“Data-Driven Thinking" is written by members of the media community and contains fresh ideas on the digital revolution in media.
Today’s column is written by Pete Kim, CEO of MightyHive.
The first- party data/big data revolution is in full swing.
Virtually everyone now agrees that activating first-party data is strategically important. The simple truth is that it’s difficult to make third-party data profitable over the long run — eventually your competitors will use the same tactics with the same data. More than ever, first-party data represents a critical foundation for future marketing plans because it’s the only data that your competition will never have. And congratulations, you are now the person who has been put in charge of this effort — tasked with devising a plan to achieve data and marketing glory!
And then reality sets in. Moving past the hype, how can you create a sane and sensible path forward? Here are five thoughts to consider.
1. First-Party Data Is More Than Just Retargeting And CRM
To date, most discussions surrounding first-party data have revolved around retargeting data and customer relationship management (CRM) data. This may be a fine place to start (see the sections below), but it’s important to appreciate the big picture. First-party data, better known as enterprise data to the rest of the technology industry, is a broad term that includes hundreds of potential data sets, each of which is an enterprise software niche unto itself. Examples include customer data, purchase data, product and inventory data and many more.
To make matters worse, every industry vertical seems to call these by different names: Customer data might be captured in CRM databases, call center systems, email lists and direct mail lists, purchase histories might be called POS data, and ecommerce transactions and product data can be captured in inventory-management systems, supply-chain management or manufacturing resource planning systems. And then there are thousands of industry specific data sets, such as vehicle telemetry and service records in the automotive vertical, and equity transactions in the world of consumer stock brokerage. The bottom line: In every company, there are a multitude of options beneath the deceptively simple name “first-party data,” and it’s more than retargeting and CRM.
2. Don’t Boil The Ocean
Now that you’re aware of the sweeping vista of enterprise data, it’s time to step back, take a deep breath … and start small. Ignore the grandiose visions of comprehensive platforms that are the standard building blocks of “marketecture” hopes and dreams.
Pick one data set, generate a hypothesis and test it. Experiment, analyze and learn, then do it again. Find what works for you by systematically making progress one step at a time.
3. Prioritize The Low-Hanging Fruit
The discussion above naturally leads to the question: Where should we start? Prioritization is the obvious path here. A simple example is provided below.
Rate all available data sets against two criteria:
- Expected ROI (eROI)
- Accessibility of data
[Two side notes: (1) Notice that data volume is not something that’s on the list. Don’t forget that many vendors get paid on the volume of media they sell or data they match. They will naturally steer you towards experiments that target large datasets and thus large audiences. (2) Regarding accessibility, it’s interesting to note in passing that we’re in the middle of a replacement cycle in enterprise software. Much of the expensive packaged ERP systems installed over the past few decades are now being replaced by updated cloud-based software. As this process continues, it will naturally lead to data sets that are easier and easier to access.]
Expected ROI is self-explanatory — how valuable do you think this data will be? The rise of retargeting as a marketing tactic has taught our industry the value of intent. As such, many of our clients often substitute “strength of intent” as a proxy for eROI.
Accessibility of data is a measure of how much effort is needed to obtain the data in question. Factors that can make data difficult to access include, but are not limited to: system age and architecture, availability of IT staff, privacy and security considerations and, of course, the omnipresent specter of corporate politics.
Unsurprisingly, website visits score high in both categories — a reassuring sanity check that confirms this model would have recommended website retargeting if this analysis had been done 10 years ago. I have also found that inbound call center logs (phone calls indicate strong purchase intent --> call center remarketing) and purchase histories (for cross- and upselling) have provided strong results in the past.
Bottom line: Take an analytical approach when determining the sequence in which you will activate your first-party data sets. Determine the two or three factors that define “low-hanging fruit” in your world and systematically apply these criteria to select the best starting points.
And here’s a final test: You should be able to describe your use case in a simple sentence. For example: “We will retarget inbound consumer phone calls with display advertisements to drive additional sales.” If you’re unable to articulate your test in a concise phrase, there’s more work to do.
4. Onboarding Is Just The Start
Now the fun begins. It’s time to move past the simple elegance of theory into the wild and woolly kingdom called execution.
First, let’s make clear that, as always, execution is a complicated process. Onboarding does play a role, but it’s a mistake to believe that this step represents a complete recipe for success. Remember that there are several data onboarding options available. After studying them, we have found that:
- The best onboarding vendor is often determined by what type of data you’re looking to match.
- Sometimes it’s a good idea to use multiple vendors, rather than just one.
Other decisions that need to be made:
- Data cleansing, de-duplication, household matching, and general processing strategies
- Media type(s)
- Budget: test duration and frequency
- Test and control methodology (pay attention to statistical significance)
- Data onboarding solution (direct to publisher, partner-based, etc)
- (Dynamic?) Creative
- Attribution parameters
Find someone to help you think through these issues. Because first-party data usage is so new, industry experience can be hard to find, but it’s out there.
5. Two Birds With One Stone: Offline Attribution And First-Party Data
One of the underrated benefits of first-party data targeting is the ability to calculate the impact of your online advertisements on offline sales in a statistically relevant manner. Long considered one of the holy grails of online advertising, offline attribution may be a surprising beast to encounter in the kingdom called first-party data. What’s going on here?
First-party data is typically native to the offline world. Because of this heritage, when first-party data is involved, offline attribution often becomes easier.
As an example, let’s take a look at call center remarketing, which targets advertisements across programmatic media within 24 hours of an unconverted inbound phone call. Using caller ID records as a source of first-party data yields phone numbers that can be used to find the same consumer online. Importantly, that same phone number has a very high likelihood of being captured during a subsequent purchase, whether it takes place online, over the phone or in-store, and can be used as the key to associate an offline purchase with an online advertisement. With the correct test and control setup, this insight not only leads to accurate offline attribution, but goes one step further: unified attribution of incremental sales lift across every purchase touchpoint.
Stated another way: If you know who you’re trying to advertise to, then unified attribution becomes a simple process of monitoring whether that person has purchased or not. For those of you who, like me, have been searching for a meaningful optimization signal other than silly proxies like CTR, this alone might be worth the price of admission. At last, with first-party data executions, the advertiser can finally hear the cash register ring.
Where Do We Go From Here?
The combination of first-party data and programmatic media will yield critical advancements in the next few years, but it will take time.
As we continue to experiment, it’s useful to remember the words of Thomas Edison: “Opportunity is missed by most people because it is dressed in overalls and looks like work.”
Make no mistake, the path forward will take years of honest, roll-up-your-sleeves effort. But for those who possess both the determination and insight to succeed, the rewards will be great.