Comparing This Month to Last Month Badly

Why month against month misleads

Months are not comparable units, and treating them as such produces conclusions about nothing.

They contain different numbers of days, between twenty-eight and thirty-one.

They contain different numbers of working days, which for a business-to-business audience matters more than total days.

They contain holidays, school terms, and weather.

A February that looks catastrophic against January is frequently February being three days shorter and containing a holiday.

The working days problem

The largest and least considered factor for most small businesses.

A month with twenty-three working days against one with twenty is fifteen percent more opportunity before anything else happens.

For a trade whose enquiries arrive on weekdays, that difference alone explains most ordinary month-to-month movement.

Which means a comparison worth making divides by working days rather than comparing totals.

Enquiries per working day is a more honest figure than enquiries per month, and it takes ten seconds to calculate.

Compare to last year instead

The straightforward improvement, and it addresses most of the problem at once.

August against last August has the same season, similar holidays, and a similar shape.

It also captures whether the business is growing, which month-to-month comparison never does.

The limitation is that it requires a year of data, and that a change in the business makes the comparison less clean.

For anybody with less than a year, comparing a rolling three months against the previous three is the next best thing, since it smooths out most of the noise.

A worked example

A business alarmed by a twenty-two percent drop in enquiries from July to August.

They were preparing to change several things on the site in response.

Checked against the previous August, enquiries were up eleven percent.

Checked per working day, the July to August drop shrank to about fourteen percent, and August had a holiday weekend accounting for most of the rest.

Nothing was wrong. August is quiet in their trade and always has been, and the business was growing.

Had they acted on the month-to-month figure they would have changed a site that was working, and attributed the September recovery to the change.

Normal variation

The concept that prevents most unnecessary reaction.

Every figure moves without meaning. On small numbers it moves a great deal.

Six enquiries one month and nine the next is a fifty percent increase and is well within what randomness produces.

Which means the useful question is not whether a number changed, but whether it changed more than it usually does.

Looking at twelve months of a figure tells you its normal range, and anything inside that range is not news.

The comparison against a target

A different kind of comparison and the one most worth making.

Traffic and enquiries compared against last month or last year tell you the direction. Neither tells you whether the number is adequate.

Adequate means enough work to fill the diary at the margin you need, which is a figure the business can state and analytics cannot.

Working backwards is straightforward: the jobs needed per month, divided by the proportion of quotes that convert, gives the enquiries required.

That number is the one to compare against, and it makes an eighteen percent drop either serious or irrelevant depending on where the figure started.

A business twenty percent above what it needs can watch a fall without alarm. One below it should have been worried the month before.

What is worth reacting to

• A sustained change over three months or more.
• A sharp drop that starts on a specific day, which suggests something broke.
• A change in one source while others hold steady.
• A change in enquiries without a change in traffic, or the reverse.
• Anything that coincides with something you did.

The second is the most actionable. A drop that begins abruptly is usually technical, and finding the date narrows the cause enormously.

The counter-case

Where month-to-month is the right comparison.

Judging a campaign that ran for one month, where the immediately preceding month is the relevant baseline.

A business with no seasonality and steady weekday demand, where months are genuinely comparable.

And any situation where you are watching for breakage rather than measuring performance, since a sudden drop is worth noticing regardless of what caused the previous month's figure.

The distinction is whether you are asking how are we doing or has something gone wrong. The second tolerates a rough comparison.

Averages hide the shape

A related trap, and it applies to any figure summarised over a period.

An average of forty enquiries a month can be forty every month, or twenty for eight months and ninety for four.

Those are completely different businesses with the same average, and they need different decisions about staffing, cash and advertising.

Which is why looking at the monthly series is worth more than looking at the annual total divided by twelve.

The same applies within a month. A figure that arrives entirely in the first week is a different situation from one spread evenly, and a monthly total conceals both.

Twelve numbers in a column tell you more than any single summary figure, and they take no longer to read.

Seasonality is information

Worth building into planning rather than treating as noise to remove.

Most trades have a shape: a quiet January, a busy spring, a dead fortnight in August, a rush before Christmas.

Knowing yours means quiet periods are expected rather than alarming, and it tells you when to do the work that never gets done.

It also tells you when to advertise, which is generally shortly before demand rises rather than during the quiet period itself.

Two years of monthly enquiry figures reveals that shape clearly, which is another argument for keeping the record from the beginning.

A comparison worth making

1. This month against the same month last year.
2. Per working day, not per month.
3. Rolling three months against the previous three.
4. Against the normal range of the last twelve months.
5. Noting anything you changed, with dates.

The fifth is what makes any of this interpretable later. A record of what changed and when turns a chart into an explanation.

What to count in the first place is covered in counting enquiries instead of visits.

Frequently asked questions

West Coast Media Solutions Inc. provides web design, web development, hosting, digital marketing, and business consulting to organisations across Canada, drawing on more than twenty-five years in the field.