CTV and retail media

When Viewing Data Meets Buying Data
Connected TV knows what people watch. Retailers know what people buy. For most brands, those two kinds of knowledge have sat in different places and never met.
Retail media is beginning to change that. It lets brands advertise using a retailer's own shopping data, and when it is joined with CTV, it promises a clearer line from the ad on the screen to the pack in the basket. It is a promising idea, and a harder one to carry out than it first appears.
Consider why it matters. A brand runs a strong CTV campaign, the reach numbers look healthy, the creative earns praise, and then someone from finance asks how many packs it actually sold. Connected TV has always been better at earning attention than at proving what that attention was worth.
Retailers, from large marketplaces to quick commerce apps to store chains with loyalty programmes, hold detailed records of what people buy. They now sell advertising built on that record.
Bring the two together and the idea is simple to state. Choose viewers based on how they shop, show them the ad on the biggest screen in the house, then check whether those households bought afterwards.
Each side covers the weakness of the other. CTV is strong on storytelling and reach, and weak at connecting an ad to a sale. Retail data knows exactly what was sold, and has little room for a story. Consider a snack brand that wants to reach households who already buy in its category but rarely buy its own pack. Purchase data can find those households, CTV can put a strong story in front of them, and the follow-up purchase data can show whether the mix worked. That is a genuine step forward, and it explains the growing interest among marketers.
It also explains why enthusiasm needs a little discipline. Joining two very different kinds of data is harder than the presentations suggest, and the difficulties sit in places that are easy to overlook.
The first complication is identity
A television belongs to a household. A shopping account belongs to a person. The two have to be connected, and the connection is rarely clean. The match usually relies on signals such as email addresses, phone numbers or device information, and every link in that chain can slip. A family may share one screen and several phones. Someone may shop from an office laptop and watch from the sofa at home.
The outcome is a match rate, and it deserves as much attention as the sales result that follows. If only a modest share of the audience can be matched, the measured effect describes that share and may say little about everyone else.
Privacy sits right beside this. Purchase history is personal, and brands should ask how consent was collected and how the data is protected. Many retailers use clean rooms, which are secure environments where two parties compare their data without handing raw records to each other. This protects shoppers, but it also limits how much the brand can inspect for itself. Trust in the process has to make up the difference.
The second is who marks the homework
A retailer earns money by selling advertising, and it also reports on how well that advertising worked. However honest the people involved, that is a conflict of interest, and it deserves a plain acknowledgement. This is no criticism of retailers, whose data is valuable and whose platforms keep improving. It is a reminder that anyone marking their own paper benefits from an outside check.
The measurement itself has a well-known trap. A shopper who was already going to buy a particular brand of tea will appear in the results as someone who bought after seeing an ad. The ad may have changed nothing. Attribution that credits whatever happened just before the purchase flatters every channel that sits close to the till, and CTV can look better than it deserves or, as often, worse.
The stronger approach is a holdout test. A comparable group of households does not see the ad, and the two groups are compared. That shows what the campaign added, rather than what was already on its way.
The third is what the data cannot see
A great deal of retail in India still happens offline, in neighbourhood shops and stores that no app tracks. Online purchase data covers only a slice of shoppers, and often the more urban and digitally comfortable ones. A campaign judged on that slice alone can look stronger or weaker than it really was.
Time plays its own tricks. A viewer might see an ad on Monday and buy the product a month later, when the old pack runs out. A short attribution window will miss that entirely. It also pushes CTV into the role of a conversion channel, when its larger strength is building the preference that makes the later purchase easy.
What a brand can do
None of this is a reason to stay away. It is a reason to ask better questions before the campaign starts. Who holds the data, and how was consent gathered? What is the match rate? Will there be a holdout group, and what result will count as success? How long is the attribution window, and why was that length chosen? Will the brand receive exposure data that it can check independently?
Agreeing these points in advance turns measurement from an argument after the fact into a shared plan. It also helps to treat sales results as one view alongside reach, frequency, and brand measures, rather than the only judge. A brand new to this can start small, with one retailer and one product, and treat the first campaign as a test with modest expectations.
Viewing data and buying data belong together. The brands that gain the most will be the ones that insist on knowing how the introduction was made.
.png)


