Measuring a festive campaign when everything goes up

In October almost every marketing dashboard in India turns green. That is the problem.

Written by the Colleviate Strategy desk

During the festive season sales go up, traffic goes up, branded search goes up and return on ad spend usually goes up too, because people who were going to buy anyway are now buying in the same few weeks the media runs. A campaign can look like a triumph and have added very little. Another can look ordinary and have held on to share that would otherwise have gone to a competitor.

Seasonal lift is not campaign lift

Separate the two before the season, in writing. Seasonal lift is what would have happened with no campaign: last year’s festive window, adjusted for the brand’s underlying trend. Campaign lift is the difference between that and what actually happened. Most festive reports present the total and imply it is the second.

Compare festival to festival

Diwali moves by up to three weeks between years, so calendar comparisons break. The vehicle trade’s 2025 figures show how badly. Retail in October 2025 was up 40.5% on October 2024, but across the festive window it was up 21.1%, because Diwali fell on 20 October in 2025 and on 31 October in 2024. Define the festive window by the festival, as the Federation of Automobile Dealers Associations does (the first day of Navratri to 15 days after Dhanteras), and compare it with the same window last year. Do the same with daily data: line up days by their distance from Diwali, not by their dates.

Hold something back

The only clean answer to what the spend caused is a comparison with people who did not see it. Keep a few matched cities, or a random slice of the audience, out of the campaign, and compare. The case for doing this is strong. When researchers compared fifteen large advertising experiments at Facebook with the observational methods advertisers usually rely on, those methods often failed to recover the experimental result, even with rich data on who saw the ads (Gordon, Zettelmeyer, Bhargava and Chapsky, Marketing Science, 2019). The bias is worst when the people most likely to buy are also the people most likely to see the ads, which describes every Diwali.

A festive holdout feels expensive, because it means not advertising to some people in the best weeks of the year. It costs less than a year of decisions made on the wrong number.

Watch share, not volume

Branded search for almost every consumer brand rises in festive weeks, so a rise in yours says little. Your share of branded search against named competitors says more, because the season lifts everyone and share shows who gained. Les Binet’s work for the IPA found that share of search tends to move ahead of market share, which makes it a useful early read while the season is still running. The same logic applies to sales: share of the category in the window, from panel or marketplace data where it exists, is a better test than growth. Our Share of Search Index for India shows the method on nine categories.

Count the pull-forward

Festive offers move purchases in time as well as creating them. Some of what a campaign sells in October would have been bought in November or December at full price. Measure the weeks after the window as well as the window itself, and expect a dip. In 2025 the effect also ran the other way: FADA reported three weeks of muted demand in September as buyers waited for GST cuts announced in advance. A price change that people know is coming moves demand to the day it arrives.

Judge it on contribution

Festive discounts, bank offers, free delivery and returns all cut into the margin on every sale, while media costs rise as every brand bids for the same audiences. A return on ad spend that looks healthy in revenue can be negative in contribution. Report what each sale contributed after discounts and the full cost of the campaign, as we set out in what ROAS tells you and what it hides.

A festive scorecard

• A festive window defined by the festival, compared with the same window last year.
• A holdout, by geography or audience, with the comparison agreed before launch.
• Share of branded search and, where available, share of category sales against named competitors.
• Contribution after discounts and the full campaign cost, not revenue over media spend.
• The four weeks after the window, to see what was pulled forward.

Agree all five before the first rupee is spent. Agreed afterwards, every one of them becomes a negotiation about whether the campaign worked.