How a year-on-year comparison works
A year-on-year comparison sets a period against the matching period a year earlier — this September against last September, this quarter against the same quarter last year — and reports the difference. Both periods sit at the same point in the annual cycle, so seasonality largely cancels out and what is left is closer to real change.
Two alignment choices decide whether the comparison is honest. Calendar alignment matches the dates. Weekday alignment matches the days of the week, which matters when enquiries collapse at weekends, because two calendar periods can hold a different count of working days. Where the driver of demand is a moveable festival, neither works on its own: Dashain and Tihar follow the lunar calendar and drift between Gregorian months, so the comparison has to be shifted to sit around the festival rather than around the date.
Why year-on-year comparison matters
It is the cheapest defence against a wrong conclusion. Month-on-month reporting in a seasonal business mostly measures the season, and the person reading it ends up rewarding or punishing decisions that had nothing to do with the movement. An annual comparison strips most of that out.
It also changes what a flat month means. Holding steady against last year while the market contracts is a good result that reads as failure in a month-on-month report, and growing against last month during a festival period is not an achievement at all.
Common mistakes with year-on-year comparisons
Comparing against an abnormal year is the most frequent. If last year contained a lockdown, a shop closure, an outage or an unusually large one-off order, the comparison inherits it, and the sensible move is to say so in the report and show an earlier normal year as well.
The second is comparing across a tracking change. A move to a new analytics property, a change in consent handling, a new tag setup or a site migration can shift what gets counted, and a data discrepancy of that kind looks exactly like a business result. The third is drawing conclusions from a period too short to be stable, because a single week against the same week last year swings on noise.
How to act on it
Report both views, let the annual one lead, and keep the month-on-month view for pacing and for spotting sudden breaks. Annotate the chart wherever something happened — a price change, a campaign paused, a site release, a tracking rebuild — so the reason for a step is recorded while it is still known, rather than guessed at a year later.
Then dig one level down before acting. A flat headline often hides one channel growing and another falling, so compare by channel, by product and by location. Building that comparison into the standard view, instead of assembling it by hand each month, is a straightforward piece of reporting setup.