The data that’s driving programmatic buying

Marketers hear no end of reports on the growth of programmatic advertising and opinion columns on why they should adopt it, but how many understand what data they should use to power their buying? We asked brands for their advice.

As programmatic trading continues to gather pace, brands are becoming more savvy about the array of data that can be deployed to maximise the efficiency and creativity of programmatic advertising.

We asked a number of brands what data they are using and the lessons they have learned – whether it be utilising first-party data better, exploring the potential of second-party data shared by partners, or knowing when to be wary of third-party and when to embrace it.

First-party data

Travel group TUI is one company that has adopted programmatic as part of its marketing mix, and head of media Sammy Austin says it has already seen great efficiencies and strong performance gains from the activity it has run to date. TUI is continuing to test new data sets
and first-party data – data generated by its own websites and customer interactions and owned by the brand itself – is key.

“We use our first-party on-site data to deliver dynamic retargeting, serving customers specific product-level advertising based on their behaviour on our site. We can also use our first-party on-site data to build lookalike models of customers, reaching new customers that we wouldn’t have been targeting before,” she says.

Austin points out that this is not just about using data to target customers, but also to exclude certain audiences to reduce wastage and create greater efficiencies. “We shouldn’t be serving ads to customers and encouraging them to book a holiday if they have just booked their holiday with us the week before,” explains Austin.

TUI also uses first-party data relating to consumers’ interactions with pre-roll video ads. “If someone has engaged with our pre-roll content, we can target them with creative that follows on seamlessly, continuing their engagement with the brand,” says Austin. This can be tailored to how much a user has interacted with TUI’s content, such as data based upon video completion rates, allowing the brand to build audiences off the back of consumers engaging with its content.

Behavioural insights gleaned from site activity similarly helps online menswear retailer Mainline Menswear to drive incremental sales from its programmatic activity. Andy Hoyle, head of marketing, says: “We gather insights on which products and pages a potential customer has viewed, and use this data to target them effectively across the web.”

The retailer also uses this data to set frequency caps and daily spend based on how users are interacting with its creative. “We look for the sweet spot of how many impressions we need to serve before we are over-exposing our brand,” says Hoyle. In addition, by reviewing the publishers that deliver the best performance, the company increases its spend on sites that drive the highest conversions.

Hoyle says one of the most interesting data sources the company has been able to tap into, working in partnership with Rakuten Marketing, is tracking and monitoring physical interactions with its banner ads, which demonstrate the engagement levels of its audience. “As an industry, we understand that not all impressions are valid and relevant. Click-through rates are not always the best indicator of performance, so the ability to know that a banner has been seen and interacted with gives us insight into how consumers are interacting with our messages.”

Being able to tie engagement rates to specific websites and prove which publisher has effective positions for a retailer’s banner ads enables ad spend to be optimised. “We know that if we test a new functionality or message in our banners, we can measure how the audience on a certain site reacts by measuring if they are engaging with our ad; we do not only have to rely on click data,” says Hoyle.

Second- and third-party data

It is second-party data, where clients and suppliers or advertisers and media owners share data with one another, that has the most untapped potential. Hoyle says that harnessing and executing second-party data will be the next step that enables Mainline Menswear to continue to grow as a brand.

It is also an area that TUI’s Austin believes has huge potential. “Being able to access data from an advertiser or publisher directly is extremely valuable.”

However, she believes there is some way to go. “Conversations are happening between different parties but there is some work to do on the logistics, the buying mechanism and pricing. Additionally, educating advertisers on the importance and the value that can be gained is key to encouraging sharing and for this to progress further,” explains Austin.

Nestlé digital lead Gawain Owen agrees and gives an example of how the company could benefit from second-party data through its SMA formula milk brand. “There might be an opportunity for us to work with a site that is exceptionally relevant to us, and they would pass their first-party data directly to us for our agency to buy media programmatically to target those people.”

He believes the reason why so few brands are tapping into this source at present is due to a lack of in-house expertise. “Data is not a skill that would traditionally sit with a standard marketer – they are fantastic at doing brand marketing, for example, but they are not necessarily numbers people. Second-party data will come along for certain but that isn’t going to happen in the near future for most people.”

Third-party data is still more commonly used to inform programmatic ad buying. For Nestlé’s Nescafé brand this might mean buying data on people who have shown a propensity to search for or buy coffee.

Nestlé also uses third-party data to inform its location-based programmatic activity, allowing it to communicate with consumers when they are in a particular store via display advertising, or in-app on their mobile device.

“We are seeing a significant rise in engagement, with higher click-through rates when people are in the vicinity of the store,” says Owen. “We take those data signals and use them to drive market share in our business.” He says that in future the company will look to use first-party data, where appropriate, to achieve this.

For TUI, third-party data is deployed to deliver bursts of display impressions when a TUI ad, or a competitor’s ad, is being aired on TV. The company works with third-party providers to harness live weather data too, allowing it to integrate this into its creative in real time.

However, Austin says that although she believes that third-party data is growing in importance and improving, particularly as more advertisers invest, the fact that third-party data sets tend to be modelled on the assumption of a mass audience brings challenges around accuracy.

“When looking at third-party data sets I want to understand how segments have been built, what the methodology looks like and what percentage of the data is modelled versus actual data.” She adds that the cost-per-impression pricing model is another challenge. “There are instances where you will pay more for the data than the media itself; this then makes the accuracy of the data even more important.”

Fact-fuelled creative

As brands learn how best to harness the plethora of data sources available to them, it is opening up greater opportunities for more effective creative. Austin says: “Using multiple data sets means we really can deliver custom creative based on what stage a customer is at in their journey and how they are engaging with our brand.”

In short, programmatic allows brands to target the right customer at the right time in the right place and at the right price. Using data to inform the creative completes this.

The data insights garnered from programmatic can feed back into brands’ creative, allowing them continually to hone it. As Nestlé’s Owen says: “We can understand which particular data segments we are buying and whether the creative appeals more to men or women or different parts of the UK, for example, and ultimately that allows us to optimise our media budget.”

He also shares the example of running YouTube ads and being able to access second-by-second viewer data. “You get enough data from the back end to understand where people drop out. We give that back to the creative agency and we make adjustments.”

Hoyle says that Mainline Menswear has been looking at how it can serve specific messages based on how long it is since the user visited its site. “For example, a user who visited the site 60 days ago should have a different experience to one who came to the site 120 days ago.”

Programmatic is about more than advertising, as Owen points out it is about using those data signals to “become smarter, to make sure your message is more targeted, and to drive greater efficiency within your media buying.

Having never bought our display advertising programmatically before we were keen to see what efficiencies buying this way could deliver for a data-rich campaign like our Game of Thrones new series launch, and it didn’t disappoint.

We worked with Infectious Media to run the launch campaign across display, video and social channels and across devices to deliver engagement at scale.

It combined the most relevant contextual data sources, such as websites reviewing Game of Thrones, Game of Thrones fan sites, the TV Guide and Radio Times, and third-party data sources such as Experian, to recognise the Now TV target audience. Infectious Media shared data across the three channels and the insights were used to update keywords and audiences that were being targeted.

A key focus for the campaign was video. Technology allowed us to run a programmatic video campaign that used a detailed set of keywords to target relevant YouTube videos that were trending in real time on Facebook and Twitter. Trending videos were established by measuring the number of comments, likes and shares they received. We saw our video completion rates soar as a result.

We also used Experian demographic data to understand existing Now TV users’ behaviour, which helped us understand what websites, time of day, day of week and devices they were active on.

Audience ‘lookalike’ models were created using this data and were used to reach relevant new audiences. Our video and display activity was also integrated via sequential messaging in order to understand what the best performing creative combinations were.

We set ambitious benchmarks based on previous direct media buying activity and we quickly saw performance improve throughout the campaign, going on to out-perform nearly all previous activity. Overall, we found the results and detailed insights delivered from our programmatic activity invaluable.

Sammy Austin and Gawain Owen will be speaking at Marketing Week’s ‘Get with the programmatic’. For more information click here.