The importance of data automation in retail media – Part 2

In our previous post, we discussed retail media networks (RMNs) and the opportunities they present for advertisers as a growing industry – with the retail media industry expected to grow 25% annually over the next five years.

In this article, we’ll cover the current landscape of retail media networks and why companies will need to automate their data operations as the industry grows.

The current landscape of retail media networks

Amazon launched its network in 2012 and remains one of the largest RMNs today. Since then, a whole host of major retailers have realized the potential of the industry, achieving gross margins of 70–90% for onsite media, and 20-40% for offsite media.

Some of the largest RMNs – with tens of millions of average unique monthly visitors – include Walmart Connect, Roundel (Target), Retail Media + (Home Depot), and Wayfair in the US; and Tesco, Boots, Sainsbury’s and Ocado in the UK.

In 2022, Amazon and Walmart reported results specifically for their advertising business for the first time, revealing just how much retail media is supporting their broader strategies. For example, Amazon Advertising makes up 68% of Amazon’s global profits, and Walmart Connect makes up 12% of Walmart’s.

However, the retail media landscape doesn’t just comprise big retailers – smaller, more niche retail media networks are also contributing to the industry’s growth – and are in fact growing at a faster pace, according to the BCG Commerce Media Survey.

By 2026, niche networks (<$100 million in projected commerce media revenue) are forecast to see 49% compound annual growth (CAGR) rates, and category specialists ($100 million–$1 billion) will see 32% CAGR. That’s compared to 22% CAGR for Amazon and 27% for other leaders (>$1 billion).

However, with this growth comes a growing need to implement scalable data operations.

Why automated DataOps is critical for RMNs

As mentioned in our last post, retail media networks are made up of onsite media (the retailer’s own collection of digital channels) and offsite media (using the retailer’s first-party data to target ads to shoppers on third-party digital channels).

Therefore, retail media networks inevitably create multiple sources of data to track, measure, and report on. As an RMN grows, so do the streams of data coming from these sources – which quickly becomes unsustainable to manage manually.

For any retailer considering setting up their own retail media network, the best approach is to build an automated data infrastructure now, in order to prepare for the growth of the business and the data that comes with it. But how?

One option is to hire a team of developers to build a bespoke data automation solution. However, this method can take a huge amount of time and resources. For instance, taking this approach would have cost Financial Times 6-9 months of risky customer development time.

Another option is to adopt an existing enterprise solution as a DataOps partner. This method reduces implementation time from months to weeks, without the need to hire a whole team of developers. Instead of overburdening existing data engineers, this approach in fact empowers them – and the business teams – to do much more with the retail media data at their disposal.

At Switchboard, we automate the data pipeline using the ETL method (Extract, Transform, Load) – which involves aggregating data from hundreds of disparate sources using pre-built connectors, transforming this raw data into usable foundational data, and loading it into any cloud destination ready for analysis with business intelligence tools.

When it comes to choosing the right DataOps partner for a retail media network, you should ask the following questions:

• Can the platform support both first-party onsite data sources, and offsite data sources?
• Can the platform help us manage our data as we scale and grow?
• Will we still own and control the data?
• Can the platform help us track and understand the data to improve our ROI and efficiency?

Answering these questions will help you narrow your options and provide clear criteria for your DataOps when the time comes.