The 8 Questions Buyers Should Ask Every Data Provider

March 13, 2019

By Matt Frattaroli

sesame/Getty Images

Smart buyers know digital audience data can make a big impact on the performance of a campaign. When data is clean, fresh, and relevant, it can give you a measurable lift in KPIs. But when it's mediocre, it'll burn through your ad budget — or worse, your client relationship.

Let's be honest. All data can have degrees of inaccuracy. (In fact, some data is outdated the very second it's created.) While there are tools available for evaluating demographic data, quality and validation go far beyond basic age and gender targets.

The good news is there is a lot of great data out there, and recognizing it can be as easy as asking the data provider a few simple questions.

Try asking these questions to any data provider you're trusting a portion of your (or your client's) campaign spend. In just a few minutes you'll have a pretty good read on the quality of a data provider's audiences.

 

1. "What's your primary business model? How is data a core component?"

What you're really asking: How much do you really care about data quality?

When the data a provider sells is critical to their business operations, its hygiene will be as important to them as it is to you. Probably even more. But if they buy data just to turn it for profit, scale will probably take priority over quality. The less important the data is in driving revenue, the more lax they're likely to be in terms of their hygiene standards.

 

2. "Is your data properly permissioned? Which laws, regulations and industry guidelines do you comply with?"

What you're really asking: Are you legit?

The only acceptable answer here is "yes". All prospect data used for digital targeting must be properly permissioned. Data that is not properly permissioned is a deal-breaker. Using it could tarnish your business' reputation, or worse, turn into a legal concern.

In the United States, compliance with the CAN-SPAM Act and individual state regulations are required at a minimum — and SOC-2 compliance is preferred. If you advertise internationally, CASL is law in Canada, as is GDPR in the EU and EEA.

Providers should also adhere to the DMA's Data Guidelines for Ethical Business Practices.

 

3. "Which data-driven organizations are you active members of?"

What you're really asking: How involved are you in the industry?

Every data provider should be current members of the Data & Marketing Association, a Division of the Association of National Advertisers, and the Interactive Advertising Bureau. Data providers that serve on the Board of Directors of either organization get brownie points.

 

4. "Was the data sourced from online or offline channels? If offline, what type(s) of matchkeys do you use to verify consumer identify?"

What you're really asking: Can the raw data be trusted?

There's no right or wrong answer here — it's just a matter of what you are looking to achieve. The greatest benefit of online sourced data is recency. If you need the ability to respond to something the second it happens, digital is your only option. However, real-time often means no-time for the checks and balances that may otherwise strengthen the quality and predictive nature of the data.

On the other hand, offline data is less recent (and less common in the programmatic space) but can be very powerful and well worth the wait. The onboarding process alone (mapping offline PII to an anonymous IDs), is inherently quality controlled at the individual level.

Let's unpack that a little further. Every consumer a data provider wants to onboard must have corresponding PII data (also known as a matchkey) in order to be successful. The more PII data known about a consumer, the higher the match rate is likely to be. So, when you think about it, offline data providers are in a way incentivized to verify their consumer records with as many email addresses, in addition to postal address, as possible.

 

5. "Are your audiences sourced from 'known' data signals?"

What you're really asking: Did you model your data?

In a perfect world, everyone would have access to pure "known" data straight from the source. But in reality, that data is often unavailable because it's either highly valuable and coveted by the owner or it's protected by data regulations.

Whether a model was used won't reveal anything about the quality of the audience. Good nor bad. You'll need to know why it was modeled and specifics about the source and strength of the signal data to understand more. Which leads us to our next question.

 

6. "If you modeled your data, what was the objective?"

What you're really asking: What's your angle? Whose interests do you really have in mind?

To be clear, modeling is the process of using algorithms, and multiple data sets, to make predictions. In the digital space, this typically refers to how likely a consumer is to [fill in the blank]. Providers group consumers with the highest propensity (or likelihood) into similar audience segments.

There are many different model types, each with a different purpose but more often than not, a data provider will turn to modeling for one of the following reasons:

  • To improve the quality of the audience. e.g. The data provider has a large universe of prospects and wants to narrow the pool down to only the most qualified.
  • To build the audience. e.g. The provider doesn't have the rights or access to the data, so they use alternate sources to predict who is a best match.
  • To increase the size of the audience. e.g. The provider has an audience but it's not large enough for scale, so they use look-alike modeling to try and find more consumers that look like the ideal prospect.

 

7. "At which level is the offline data collected? What about onboarded?"

What you're really asking: How strong is your signal?

Don't assume the collection level of the offline data is the same as the level it's onboarded. Often times it's quite different. Take credit card companies for example. They have massive databases of billions of consumer transactions on the individual level — but legally, they can't make this raw data available in the ecosystem for targeting. To work around this, some providers for example, will choose to onboard their individual level data at a broader level for targeting, such as zip+4. This protects the individual consumers' privacy while still using their purchase as a signal for the neighborhood.

 

8. "How do you verify the quality of your data?"

What you're really asking: Can I trust you?

In a way, this is a trick question. As mentioned earlier, there are services like Nielsen Digital Advertising Ratings (DAR) that measure basic demographic data against industry benchmarks but they are in no way comprehensive analyses of data quality because there are hundreds, if not thousands, of additional data points to take into consideration.

A couple companies have emerged recently that evaluate audiences by conducting surveys and comparing the consumers' self-reported information to the data a provider has about that person. Alliant was invited by Liveramp and Lucid to beta test their new methodologies and will have more to come on that soon.

If nothing else, you can always request an independent evaluation, which would give your in-house statisticians the opportunity to analyze the provider's data first-hand. Case studies can also be helpful but keep in mind they are inherently biased and most likely won't tell you the full story.

In all reality, if you're a data geek like us, you could ask questions for days. But even if you're not (or just don't want to admit it), these talking points will provide the answers you need to make an informed decision. And don't worry about feeling like an interrogator. If the provider takes quality seriously, they'll be more than happy to tell you all about their pain-staking processes for keeping their data "so fresh 'n so clean".

Here's a bonus tip: Build a network of a few data providers that offer a help desk, and lean on them for support with all of your campaign needs. Whether you need an audience recommendation or some flare for your presentations — when you need help, we've got you covered.

 

Matt Frattaroli is the VP of digital platform and agency partnerships at Alliant.


The views and opinions expressed in Marketing Maestros are solely those of the contributor and do not necessarily reflect the official position of the ANA or imply endorsement from the ANA.


You must be logged in to submit a comment.