This festive season, brands are approving systems, not just ads

Every festival has its rituals. Even as consumers spend September to November scanning newspaper pages, app surfaces and browser tabs for the best deal on a phone, television or improbable quantities of dry fruits, despite knowing they'll anyway get inconvenient amounts of dry fruits, marketers (and the journalists and industry watchers who observe them) perform a parallel ritual during 'festive': scanning an equally formidable pile of reports to find out what India is buying, how much it is buying, where it is buying it, and what persuaded it to click, er, buy. 

This year, the numbers suggest there is simply more to chase.

Datum Intelligence expects festive advertising expenditure to reach ₹60,000–62,000 crore in 2026, up 13–17% year-on-year and accounting for 34–35% of India’s entire advertising year. Simultaneously, commerce advertising is expected to grow 27%, more than twice the rate of the overall market. 

The consumer journey is proliferating just as quickly as the discounts flashing across your screen(s). Google says nine in ten Navratri and Diwali shoppers who made a purchase researched beforehand, conducting an average 8.3 searches and watching 7.7 online videos along the way. Those using generative-AI chatbots interacted with them 7.5 times on average; 51% said AI helped them make a decision and 38% said it shortened decision time. 

Also Read: Google is putting Search on the front page as AI starts selling ads of its own

Has mobile become the primary digital marketing interface?

AI can make 100 versions of an ad. Who checks version 63?

 InMobi finds 77% of consumers have increased their festive shopping budgets, while four in five want AI-powered shopping assistants. Some 67% browse on their phones while watching television, and 29% expect to make purchases only a week or less before the festival. 

IAS adds another layer to the festive pie: 38% of Indian consumers plan to use AI specifically to research products and gift ideas, up from 32% last year, while 83% identify mobile as their primary online shopping device. 

 All this means a lot of eyeballs on a lot of surfaces, even as brands and platforms rush to fill those surfaces with ads generated at the drop of a prompt.

As exchange4media reported recently, that explosion in creative volume has already created a new bottleneck: AI can produce the 50th or 100th variation far faster than human teams can meaningfully inspect it. This raises the next question: if every asset cannot be approved the old-fashioned way, what replaces that process?

For marketers, this does not merely translate into more ads. It means more combinations of audience, language, city, device, format, product, price and offer, frequently produced while the campaign is already live.

And that is turning festive advertising into a useful stress test for a question AI has made unavoidable: when one campaign can become dozens of different pieces of communication, what exactly does the brand approve?

Brands Speak

At Shapoorji Pallonji Real Estate, the answer increasingly begins before the individual advertisements exist.

CMO Rima Kirtikar says teams sign off on the master creative, template architecture and a defined rulebook. Elements such as RERA registration numbers, QR codes, project names, approved imagery and legal disclaimers remain locked, while copy length, language, format and local context can flex within defined limits. 

“That said, we don't treat it as a blind hand-off,” Kirtikar says. Legal and compliance teams concentrate on “high-risk assets such as new claims, offers, regulatory-sensitive markets and anything that falls outside the template.” 

Her shorthand for the change is telling: “approve the system, audit the output.”

Financial services company Airpay follows a two-layer model. According to Shreyas Deshmukh, AVP, Marketing (B2C), the master creative and its rulebook undergo the most rigorous approval. Core messages, product claims and disclaimers are fixed; language, format, visuals and tone can vary. An adaptation that changes a locked element goes back through full review. 

But then, even automation has a limit.

“A tool can flag that a line has changed, but assessing whether that change could mislead someone about their money remains a human responsibility,” Deshmukh says. 

That distinction becomes especially useful in a festive environment where prices and offers can change rapidly, different regions may receive different promotions, and the same campaign may need to work across short-form video, commerce apps, search and regional-language mobile placements.

Karan Jaju, co-founder and business head at CloudPe, says creative is consequently no longer being treated as a fixed collection of campaign assets. AI lets the same idea travel across audiences, platforms, languages and locations, forcing brands to reconsider how approval itself is structured.

“Automated checks can flag routine issues across hundreds of versions, allowing teams to focus on exceptions, brand fit and legal concerns,” Jaju says. “The real opportunity is to gain speed and scale without losing control over what ultimately reaches the consumer.”

Tech Talks 

The challenge becomes more complicated when the different parts of a campaign are produced through separate systems. Ravi Dugal, President of Flipick, says prompts, scripts, storyboards, assets, translations, voice-overs, text-to-speech and rendering can all sit in disconnected tools, leaving reviewers to reconcile the finished outputs at the end.

In one recent retail campaign handled by Flipick, 12 versions were produced across English, Hindi and Marathi and four different sizes. Automated audio checks found every English version failing across all four shots, with only 43% to 69% of the script making it into the audio. Each Hindi version contained one excessively rushed shot, while the Marathi versions passed. 

Dugal argues that moving those stages into a single pipeline allows checks to happen both before and after rendering. “Most problems can be flagged, and many fixed, before a person ever reviews the output.” 

That means QA itself begins moving upstream.

Kavir Advani, co-founder and lead growth strategist at content and data advisory firm NorthSight, argues that approved claims, offers, tone of voice and regional requirements should be defined before the first variation is created. Automated systems can then police what is objectively checkable. 

“People can focus on areas that require human judgement, such as tone, cultural relevance, and whether the content feels right for the local audience,” Advani says. 

At Tonic Worldwide, automated checks in its GROTH platform similarly flag potential problems so reviewers can concentrate attention where it is required. Founder and COO Samir Asher says branding, key assets, translations, offers, disclaimers and platform crops still need human scrutiny. 

“A version can look perfectly fine at first glance and still be wrong,” he says, pointing to city-specific offers, translations that change meaning and crops that remove important information. 

Some teams are pushing those safeguards even further toward the beginning of production. J7 founder and CEO Venkat Mallik suggests “prompting a checklist into the AI campaign before the assets are generated” or constraining the system to a particular creative style. 

Bottom Line

None of this quite eliminates the old-fashioned final pair of eyes.

Sailesh Grover, business head at OSL Luxury Collections, which represents Corneliani, notes that rule compliance and contextual correctness are not always the same thing.

“An asset can follow all the basic guidelines and still not work in the context where it is being used,” Grover says. “So there needs to be some human review before anything goes live.” 

That is perhaps the more consequential change beneath this year’s festive advertising rush.

AI is usually described as making production faster. What it is also doing is changing the object being approved.

The person signing off may increasingly be asked not to scrutinise every banner, film and language adaptation individually, but to establish the boundaries within which they can be made: what the machine may change, what it may never touch and which exceptions still demand judgment.

When roughly a third of India’s advertising year is compressed into the festive season, that is more than a workflow tweak.

The machine can make another festive ad in seconds. The harder job is deciding, in advance, what it is not allowed to make.