Sprite's AI Localisation: Right Strategy, Wrong Landmark
Sprite (Coca-Cola) · One Film, Many Indias (AI hyper-local campaign)
Sprite broke the Indian FMCG habit of one country, one film, one story by using AI to generate city-specific versions of a single ad fronted by Sharvari Wagh, varying only how each city's heat looks while holding one brand narrative constant. It is worth studying because it is a visible early test of AI-driven creative versioning by a major Indian brand, and because a reported slip, Sydney's Harbour Bridge appearing in a Kolkata frame instead of the Howrah Bridge, turned it into a live case on the limits of automated localisation. Industry reception was mixed.
The objective
Sprite wanted to raise the relevance of one national campaign across very different urban markets without fragmenting the brand story. The Indian FMCG default is a single film run identically everywhere, which buys consistency and scale at the cost of local resonance. Sprite tried to keep both: one narrative with Sharvari Wagh as the single face, plus creative that felt specific to each city's experience of heat. The stated intent was to lift creative precision to the level media planning already operates at, so the version a viewer sees is matched to their market the way media buys are, and to shift the brand's competitive focus from media buying towards depth of consumer insight.
The insight
Heat is the category's core occasion, but heat is not one experience across India. Kolkata's is heavy, humid and monsoon-like; Mumbai's is a sticky coastal summer thickened by dense traffic; Delhi's is dry and intense. A generic hot day flattens these into a lowest common denominator, so the refreshment promise lands with less force everywhere. If Sprite is relief from heat, showing each city its own heat should make the relief feel more personal and more believable. That reframes localisation: not translating dialogue or swapping celebrities, but adapting the environmental context of the problem the product solves, while story, protagonist and promise stay identical.
The strategy
The strategy was one narrative, many contexts. Hold the film, the storyline and the face constant, and use AI to vary only the environmental backdrop per city, so Sprite is always the answer to that market's specific heat. This breaks the one country, one film norm while keeping its benefit, a single coherent brand meaning. The rationale was to align creative precision with media-planning precision: media has long been targeted city by city while creative stayed uniform, and this closes that gap. It also moves where the brand competes, from who buys media more efficiently to who reads the consumer more closely, with AI as the production method that makes per-city versioning affordable enough to attempt.
The execution
Execution built one master film that AI adapted into city-specific versions. The Kolkata edit rendered heavy, humid, monsoon-like conditions; Mumbai showed dense traffic and a sticky coastal summer; Delhi depicted dry, intense heat. In each, Sprite plays the same role, relief from that city's particular discomfort, with the narrative and Sharvari Wagh unchanged. The same execution exposed the method's fragility: a Kolkata frame reportedly showed Sydney's Harbour Bridge rather than the Howrah Bridge. For a campaign whose whole premise is that AI can make creative feel locally true, a wrong landmark in one of its three showcase cities cut against the promise, and it became the detail most discussed in trade coverage.
The media mix
Trade coverage frames this as a creative and production innovation rather than a media one, and channel-level media detail is not in the public reporting, so it should not be invented. What is clear is the intended link between the two: per-city versions only pay off if delivery is targeted enough to serve the right version to the right market, so the idea presumes the geo-targeted, city-level media planning that already exists in India. The stated logic of matching creative precision to media-planning precision effectively treats the media plan as the distribution grid for creative variants. The honest framing for a student is that the media architecture enables the idea, but the documented specifics concern the creative system, not the buy.
The results
No business or campaign metrics are public, so no claim on sales, reach or lift can honestly be made. What is documented is the reception, and it was mixed. On one side, the work was read as a genuine break from the one country, one film convention and a visible test of AI-enabled versioning by a major brand. On the other, the reported Sydney Harbour Bridge in a Kolkata frame handed critics a concrete failure of the very authenticity the campaign sold. Commentators also asked whether this is real innovation or just faster green-screen and stock-footage work, and whether multiple city versions strengthen relevance or fragment the creative. The campaign is currently about as known for the debate as for the films.
Why it worked
As strategy the idea holds up on standard frameworks, with one caveat. On STP it keeps segmentation and targeting at city level while positioning stays fixed, Sprite as relief from heat, with only the depiction of heat varying, which protects distinctiveness while sharpening relevance. On category entry points, heat occasions differ by city, humid Kolkata, sticky Mumbai, dry Delhi, so showing viewers their own entry point should make Sprite more mentally available in that moment. It also converts a media capability, city-level targeting, into a creative advantage, which is the campaign's own stated rationale. The caveat is decisive: these mechanisms only pay off if the local detail is accurate, which is exactly where the execution reportedly stumbled.
Watchouts
The watchouts here are documented, not hypothetical. First, authenticity risk: the reported Kolkata frame showing Sydney's Harbour Bridge instead of the Howrah Bridge shows AI-generated local context can be confidently wrong, and one wrong landmark can undo the credibility the whole approach rests on, because locals spot errors about their own city fastest. Second, the innovation question: critics asked whether this is genuinely new or just faster green-screen and stock-footage substitution, which implies the claimed insight depth may be thinner than the rhetoric. Third, fragmentation: more versions of one film can dilute rather than build the creative if the variations add cost and risk without adding meaning. Fourth, reception risk: industry response was mixed, so citing this as an unqualified success would misread the record.
- 01Hyper-localisation here works by adapting the context of the consumer's problem, each city's specific heat, while positioning, narrative and brand assets stay constant.
- 02AI's strategic value in this campaign is closing the gap between city-level media precision and one-size-fits-all creative, shifting the contest from media buying to insight depth.
- 03Authenticity is the whole bet: one wrong landmark, Sydney's Harbour Bridge in a Kolkata frame, can undo the credibility per-city versioning is meant to build.
- 04More versions do not automatically mean more relevance; each variant must add local meaning or it only adds cost, risk and fragmentation.
- 05With no public results, present the campaign by its documented mixed reception and strategic logic, never by invented numbers.
In a GD on AI in marketing or personalisation at scale, use Sprite as the two-sided example: real strategic logic, real execution failure. Set it up in one move, then let the landmark error carry the critique. A sayable line: 'Sprite got the strategy right, matching creative precision to media precision, but the Kolkata cut reportedly showed Sydney's Harbour Bridge instead of the Howrah Bridge, which tells me AI localisation fails exactly where it claims to win, on local truth.' In a PI, extend it: keep the versioning architecture but add human, city-level verification before release, and judge such work on insight depth, not production speed. Quote no metrics, because none are public.
Sourced from the trade press
Full credit to the original publishers. This decode is MarQuest's independent read of their coverage.