving it, since somebody who does not know something is a problem cannot address it. Most disappointing hires received their first clear feedback at the point the decision had already been made.
Watch how they respond to correction rather than only whether they need it, since that response predicts the next six months. Somebody who takes a point and applies it is developing, and somebody who explains why it was not their fault is showing you a pattern.