This article originally appeared in Campaigns and Elections Magazine

sk a dozen consultants to define big data and you’ll likely end up with a medley of responses. That’s because big data is a nebulous concept open to interpretation, which sometimes leads to misconceptions.

Most would agree that it centers around the terabytes and terabytes of data generated and stored every second. Everything from Nest thermostat programs to CVS purchases to Facebook posts are telling some analyst somewhere more about who you are and what you care about.

But does big data give campaigns the chance to learn more about voters and how they should be messaged? It can — sometimes. The rub with data is that quality can sometimes matters more than quantity. Determining the quality of data starts with one simple question: What’s its source? Broadly, there are three ways to categorize the source of data: first party, second party and third party.

First-Party Data is King

First-party data is your data. This includes everything from voter IDs to online petitions to sign-up sheets at district events. It’s information that voters have agreed to share with you. Need evidence that first-party has incredible value? Look at Facebook. The social networking site has become a $255-billion company through the value its first-party data creates for marketers.

First-party data can also be generated from tracking engagement through open and click-thru rates on opt-in lists, participation in tele-town halls, or determining through cookies which pages of your web site users are visiting.

First-party data is richer and of higher quality than any other data because it represents how voters are interacting directly with your campaign. First-party data is the only type of data that represents a true one-to-one experience.

It’s not a big leap to say that data acquired right from the source will provide the most reliable and accurate depiction of a voter’s preferences. Unfortunately, many campaigns don’t see it this way. The most common mistake we see made by campaigns is the failure to fully realize the value and power of their own data.

Big data is seductive. Some campaigns see the incredible granularity of third-party data on the same level as first party data. It’s not even close. Third-party data absolutely has value (more on this later), but if you’re relying on it to make vital strategic decisions, you’re taking an enormous risk.

First-Party Data Isn’t Easy

If it were easy everyone would do it. There are two major challenges with first-party data.

  1. It doesn’t scale easily. Building a rich dataset of first-party data takes patience, coordination and careful strategy. Many campaigns lack the time and resources needed to build enough meaningful first-party data to where they can rely on it solely.
  2. It’s difficult to organize. First-party data is disparate and fragmented. To get the most out of first-party data, a central data management platform is needed to integrate and standardize datasets.

Second-Party Data: We’re All Data Companies

From the marketers’ perspective, Facebook’s data is not first-party. It’s actually second-party. Second-party data is somebody else’s trusted, first-party data that has been acquired or exchanged through a mutual agreement.

The most obvious example of second-party data in the campaign world is a voter file acquired through a state or local government. Other examples are contact lists from like-minded non-profits, advocacy groups and preceding campaigns. Most states also make lists of licensed professionals and gun owners available, although there are often restrictions on how this data may be used.

Sometimes these lists can be almost as valuable as first-party data. That’s because it’s also generated through a one-to-one experience. Considering it’s derived directly from the source, second-party data is considered to be more reliable than third-party data.

But what makes it less valuable than first-party data is that it’s once removed. Since the goals of two organizations are never perfectly aligned, there’s no guarantee that contacts from a second-party list will interact with your campaign the same way.

Let’s say you obtained a list of confirmed supporters from the campaign of a retiring incumbent in your party. It’s a safe bet that much of the list will likely support your candidate, but there will be a few who won’t.

Second-party data isn’t perfect, but after first-party data it’s the next best thing.

Third-Party Data: Trust But Verify

Things like consumer purchases, income levels, religious preference and gun ownership can provide insights into how likely a voter is to support your campaign. This is the kind of third-party data available to campaigns. One advantage with this set is that it easily scales. Need a list of evangelicals in Des Moines today? Done.

Now the cold water. Third-party data can be aggregated from dozens of services through a series of appends and scrubs. With every append, the data gets diluted. Sometimes, the truth will be lost in aggregation.

Third-Party data is the seductive offering of big data. It’s full of promise and often delivers, but an over reliance on third-party data at the expense of first-party data is a mistake.

Going From Third to First

In the campaign world, email appends to the voter file is prototypical third-party data. The lists are typically sourced from people who have opted-in to some list, somewhere, which was likely not related to politics at all. It’s a familiar tale: The list is purchased by an unrelated campaign and blasted with impunity. Most of these people have never heard of you. If they have, they never said they wanted to hear from you. Spam complaints ensue. The campaign is kicked off their platform. The end.

Email appends aren’t necessarily a bad idea if used the right way. Typically, we advise clients to leverage purchased emails in a social listening tool like Matching voter-appended emails to social media profiles enables them to better understand the issues they care about. This gives the campaign an opportunity to convert the purchased email to a true opt-in (first-party data).

For example, you can create a custom Facebook audience, targeting ads to all voters discussing a specific issue on Facebook with a relevant call-to-action to capture data. If your social-listening tool is integrated with a voter file DMP (Data Management Platform), the data can be pulled in, enabling you to tag the voter with the issue, and trigger follow-up contacts. The initial contact served as a soft introduction and decreases the chances of a spam complaint while increasing the likelihood of converting this third-party data to first-party data.

While second and third-party data should be key components of any data-driven campaign, they should never be viewed as a substitute for first-party data.

Justin Gargiulo is the founder and CEO and VoterTrove, a data management and voter contact platform offered exclusively to center-right campaigns and causes