There’s a myth in the marketplace of how “third-party data is dead,” especially given the significant changes in policies, privacy and compliance requirements. Some brands think third-party data leads to cost inefficiencies, but typically, these brands are leveraging data from an unknown, non-validated source resulting in poor data quality. Among the other different types of data, first-party data and second-party data, third-party is unique.
From our perspective, third-party data is thriving. In fact, according to the IAB and Winterberry Group, spending on third-party data increased by 17.5% in 2018 to reach $19.2 billion and preliminary research indicates the spending significantly increased in 2019.
What marketers need to focus on is cleaning-up their third-party data sources, leveraging it effectively and understanding how it compliments their first-party data. By doing so, marketers can get to know their customers, improve engagement and campaign response and increase the effectiveness of targeting to attract new customers, with better ROI on campaigns.
There has been much discussion about this data and its ineffectiveness and lack of quality. There is definitely a range of providers in the market, with varying degrees of quality and effectiveness, especially in the digital ecosystem.
Read Epsilon's POV on Google's March 2021 alternative identifier announcement
For years, there was a focus on reach in programmatic, which impacted quality. There was also limited scrutiny on data providers in terms of sourcing, the accuracy of data as well as the transparency into what was really being presented.
Thankfully, as marketers have become more sophisticated and with a shifting consumer and privacy landscape, this is changing. My hope is that this will clean up the industry, remove the bad players and shine a light on quality providers.
Third-party data should not be thought of in one sweeping statement—as in anything, there are valid and reliable sources that add to the overall picture of a consumer that first-party data alone cannot deliver. A broader profile of consumers, including their interests, demographics, household composition and purchases, are very useful to help understand each person and improve personalization.
I’ve seen recent articles touting that first-party data can deliver it all, and any other sources are inferior and inaccurate. I have yet to come across a 100% accurate picture of a consumer from a single data source and have even seen inaccuracies from first-party data as they represent only a particular view of a consumer or are based on behavioral inferences from browsing behavior.
For example, Facebook thinks I have a toddler because I liked a friend’s picture, yet my kids are in their late teens. Additionally, they’re unaware that my brother has a dog since he purchases dog food at his vet, not a grocery or mass merchandise store. There are often inaccuracies, and marketers need to be aware of the degree of the flaws and how they impact both performance and the consumer experience with your brand.
So what’s a marketer to do? Educate yourself and ask lots of questions to ensure you know what you are getting, so you can be realistic about how it fits in with your plans and how different data sources can complement each other.
Consider these five tips:
1. Think about what data you need and will use
Data is important to drive targeting to eliminate waste and also make the content more personal to each consumer. Distinguish between these two as they serve different purposes and may result in selecting different data sets. Targeting is about reaching the right person based on where they are in their consideration set along with a brand’s objective and awareness versus conversion.
This is a balance of scale versus accuracy. Anything too big can be meaningless, but anything too targeted can limit the impact and reduce efficiency.
Data for personalization makes the interaction with a consumer more meaningful and impactful—consistently providing what consumers say they want and expect. But a brand needs to be willing to vary both the creative and messaging for this to work to ensure this can be executed. It can be as simple as showing a dog image versus a cat image for pet owners (which I have seen show lift) or a family image versus a single male, and yet so few brands do it well or even at all.
2. Ask questions of every data provider to fully understand what is being provided
Take the time to review all data providers you’re working with and make sure to ask the right questions when evaluating each partner. And remember, no data set is perfect. Make sure to know the pros and cons and understand each information source and allow that to inform how it’s used.
For example, transactional data is precise in terms of knowing what the consumer bought with a retailer, but understanding how much these transactions represent of the consumer’s total spend and how the data is managed is important. During your assessment, make sure to review:
- What is the source of the data? Where was it collected? How much data is collected? Is it combined with other data?
- How the data is linked to consumers? Do they use a cookie, IP address or device? And is it at the individual or household level?
- Know the data derivation. Is it declared, derived, inferred, modeled or observed?
- How often is the data refreshed?
- For transactional data from retailers, it’s important to understand:
- How are transactions captured and linked to an individual or household?
- How many transactions can the retailer capture of total spend in their stores or a given consumer? Is it based on tender (plastic versus cash) or a loyalty card?
- How many cardholders spend the majority of their possible spend with that retailer?
- How much does that retailer represent of your brand’s sales? If it is grocery, how much else is purchased in mass, club, specialty, convenience or club?
- Are multiple loyalty cards per person or household co-mingled or distinct?
3. Think quality over quantity
Did you know that poor data costs the US economy over $3 trillion each year and can cost businesses at least 30% of revenues? It’s true. Oftentimes marketers do not have the resources (or the time) to do a full data quality assessment to determine if their data is of quality.
In our data quality e-book, we outline 10 key assessment criteria to consider when evaluating data quality. Remember, not all third-party data is created equally. Advertisers and agencies must be careful to validate the accuracy and validity of the partners they work with versus assuming all are either ‘bad or good’.
4. Make sure data is transparent
Understanding how data audiences are built is essential and making sure everything stated in the document/label is true is important for data success. With the recent initiatives like the IAB Data Label, the ecosystem is working to create transparency to eliminate poor data providers and hold all data providers accountable. Check that your data provider is certified or in the process of certification to ensure what you think you are buying is actually what you are getting.
5. Ensure data is compliant
It’s important for data providers to share how they comply with current legislation in addition to how they’re preparing for new legislation. Make sure to continuously review privacy policies and opt-out language. At Epsilon, we protect and respect data, and we’re always thinking privacy. In our role as a trusted partner to clients, Epsilon remains committed to having a leadership role in shaping the changing privacy landscape.
Data is powerful. It’s the brains behind our marketing success. It’s critical to understanding consumers, reaching new buyers and to better communicate with existing customers. There is no silver bullet for the perfect data set, even first-party data.
This data is valuable when it’s of high quality, and multiple partners will always be necessary for full coverage of consumers and useful insights about them that are not skewed. Take the time to understand every partner you work with, and only work with those that are honest about their strengths, weaknesses and gaps.
To learn more, download our data quality e-book.