Why your conversion rate keeps changing month to month

4 days ago
3 min read

Updated: 3 days ago

Introduction

Forty per cent in January, twenty-two in February, thirty-five in March. Somebody asks what went wrong in February and a plausible story gets constructed — the weather, a competitor, the new enquiry form. Meetings follow. Changes are made. April comes in at thirty-one and nobody can say whether the changes helped.

At small volumes this is almost entirely arithmetic. With twenty enquiries a month, two jobs going either way moves the rate by ten points, and two jobs going either way happens for reasons that have nothing to do with how the business is run. The rate is real, but the month-to-month movement in it mostly is not.

Understanding that changes what you do with the number, and mainly it means looking at it less often over longer periods.

1. Why your conversion rate keeps changing is mostly the denominator

The arithmetic.

Small denominators produce large swings from small absolute changes. Twenty enquiries and one extra win is five percentage points. Nothing about the business changed.

2. The mix of enquiries moves

The composition effect.

A month heavy with price-shoppers converts worse than a month heavy with referrals, and the mix moves for external reasons. This is real but it is not performance. Track conversion by source and it becomes visible.

3. Timing crosses the month boundary

The accounting artefact.

An enquiry that arrives on the 28th and closes on the 3rd depresses one month and inflates the next. Long cycles make this worse. Cohort the enquiries by arrival date to remove it.

4. Seasonality is real and repeats

The predictable part.

Most businesses have months that are reliably better, and comparing to last month rather than the same month last year mistakes a season for a trend. One year of data makes this checkable. Two makes it dependable.

5. Definitions drift quietly

The measurement cause.

If someone started logging enquiries that previously went uncounted, the rate falls without anything getting worse. This is one of the commonest causes of an alarming number. Check what changed in the counting first.

6. Look at rolling three months

The practical fix.

A three-month rolling figure absorbs most of the noise while still moving when something real happens. It is the single most useful change to how the number is presented. Plot it beside the monthly figure once and the point makes itself.

7. Watch the counts, not only the rate

The complementary view.

Enquiries and wins as raw numbers often show what the ratio hides, particularly when both are moving. A falling rate with rising wins is usually good news. The rate alone cannot tell you that.

8. Set a band, not a target

The management approach.

Decide the range within which you will do nothing, and act only when the rate leaves it for two consecutive periods. This prevents the monthly reaction cycle. The band is easier to agree than a target.

9. Investigate a run, not a point

The rule of thumb.

Three months in the same direction is a signal; one month is weather. Almost every useful investigation starts from a run. Almost every wasted one starts from a single figure.

Be careful about explaining a bad month convincingly. A good explanation is available for any number, and its availability is not evidence that it is correct, which is why the band and the rolling figure are worth more than the discussion.

Conclusion

Treat month-to-month movement in a small funnel as noise until it persists.

Recognise that a small denominator produces large swings, expect the enquiry mix and the month boundary to move the figure, compare against the same period last year rather than last month, check whether the counting definition changed before assuming performance did, present a rolling three-month figure alongside the raw counts, agree a band within which you take no action, and investigate runs rather than individual months.