# Surf Excel Ad Testing Case Study: How the Vintage vs Modern Test Was Designed

How a test of a vintage functional Surf Excel ad against a "Daag Acche Hain" film was designed: proto-monadic order, 200 parents, screening. No results.

Published 2026-09-04 · Updated 2026-10-12
Canonical: https://hercules.works/blogs/surf-excel-ad-testing-case-study-india

This study was set up to compare a vintage functional Surf Excel ad ("Pour Rub Pour") with the brand's "Daag Acche Hain" purpose-led parenting narrative. Its design shows each ad in turn, in randomised order, to urban parents in six Tier 1 metros. The report, generated on Hercules on 28 July 2026, describes the design and a 200-person sample but contains no creative results: every results section is empty, and the report also says the dataset has 0 respondents. This page explains the design and what it could and could not answer.

## A Case Study Without Results

Most case studies lead with findings. This one cannot. The report sets out the question, the design, a sample profile and the quality screening, but every creative-results section reads "data unavailable", and the report says data collection is the next step. We have no later results to report.

The design is still worth reading if you plan to compare two ads from different creative traditions. Both ads are Surf Excel's own advertising, used as test material. The report is written as advice to the brand but does not say who commissioned the study. Any results, once collected, would describe respondents' reactions to two films, not the brand's sales.

## The Creative Question

The report frames two creative territories. One is vintage functional positioning: the "Pour Rub Pour" ad, which the report calls a 2016 campaign. The other is the modern "Daag Acche Hain" ad, which the report calls a purpose-led parenting narrative. The report's question is whether the brand should re-integrate tactical functional hooks into its long-term progressive narrative.

The questionnaire turns that into measures for each ad: how eye-catching it was (1 to 10), whether it held interest to the end, how well it fits the brand, and how likely the viewer is to consider buying after it. After both ads come a head-to-head preference, a final brand choice against competitors, and the words people associate with Surf Excel.

## The Design: Proto-Monadic, Randomised Order

The report calls its design proto-monadic: each respondent sees one ad and answers about it before seeing the second, and the order of the two ads is randomised. It sits between the two common alternatives. Our summary:

| Design | How it works | Strength | Weakness |
|---|---|---|---|
| Pure monadic | Each person sees one ad only | Cleanest read of each ad | Needs a separate sample per ad; no within-person comparison |
| Paired comparison | Both ads, then a choice | Cheap and direct | The first ad colours reactions to the second |
| Sequential (proto-)monadic, randomised (this study) | Ad A rated, then ad B rated, then compared; order alternates | First-seen scores are untouched by the other ad; head-to-head possible | Second-seen scores carry an order effect; randomising spreads it but does not remove it |

The report says randomising the order eliminates sequence bias. It does not: it balances the bias across the two ads. If you use this design, read each ad's first-seen scores on their own as well. They form a clean monadic comparison with half the sample per ad, and the gap between first-seen and combined scores shows how much order mattered.

## The Sample the Report Describes

| Feature | As reported |
|---|---|
| Completed surveys | 200 (target 200) |
| Raw completes before screening | 400 |
| Excluded after attention checks and data-quality screening | 200 (50.0%) |
| Who | Parents in Tier 1 metros who take part in laundry decisions |
| Cities | Mumbai, Delhi, Bangalore, Chennai, Kolkata, Hyderabad |
| Gender | 100 men, 100 women |
| Aged 25-35 | 41.5% (n=83) |
| Aged 35-45 | 41.5% (n=83) |
| Aged 45-55 | 17% (n=34; recruiting capped at age 48) |

**The report contradicts itself on the sample.** Its methodology page records 200 analysed responses after excluding 200 of 400. Its executive summary and closing section say the dataset contains 0 respondents (n=0). Since no results appear anywhere, the 200-person profile above may be a recruitment plan rather than achieved fieldwork. We could not settle which from the report.

The report explains the 50/50 split as reflecting joint grocery decisions in dual-income homes, and the all-metro design as matching where liquid detergent and washing machines are most common. It cites no source for either premise, so treat them as the study team's assumptions. Recruitment is not described either, so any results would carry no margin of error.

A 50% exclusion rate is high. When you see one, ask what each check was, how many failed each, and whether those excluded differ by age or city from those kept, because heavy screening can tilt a sample.

## The Planned Analysis

The report plans to compare purchase intent for Ad A and Ad B with a paired t-test, overall and within two age cohorts (25-35 and 36-48). A paired test fits here, because the same person rates both ads. To compare the two cohorts with each other, you would need an independent-samples test or a regression with age as a factor. Our [significance testing guide](/blogs/statistical-significance-testing-india/) covers both.

Two details to fix before reusing the plan. The reported age bands overlap at 35, and the analysis cohorts (25-35 and 36-48) do not match them. Define one set of non-overlapping bands before fielding. And at about 83 people per band, each cohort sits under the 100-person line that the report itself sets for reading subgroups as more than directional.

## What This Study Cannot Tell You

- **Which ad works better.** There are no results.
- **Anything about "vintage" or "modern" advertising in general.** Even with results, it would compare two specific films.
- **Non-metro India.** The design covers six Tier 1 metros only.
- **Real exposure.** Viewers would watch both ads in a survey, not on television or in a feed.
- **Sales.** Purchase consideration after an ad is stated intent, not buying.
- **Nostalgia.** The report expects the older group to have "experienced the original 2016 launch first-hand". Age alone does not show that anyone saw the ad; ask them.

## Reusing the Design for Your Own Two Ads

If you have two creative routes and one budget, this design is a reasonable start. Before you field it:

1. **Write the decision first.** "Should we add a functional hook to the brand film?" is a question you can act on; "test these two ads" is not.
2. **Record ad order for every respondent**, so you can read first-seen scores separately.
3. **Ask whether people remember an ad** rather than inferring memory from age.
4. **Size each cohort you will compare** at 100 or more.
5. **Agree the exclusion rules in advance** and report how many failed each one.

Our [ad testing guide](/blogs/ad-testing-platform-india/) covers stimulus and question choices for video ads. The [Vim vs Beco study](/blogs/vim-beco-brand-switching-study-india/) is a comparative ad test that did report results.

## Running This Study on Hercules

At the planned size, 200 completes on the Plus plan cost 200 × ₹150 = ₹30,000 before any discount. To give each of three age bands 100 people, plan 300 completes: ₹45,000 on Plus, or a 400-response Pro bundle at ₹48,000 if you want a margin for screening.

Our pricing is per response, and the pricing pages do not say how responses removed in quality screening are counted. If you plan to over-recruit, as this study did at 400 for 200, ask before you budget.

## Frequently asked questions

### Which Surf Excel ad performed better?

The study does not say. Its report contains no creative results, every results section is marked "data unavailable", and the report says data collection is the next step.

### What is proto-monadic ad testing?

Each respondent sees one ad and answers about it before seeing the next, with ad order randomised across the sample, and then compares them. First-seen scores are untouched by the other ad; second-seen scores carry an order effect that randomising balances but does not remove.

### Why were 200 of 400 responses excluded?

The report says they failed attention checks and data-quality screening, but gives no breakdown by reason. It also says elsewhere that the dataset has 0 respondents, so the screening figures may describe a plan rather than completed fieldwork.

### Is a paired t-test right for this design?

For comparing Ad A with Ad B among the same respondents, yes. For comparing younger with older parents, use an independent-samples test or a regression, because those are different people.

### How many respondents does a two-ad test need?

Enough for each group you will compare. The report reads any group under 100 as directional, so three age bands need about 300 completes after screening.
