The explosion of big data over the past decade has given marketers the ability to make data-driven decisions in place of gut instinct and anecdotal evidence. And as marketing strategies go, the benefits of taking a data-driven approach are clear. After all, what business doesn’t want to invest it’s marketing budget by targeting users whom the data shows are valuable prospects primed to convert?
That said, if the data you’re using isn’t quality, you may as well be basing your decisions on guesswork. Junk data is just as useless as no data. If you want to maximize its potential, you must first do the work to understand its quality. The higher the quality, the more valuable the data, and the more confident you can be in using it to influence your strategic approach.
Here are four criteria you can use to assess the quality of your data:
1 – Freshness
Your data is only useful to you if it’s current. When your data becomes out-data-ed (see what I did there?), it may no longer be accurate and, as such, should no longer be used as a basis for decision-making. Obviously, certain demographic data, like gender, won’t change over time; but even data that seems relatively stable, like marital status or home address, will often change.
Behavioral data, especially, can become stale quickly, and this is where freshness is particularly important. For example, a user who shows interest in a specific product should be retargeted within a relatively short period because that’s the time when they’re still considering taking an action. If your data is stale, you might end up targeting users with marketing messages that are no longer relevant to them.
2 – Depth
Depth of data is important because it allows you to have a broader and more precise picture of your audience. Data that represents just a single aspect of a consumer’s profile is only useful so far as it is both representative of, and useful to, your specific purpose. Alternatively, when multiple data points can be connected to construct a more complete picture, more sophisticated and accurate targeting is possible.
For example, imagine you are a clothing retailer searching for ways to personalize your website experience. You have demographic data, so you can customize your homepage based on gender — women’s clothing for female website visitors and men’s clothing for male visitors. That’s personalization based on a single data point and it may be useful to users to some extent.
Now what if you have both demographic and location data tied to the same user? You now have the opportunity to personalize in an even more meaningful way. If it’s the middle of January, for example, you can show men’s winter clothing to a male visitor from New York, and warmer weather options to a female visitor in San Diego.
3 – Source
Not all data is created equal. Certain types of data are more reliable than others, and it’s critical to understand the data source when assessing its quality.
First-party data, or data you collect yourself, is naturally the highest quality. First-party data includes whatever is collected from your website, CRM, mobile app, and call centers. It’s data from your actual customers and website visitors, so it’s strongly representative of the types of consumers who engage, or consider engaging, with your brand’s products or services.
Second-party data is basically somebody else’s first-party data that they’re selling to you. Since you know the original source, second-party data is generally reliable.
Third-party data is aggregated by, you guessed it, a third party, and can be sold widely. Because this data is aggregated from many sources that aren’t always transparent to the buyer, its quality is mostly unknown. Additionally, there may be many buyers of the same data, which may diminish the value of the data.
Quality differs depending on the source, so it’s important to understand where your data comes from when assessing its value.
4 – Methodology
The most reliable data is that which is stated or declared by the user. When a user signs up for a service, for example, they might have to give their gender, birthdate, income level, and other personal details. To the extent that we can assume users are being honest when they answer these questions, this type of declared data is the most trustworthy because we got it from the horse’s mouth, so to speak.
Another type of data is inferred data, which is deduced or assumed based on user behavior. For example, if you visit a website and look at five pairs of winter boots, we can assume you are interested in purchasing winter boots.
Both declared and inferred data can be used to create “lookalike audiences”, which are audiences that are modeled from known audiences. As you can imagine, this modeled data will vary in quality based on the accuracy and size of the original audience it was modeled from.
In and of itself, data is not a magic wand that can be waved to improve all of your campaigns and solve every last one of your marketing woes. Data must be strategically employed and continually assessed for quality in order to have its intended impact. So, before you take your data as gospel, consider how old it is (freshness), how comprehensive it is (depth), where it’s collected from (source), and how it was collected (methodology).