How YoY, MoM and WoW work
All three take the period you are looking at and place an earlier period beside it. Year on year compares a month or quarter with the same stretch of the previous year. Month on month compares it with the month just gone. Week on week compares a week with the one before.
They are not interchangeable, because each one controls for something different. Year on year holds season constant, so a hotel comparing a monsoon month with the same monsoon month last year is comparing like with like. Month on month holds nothing constant except recency, which makes it good for spotting the effect of a change you made and bad for judging a seasonal business. Week on week is the most sensitive and the noisiest, useful while a campaign is being launched or repaired and misleading as a measure of progress.
Why the comparison period matters
A single number means nothing on its own. Enquiries this month are only good or bad against something, and the something you choose decides the story. The same month can look like growth against last month and a decline against last year, and both statements can be true at once.
Choosing deliberately protects you from two opposite errors: panicking over an ordinary seasonal dip, and celebrating a seasonal rise as if the work caused it. For slow channels such as search and content, the year-on-year view is usually the honest one, because a month is barely long enough for a change to show.
Common mistakes with period comparisons
Comparing months as if they were equal is the everyday one. Months differ in length and in how many weekends and public holidays they contain, and for a business whose enquiries arrive on working days that difference alone moves the total. Festival timing does the same job on a larger scale: Dashain and Tihar shift against the Gregorian calendar, so a Nepali business can compare October with October and still be comparing a festival month with an ordinary one.
The other frequent error is comparing an incomplete period with a complete one — a report written mid-month that quietly compares part of this month with all of last. Late-arriving conversions cause a milder version of the same problem, making the most recent period look worse than it will eventually be. And a comparison against a broken period, such as a month when tracking was down, produces a spectacular improvement that never happened.
How to act on it
Pick the comparison that matches the question. To judge whether the business is growing, use year on year. To judge whether a specific change worked, use month on month, and only if nothing seasonal moved at the same time. To watch a launch or an incident, use week on week and expect noise.
Show more than one where it is honest to do so, and note the reason for a distortion — a shifted festival, a shorter month, an outage — beside the figure. Where the weekly view jumps around too much to read, a rolling average smooths it, and the entry on seasonality explains which swings to expect before you try to explain them away.