# Bata Fires Ad Agencies for AI: Can Algorithms Really Sell Shoes Better and Cheaper? > Source: > Published: 2026-09-03 10:21:48+00:00 # Bata Fires Ad Agencies for AI: Can Algorithms Really Sell Shoes Better and Cheaper? The brief walked into Bata’s marketing department. Nobody did. An artificial-intelligence agent read the campaign calendar, generated the creative, examined its own work, published the approved version and waited for consumers to teach it what to do next. The copywriter did not arrive late. The art director did not defend a difficult idea. There was no agency retainer to pay because there was no agency on retainer. The agency has been given the boot. The bot now has the brief. Bata India has rebuilt its everyday creative operation around autonomous AI agents supplied by marketing-technology company Zocket. Chief strategy and business development officer Badri Beriwal reportedly said that system now handles work once distributed among creative agencies, social-media partners and production houses, while Bata’s internal marketers retain control of strategy and approval. ## What Rahul Gandhi Wants 28 Aug 2026 - Vol 05 | Issue 35 He is desperately courting the angry youth. That won't be enough to credibly challenge Modi [Read Now What Rahul Gandhi Wants](/magazine/what-rahul-gandhi-wants) “One hundred per cent of our creatives are now created without any agency,” Beriwal reportedly remarked. The claim is larger than another brand announcing an AI experiment. Coca-Cola has generated artwork. Mango has created fashion images. Toys“R”Us has made a film with OpenAI’s Sora. Those were campaigns that used artificial intelligence. Bata is attempting to make artificial intelligence the campaign department. The difference is existential. A machine is absorbing the repeatable parts of the agency until the client begins to wonder why the agency is there. ## Eighty per cent cheaper. But is it better? Bata says work that once required weeks can now be produced in hours. It estimates annual savings of nearly Rs 1 crore and puts the overall reduction in creative-production costs at almost 80 per cent. Locations, crews, large shoots and multiple rounds of agency coordination can be replaced by prompts, models, brand rules and a feedback loop. Agencies may still be hired for selected projects that need human expertise, but the dependable monthly retainer has walked out of the door. For a business of Bata’s scale, the appeal is obvious. The company has nearly 2,000 stores, sells close to 50 million pairs a year and operates brands ranging from Power and North Star to Hush Puppies and Bubblegummers. Campaigns must be adapted across formats, products, regions and retail moments. Much of that is industrial content production wearing a creative job title. AI is extraordinarily well suited to that factory. The timing matters. Bata reported revenue of Rs 978.9 crore in the quarter ended June 2026, up about 4 per cent year on year, while saying its advertising investment had increased by nearly 25 per cent. Spending less on production could move more money into media and serve more locations. Bata’s latest results show why efficiency is part of the growth machinery. The problem begins when efficiency is mistaken for imagination. Beriwal offers a striking proof of success: consumers looking at Bata’s Instagram feed cannot tell which pictures were traditionally shot and which were generated. That may mean the technology has become brilliantly convincing. It may also reveal something less flattering about contemporary advertising. Perhaps AI has not become extraordinarily creative. Perhaps advertising has become extraordinarily predictable. If a feed consists of familiar poses, polished faces, immaculate lighting and products arranged according to an established grammar, a model should reproduce it capably. The more conventional the assignment, the easier the replacement. The machine need only recognise what this category of image usually looks like. The dangerous question is not whether AI can make an ad. It plainly can. It is what agencies were charging for if their output can be recreated through brand guidelines and a feedback loop. AI can measure the cost removed from advertising immediately. "But the cost of lost distinctiveness may take years to appear,” says Ashita Aggarwal, professor of marketing at SP Jain Institute of Management & Research. A brand can become faster, cheaper and more consistent while slowly becoming less surprising. By the time the dashboard detects that loss, consumers may already have stopped noticing. ## When the machine briefs itself An agency, at its best, is an outsider paid to notice what the company has normalised. It can question the brief, resist a safe answer or tell a senior marketer that the product claim is boring. The friction clients dislike is sometimes the source of the idea they could not have produced internally. Bata’s model compresses that distance. AI agents write briefs, create outputs, review them and learn from performance. Human marketers still supervise the system, which is an important qualification. Yet when the same technological loop helps define the question, manufacture the answer and evaluate the result, efficiency can become self-confirmation. The bot knows which red shoe received more clicks. It may not know that the category has begun to look alike. Performance data rewards what worked before. Generative models reproduce learned patterns. Together, they can become a superb machine for walking confidently into the brand’s past. An agency is not valuable merely because it supplies people to produce assets. “Its deeper value is an independent point of view," says Aggarwal. If software creates, scores and improves its own work, the client must preserve a human mechanism that challenges the system’s assumptions, not merely approves its output. The most memorable international AI campaigns have usually contained a conspicuously human proposition. In Heinz’s AI Ketchup experiment, DALL-E repeatedly produced bottles resembling Heinz. The technology became evidence of brand dominance, but agency Rethink supplied the joke. Coca-Cola’s Create Real Magic opened famous visual assets to artists using GPT-4 and DALL-E. Mango’s first AI-generated Teen campaign began with photographs of real garments and passed through design, styling, photography and art teams before selection and retouching. Mango called AI a co-pilot. Toys“R”Us made its Sora brand film with creative agency Native Foreign. The machine performed. Humans decided what the performance meant. Bata’s more radical experiment asks whether that interpretive layer can sit almost entirely inside the company, with outside agencies summoned only when a special project requires them. That may not kill the agency business. It could divide it. Routine production becomes software; rare strategic intervention becomes premium consulting. The agency does not disappear completely. Its monthly invoice becomes a licence fee, while its human talent is rented only for moments the machine cannot resolve. ## The celebrity who need not attend The most visually unsettling part involves famous faces. Beriwal says celebrities and influencers grant AI rights for digital shoots, with generated pictures sent to the talent for approval. Yesterday, the celebrity lent the brand a face. Today, the face can report for work without the celebrity. The business case is legitimate. A physical shoot requires calendars, travel, sets, photographers, stylists and a formidable bill. A licensed likeness can create variations rapidly, provided consent and contractual boundaries are clear. The talent is not secretly deepfaked; the usage is authorised. But endorsement changes when presence becomes infinitely reproducible. AI turns the star into adaptable inventory, able to model more shoes in more locations without experiencing any of them. The image becomes cheaper as its claim to human reality becomes thinner. Consumers may not notice. Bata regards that invisibility as success. The real test is not whether consumers can detect the artificial image. It is whether they remember the advertisement, recognise Bata without the logo and feel anything after the thumb has moved on. An 80 per cent cost reduction is an excellent procurement result. It is not automatically a marketing result. Marketing must demonstrate attention, memory, preference and sales. If every rival gains access to similar models, speed becomes hygiene rather than advantage. The technology that lowers Bata’s costs also lowers the cost of imitation. That is the paradox beneath the triumph. AI makes brand consistency easier at the exact moment it makes brand sameness cheaper. Bata deserves credit for exposing its marketing model to a real test. Human marketers remain accountable. Agencies still have a door for exceptional assignments. Localisation at a 2,000-store scale is a genuine problem, and consumers do not owe production houses a living merely because advertisements were historically expensive. But a century-old brand needs more than flawless repetition. Incorporated in India in 1931, Bata helped build Batanagar and became a neighbourhood institution. Its history is built on manufacturing, reach and trust. AI can learn the colours, angles, typefaces and product hierarchy associated with that inheritance. It can reproduce Bata-ness. Can it invent Bata’s next meaning? The answer will not be found in the Rs 1 crore saved or the assets published. It will appear when Bata produces an idea consumers could not have predicted, competitors cannot copy and a dashboard would never request. Bata built its business by asking customers to sit down, try the shoe and test the fit. Its AI can fit the brief perfectly. Whether it fits the future will take a little longer to discover.