Cadbury’s Shah Rukh Khan-My-Ad: from brand asset to retailer service

One campaign, three different jobs

Wavemaker describes a commercial problem behind the generosity: Cadbury needed local retailers to stock Celebrations as well as shoppers to consider festive gifting. That makes this a useful partner-marketing case, not simply an example of personalized video.

The advertiser’s job was to make its generosity positioning useful at the point of purchase. The retailer’s job was to attract relevant customers without commissioning a costly campaign. The shopper’s job was to find a credible, convenient place to buy. These interests overlap, but they are not interchangeable. A retailer can enjoy its film while receiving no extra customers; a shopper can visit a featured shop without buying Cadbury. Each participant therefore needs a separate success measure.

From a local mention to a usable advertising asset

The 2020 predecessor supplied the starting point, not the results for this case. WPP explicitly dates the Shah Rukh Khan update to 2021. The public accounts establish the following sequence, but do not disclose an exact production calendar.

• 2020: the earlier campaign placed local-store information into advertising. This is background only.
• 2021 production: generative face and voice adaptations made the ambassador say store names. The One Show describes four category-specific digital avatars.
• 2021 distribution: location determined which nearby businesses appeared in served ads. A microsite also enabled users to make downloadable versions for social sharing.
• 2022 and later recognition: awards followed the campaign. Ogilvy documents two specific Gold Lions in 2022 and a Creative Effectiveness Grand Prix in 2023. Award year is not launch year.

Why saying the shop name changes the proposition

The personalized name connects two scales: a nationally familiar performer and a business someone can actually visit. The creative payoff is the unexpected specificity. For the retailer, it also turns an otherwise distant brand asset into something worth showing customers.

The important design question is what remains valuable after the novelty fades. If viewers remember the technology but cannot identify the shop, category or next action, the execution has entertained without completing its commercial job. A smaller adaptation should keep the useful information legible rather than spend its entire attention budget demonstrating a technical trick.

A generator is only one part of distribution

WPP identifies a planned media effort involving Wavemaker, Google and YouTube. Retailer sharing added another route. These routes should not be collapsed into a claim that the campaign spread entirely for free.

The useful chain is invitation, completed asset, approval, publication, relevant exposure, then customer action. Each transition can fail. Incorrect location data wastes relevance; a confusing download step loses partners; unreadable shop details waste exposure. A store with no stock cannot fulfill demand. A campaign team should assign an owner to each transition, rather than declare success when the generator works. The retailer’s willingness to publish is an activation problem, distinct from the viewer’s willingness to visit.

Keep the numbers attached to their definitions

WPP reports 130,000 created ads, 2,000 stores and 94 million YouTube/Facebook views. These are different units, not stages of a verified conversion funnel. Wavemaker separately describes more than 2,500 business owners. Neither public page explains how those populations reconcile.

The One Show entry mentions 35% business growth without supplying a sufficiently clear denominator, comparison period or attribution method. It should not become a claim of incremental retailer sales. Likewise, combined platform views cannot establish unique local shoppers. The cited public material supplies no auditable spend, production budget, customer-level deduplication or controlled sales-lift design. It supports a documented activation, not a forecast of another team’s return.

What else could explain an apparent win?

Celebrity attention, paid exposure, festive demand, reopening and existing retailer relationships could contribute independently of personalization. Shops willing to create and share an ad may also be more active marketers than shops that decline. Comparing those groups after launch would mix the treatment with selection. Even a strong campaign-level result would not tell us whether the spoken name, the local targeting, the generous offer or the distribution effort did the most work. Separate those questions before designing a replication.

Test a partner service before building a platform

Recruit willing local partners for one genuine shopping occasion. Make an original short film with a consenting host, using controlled shop-name, location and offer panels. A spoken personalized name is optional. Do not clone a celebrity or imply that the host personally endorses every store.

Agree in writing on channels, duration, editable fields and any synthetic use. Obtain retailer approval of its details, disclose the commercial relationship, and label synthetic adaptations clearly if used. Prepare a generic version with the same host, duration, offer and destination. The test asks whether local naming adds value beyond an otherwise comparable asset, rather than whether any advertising beats silence.

Make approval and correction part of the product

Use a simple partner register and manual production first. Automation becomes useful after repeated work has been understood.

• Partner owner: confirm shop spelling, category, location, opening hours, offer validity and permission to participate.
• Creative owner: check every export on a phone, including captions, text contrast, pronunciation and destination links.
• Activation owner: provide the approved file and a short posting guide; record publication separately from download.
• Support owner: offer correction and withdrawal routes, expire old offers, and keep permission records access-limited. Collect only the information the pilot needs.

Choose the comparison before seeing the result

Where possible, randomly assign comparable, non-overlapping delivery units to generic or shop-specific creative, keeping spend, timing and targeting comparable. If partner-level assignment is necessary, balance baseline activity and record non-publishing partners in their assigned group. Organic posting differences remain a limitation; do not call that a clean creative test.

Predefine one primary outcome, such as tracked offer redemptions per assigned partner over the same window. Record exposure, qualified store-page visits, publication rate, errors and staff time as separate diagnostics. Use aggregate reporting and disclose tracking. With few partners, report uncertainty and operational learning rather than a definitive lift percentage. Pause for rights or accuracy problems. Expand only when partners use the assets and the additional effort is justified by a credible improvement over the generic version.

Three exercises for a better brief

• Write three briefs: one sentence each for advertiser, retailer and shopper. Give each an observable success measure and one possible conflict with the others.
• Draw the activation funnel from invited partner to redeemed offer. Mark the person responsible, required evidence and most likely failure at every step.
• Audit a result slide: separate outputs, exposure, behavior and business outcomes. Rewrite its strongest causal claim using only what the evidence actually establishes.

How to read the evidence behind this case

Use WPP for the edition boundary and reported totals, Wavemaker for activation context, The One Show for entry credits, and dated Ogilvy announcements for awards. These are participating-company accounts or submitted award material, not independent impact audits. The explanations and pilot above are original analysis.