AI in marketing strategy

AI in Marketing Strategy: Better Decisions, Not Just More Content

A practical AI marketing strategy for Indian brands: turn customer signals into creative tests, smarter media decisions, inventory planning and profitable growth.

By Ish Batra · · 7 min read

AI dashboard turning shopper insights into ecommerce revenue opportunities

Most brands are using AI to make more content. The bigger opportunity is to make better marketing decisions.

DS Group, WPP Media and MICA launched DCODE 2.0 on 17 September 2026. The playbook places AI across search, creative, paid media, e-commerce and performance marketing. It got me thinking about what an AI in marketing strategy actually needs to deliver for a brand.

AI is becoming part of how marketing teams work every day. Yet many brands still measure their progress by asking: How many creatives did we generate? How much production time did we save? How many AI tools are we using?

Those numbers tell us whether production got faster. A marketer needs another answer: did AI help us understand customers, make stronger decisions and grow profitably?

Here is the framework I would use for an Indian brand selling through its website, marketplaces or quick-commerce platforms. This is my interpretation and practical application, not a framework attributed to the DCODE 2.0 authors.

1. Find customer insights in reviews and support conversations

Start with what customers are already telling you:

  • Product and marketplace reviews
  • Search queries
  • Support conversations
  • Reasons for returns
  • Comments on ads and social content

Ask AI to group repeated objections and questions. Then read the actual customer messages behind each group. A neat summary is useful only if the source material supports it.

What problem is the customer trying to solve? What stops them from buying? Which words do they naturally use? Which product claims cause confusion? Why do some products come back?

For example, if shoppers repeatedly ask whether a kurta includes the dupatta, that is a product-page clarity problem before it is an ad-targeting problem. Fixing the listing may be a better first decision than creating ten more ads.

Time from a customer signal to a specific decision the team can test.

2. Test creative ideas instead of producing more ads

Customer insights should lead to different messages for different motivations.

A fashion brand could test an occasion-led concept against comfort, price, craftsmanship and social proof. Generating 50 visual variations of one message is easy with AI. It also leaves the central question unanswered: which reason to buy actually moves the customer?

Brief the team on the idea behind each ad, who it is for and what you expect it to change. Where budgets permit, compare distinct concepts under reasonably similar conditions. Look beyond clicks to orders, new customers and contribution after ad costs.

What the brand learns from each creative concept and whether a winning concept improves a business outcome.

3. Give automated media buying better business signals

Ad platforms can adjust bids, audiences and placements much faster than a person can. They still optimise for the signals and objectives available to them.

If every purchase looks equally valuable in the ad account, the system may bring in more buyers for discounted or low-margin products. Platform ROAS can improve while the business earns less per order.

Where the platform supports it and your data is reliable, use accurate purchase values and appropriate customer signals. Outside the platform, review new versus returning customers, product margins, cancellations, returns and repeat purchases. Do not assume the ad dashboard contains your full profit calculation.

Contribution profit and the value of customers acquired, alongside platform ROAS.

4. Connect demand to stock and operations

Marketing decisions affect assortment, inventory, regional pricing, promotions and bundles.

Suppose demand for one SKU starts rising in Bengaluru. Increasing ad spend before checking stock could buy clicks for a product that soon becomes unavailable. The campaign might look promising briefly, while customers see stockouts and the team wastes media budget.

A more useful sequence is: Demand signal → stock available locally → relevant creative → media budget → margin check.

That matters especially when a brand sells the same product through D2C, marketplaces and quick commerce, each with different fees and fulfilment economics.

Profitable sell-through by product and location, alongside campaign revenue.

5. Check if reporting gains mean real growth

AI can help platforms attribute conversions that were previously missed. A higher reported number does not, by itself, prove that ads created more orders.

Imagine one platform’s reported conversions rise from 8,000 to 9,500, while total company orders remain at 11,000. This is a hypothetical example. The platform may have become better at recognising purchases that already happened. We would need more evidence before claiming 1,500 incremental orders.

Compare ad reporting with backend orders and returns. For important budget decisions, use an appropriate holdout or geographic experiment. Larger advertisers may also use marketing mix modelling. Then ask whether the change produced incremental contribution profit, not just a larger attributed sales figure.

6. Use a simple AI marketing decision loop every week

My view is that the first phase of AI in marketing was about producing work faster. The more valuable phase is improving the decisions a team makes with what it learns.

Customer signal → insight → creative hypothesis → controlled test → commercial result → next decision.

That loop is more useful than producing another 100 ads in a few minutes. It also gives a marketer a clear question to ask after each test: what will we do differently next week?

If AI doubled your team’s content output tomorrow, would it also double what your team learns about the customer?