Diagnose before you reprice

The reflex after a bad month is to lower the price, and it is the wrong move roughly half the time. A quiet month has two possible causes, they need opposite responses, and one column in your records tells you which one you are looking at.

The column is enquiries. Bookings alone cannot distinguish between nobody arriving and everybody leaving.

The split

Write down two numbers for the month: enquiries received, and jobs booked.

Few enquiries, normal conversion. Nobody is finding you. This is a visibility problem, and the price is almost certainly not the cause, because a price cannot deter someone who never saw it.

Normal enquiries, poor conversion. People are finding you and deciding against you. This is a conversion problem, and the cause is price, clarity, response time or trust - in that order of how often it turns out to be each.

Both down. Usually seasonality or a platform change rather than anything you did, and the correct response is to check whether it is happening to everyone before touching anything.

Both normal, income down. Not a quiet month at all. Your job mix shifted toward cheaper formats, which is a rate card question rather than a demand one.

The reason this works is that the two failures look identical from the bank statement and completely different one step upstream. Tracking the enquiries that did not convert is a single extra column, and the case for keeping it is that it is the only thing that answers this question at all.

If it is traffic

Price cuts do nothing here, and they cost you on the way back up.

The things that actually move visibility are, in rough order of effect: completed jobs, recency of activity, review count, and response time - all behaviours rather than copy, because behaviour is what a marketplace can measure. That ordering has an uncomfortable implication, which is that a quiet month makes the next month quieter, since low activity is itself a ranking input.

So the useful moves are the ones that produce activity: taking a smaller job you would normally pass on, adding a second format, updating the listing so the last-modified date moves, being present in the busy hours. How marketplace ranking actually works covers which of those signals a platform can see and which it cannot.

Check the boring explanations first, though. A listing paused during a holiday and never resumed, a category change, a verification lapse, or a format that quietly stopped being surfaced accounts for more quiet months than any subtle demand shift does.

If it is conversion

Now the price is a legitimate suspect, and it is still not the first thing to check.

Response time comes first, because it is the cheapest to fix and the most common cause. An enquiry answered in nine hours converts far worse than the same enquiry answered in nine minutes, and the fix is a window rather than more hours - the mechanics are in setting availability windows.

Clarity comes second. Buyers abandon when they cannot picture what arrives: no stated turnaround, no stated length, no visible price, no worked example. Each of those is a reason to keep looking, and none of them are about cost.

Price comes third, and only if the first two are clean and your quotes are being answered with silence rather than with counter-offers. Even then, test one variable and leave it a fortnight, because changing three things at once tells you nothing at four weeks' cost.

One month is not a signal

This is the part that matters most and is hardest to act on.

Bookings in a one-person service business are lumpy by nature. A month with a third fewer jobs than the previous one is within normal variation for most earners, and the sample is far too small to distinguish a trend from noise. A simple model for counts like these is the Poisson distribution, whose standard deviation is the square root of its mean, as NIST's Engineering Statistics Handbook sets out. At an average of 12 bookings a month that is about 3.5, so a month of 8 sits barely more than one standard deviation below normal.

Two consequences. Do not reprice on one month; two consecutive months moving the same way is the minimum, three is better. And review your numbers on a fixed date rather than when you feel worried, because reviewing when worried guarantees you review at the least informative moment and act on the noisiest data.

Some of the variation is calendar rather than performance, and knowing which fortnight is normally slow removes most of the panic - the seasonal pattern across a year is the reference for that.

The uncomfortable third possibility

Occasionally the diagnosis is neither traffic nor conversion but the market moving underneath the format.

The signal for that is specific: enquiries holding, conversion holding, and the prices you can get drifting down across several months while nothing about your own listing changed. That is crowding or substitution, and no amount of responsiveness fixes it.

Substitution is the one to check first, because the floor of this market has already moved once. Anything a model produces instantly is now priced as though a model produced it, and the technical account of what automated scoring reliably delivers is the clearest way to see whether your format sits above that line or on it. The other honest check is whether the thing buyers wanted was a number at all - what a score is actually communicating is worth reading against your own listing, because a format that promises precision it cannot carry converts badly for reasons no price change will fix. And where conversion is failing at the message stage rather than the listing stage, the buyer-side account of how a commission normally opens shows what the other party thought was supposed to happen next.

If the quiet month coincides with a platform change - a new fee, a new ranking rule, a category reorganisation - that is a specific and findable cause rather than a mystery, and Rate Cock's judges page is the current statement of how its own listings work, which is the thing to compare against what you remember.