Offline attribution, explained.

How do retail companies measure offline attribution?

There are several ways retail companies measure the effectiveness of their offline marketing. The three most common are:

• Surveys and focus groups. Gathering customer feedback shows how well a campaign resonates and whether it has the desired impact, but it relies on memory and on small samples.
• Sales data. Comparing sales before and after a campaign reveals spikes during the flight. It shows that something changed, not necessarily why.
• Tracking codes. A unique coupon code in a print ad lets you count redemptions. It is precise, but only for shoppers who redeem.

Each method answers part of the question. None of them sees what happens between the ad and the shelf.

Do retailers use modelling attribution?

Many do. Modelling attribution is a statistical technique that estimates how much each marketing channel and campaign contributes to sales or to another outcome. The two main families are attribution modelling, which splits credit across touchpoints in a journey, and marketing mix modelling, which estimates the effect of each channel from aggregated sales and spend over time.

Both help decide where to allocate budget, including budget for offline media.

Is modelling attribution accurate?

It can be useful, but like any statistical technique it has limits:

• It depends on data quality. If sales or marketing data is incomplete or inaccurate, the results are less reliable.
• It carries uncertainty. Results are estimates, with a margin of error that is often not communicated.
• It struggles offline. Without observing the shopper between exposure and purchase, the model has to infer that step.

That is why companies should verify modelled results with other evidence, ideally with direct observation of in-store behaviour.

Examples of attribution models

Imagine a shopper who saw a flyer at home, walked past a screen in the store and then stopped at an end cap before buying. This is how each model would split the credit:

• First-touch gives all the credit to the first touchpoint the shopper interacted with, the flyer.
• Last-touch gives all the credit to the last touchpoint before the sale, the end cap.
• Linear gives equal credit to every touchpoint in the journey.
• Time-decay gives more credit to touchpoints closer in time to the sale.

There are many other variations, such as position-based and data-driven models. Read more in the attribution models article of our Help Center.

What is the most common approach?

Last-touch attribution. It is simple, easy to implement and easy to explain, and it is the default in many analytics tools. Its weakness is that it ignores everything that built the intent before the final touchpoint, which in retail usually means ignoring in-store media altogether.

Closed-loop attribution in store

Shoppermotion observes the missing step. Tags on trolleys and baskets record whether each shopper passed a placement (an impression), slowed down to look (a view) or stopped to engage (a click). At checkout, the trip is matched with its ticket.

Comparing exposed shoppers with a matched group who were not exposed gives the incremental sales of each campaign, and from there its ROAS. No surveys, no coupon codes and no guesswork about what happened in the aisle.