How Attribution Works for TV and Connected TV
TV doesn't produce clicks, so click-based attribution misses it. How TV attribution uses ACR data and household matching, where it falls short, and how to measure TV's real impact.
Digital marketers have long been able to follow a customer from ad to purchase, at least within the channels that produce clicks. Television has never fit that picture. Someone sees a commercial on the living-room TV, picks up a phone an hour later and searches for the brand. Click-based reporting gives the sale to search, and TV doesn't appear in the journey at all.
For years, TV measurement relied on ratings, broad audience estimates and brand surveys. Useful for planning, but far from the channel-level evidence digital teams expect. As connected TV has grown and more TV is bought programmatically, marketers want TV measured with the same rigor as every other channel.
Why TV breaks standard attribution
Multi-touch attribution splits credit for a conversion across the touchpoints it can track. It works when each touchpoint leaves a trace on the path to purchase: a click, a tagged visit, a known user.
TV leaves almost none of those traces. There is no link to click, the viewer is often on a different device when they act, and the response can come days or weeks later. A model built on clicks will consistently give TV too little credit and give the channels that capture the demand, such as search and direct traffic, too much.
How TV attribution works
Modern TV attribution rebuilds the missing link in three steps.
1. Count TV exposure like a digital impression. Many smart TVs include automatic content recognition (ACR), which identifies what is on screen, including commercials. Aggregated across many sets, ACR data shows which households saw an ad and when. For connected TV, ad servers record impressions directly. Either way, the goal is to measure TV exposure in a digital-like way rather than relying only on ratings estimates.
2. Connect exposure to response. Exposed households then have to be linked to what happened next: site visits, app installs, sign-ups and purchases. The usual method matches the household's IP address, seen through the TV or streaming device, with visits from phones and laptops on the same home network. For linear TV, a second method looks for short spikes in web traffic in the minutes after a spot airs. Both depend on tracking web activity accurately and at scale, which means connecting the TV side and the website side carefully.
3. Analyze it the way TV is bought. TV is planned and bought by network, daypart, program, reach and frequency, not by keyword or audience segment. Analysis that reports results in those terms, such as which networks and dayparts drive response, and at what frequency the response levels off, is what lets a TV buyer act on it. Applying digital metrics like click-through rate to TV produces numbers nobody can use.
Done well, this puts TV on the same page as other channels: exposures, responses and cost per response, by network and by creative.
Where TV attribution falls short
TV attribution is a real step up from ratings alone, but it has limits worth understanding before you move budget on it.
- It only sees part of the audience. ACR covers only the TVs that have it switched on, and IP matching misses households where devices can't be linked. Results are projected from a sample.
- Matching is imperfect. Shared and changing IP addresses, VPNs and mobile networks all weaken the link between a TV household and the devices that later convert.
- It measures exposure, not cause. Households that see more TV ads may watch more TV, live in targeted areas or already be in the market. Showing that exposed households bought more doesn't prove the ad made them buy.
- It still splits credit. Like any attribution, it divides credit among touchpoints. It doesn't say how many sales would have disappeared if the TV campaign had not run.
- Long-term effects get missed. Short response windows capture the immediate reaction but not the brand-building that pays off months later.
Privacy rules add a further constraint. ACR and IP matching depend on household-level data, and the rules for collecting and sharing it continue to tighten.
Measuring what TV actually adds
The question that matters for budget is incrementality: how many sales happened because of TV that would not have happened otherwise. Two approaches answer it directly.
- Marketing mix modeling works from aggregate spend or impressions and sales over time, so it includes TV without needing clicks, devices or households. It estimates TV's contribution alongside every other channel, including the search and direct sales that TV often drives.
- Geographic tests run TV in some markets and hold it back in comparable ones, then compare total sales.
The best practice is to use them together. Attribution data on networks, dayparts and creative helps optimize the TV buy itself. Incrementality measurement tells you how much TV is really worth and how much to spend. Our posts on why CTV success starts with measuring impressions, not clicks and why CTV must be measured like television go further into how to plan and measure it.
TV has always driven sales that click-based reports can't see. The goal of TV measurement isn't to give TV more credit. It's to give it the right credit, so it competes for budget on the same evidence as everything else.
Frequently asked questions
Can multi-touch attribution measure TV advertising?
Not on its own. Click-based attribution can't see an ad that no one clicks. TV attribution adds the missing piece by recording which households were exposed to an ad, then linking those households to later website visits or purchases.
What is ACR data?
Automatic content recognition (ACR) is technology built into many smart TVs that identifies what is playing on the screen, including ads. Aggregated across many TVs, it shows which households saw a given commercial and when, which lets TV exposure be measured somewhat like a digital impression.
How is a TV ad linked to a website visit or sale?
The most common method matches a household's IP address, seen through its smart TV or streaming device, to visits from other devices on the same home network. Another approach looks for short-term spikes in web traffic in the minutes after a linear TV spot airs.
What are the limits of TV attribution?
It only sees households in the ACR or device panel, matches are imperfect and it shows correlation between exposure and response, not cause. People who see more TV ads may already be more likely to buy. Measuring incrementality, the sales that would not have happened without TV, needs a model or test built for that question.