Author: Evan Magliocca

Programmatic advertising: what’s the difference between good and bad data?

Over the last decade, marketing has drastically shifted its foundation from innate intuition to calculated analysis. It shifted from an art form to a science.

Strategies are now driven by unprejudiced analysts instead of committal creatives. More importantly, strategic data insights give each marketer the ability to provide the best experience possible for the customer.

Yet one area of digital consistently lags behind the rest. As disruptive and momentous as the data revolution has been, digital advertising has been slow to embrace it.

Twenty years ago, advertising was the most data-driven industry in the market. It utilized demographic data through publishers to identify target markets. The problem is, many digital advertisers haven’t evolved since then. They’re still relying, in many cases, on demographics.

They’re content.

And as we all know, no brand can settle in the digital age; it’s a constant evolution. Marketers probably tell their kids bedtime cautionary tales about the brands that lost their edge because they stopped moving forward. The list is long.

But there’s light at the end of the tunnel for digital advertising. Programmatic’s ascendance has opened a path forward to evolve from the era of Mad Men to the age of Silicon Valley. Although, data points run a broad spectrum of quality.

Data is quite similar to making sausage: you don’t often know what you’re getting and there’s good quality in some, but there are also some shady practices to dupe the unsuspecting buyer.

You get what you pay for with quality and you need to unpack the data to see what you’re buying.

The majority of data in digital is currently unrefined, broad and hard to action – it’s filler. Marketers need to avoid that data pitfall within the digital sphere.

Bad data

Most of the unrefined data that brands utilize to target consumers digitally is considered third party data. It’s the data that publishers love to offer brands based on their readership.

Third party data, in essence, is demographic. It’s your age, your gender, what you read, where you live, and while demographic data is better than no data, it’s often a false promise that looks glamorous when presented, but lacks in execution and results.

In other words, demographics are filler, they can help plug cracks in your customer view, but they will never give the information necessary to execute a successful campaign.

Let’s consider a quick example of third party’s downfall. Let’s say you and I have some similarities in our demographics and preferences. We are both 54-year-old males, we live in the same city and we both subscribe to a sports publication.

While it may seem like we have a lot in common, what can we truly correlate from those identifiers? We may be the same age, but where the Rolling Stones might be more aligned to my age, I might actually listen to the same music as today’s 15-year-old kids.

We may live in the same city and have similar jobs, but that doesn’t mean we both wear suits and shop at Brooks Brothers.

With third party, there are no actionable insights to draw conclusions on behavior. If marketers want to have a true view of their customer, they need transactional data to be the foundation.

Great data 

There are two forms of data that are reliable and actionable for marketers – first and second party. The important difference between good data and fluff is that with good data, there’s a retail action performed by the customer.

It is based on transactions; the only measurable form of data marketers can develop strategy around.  A transaction is a sign of intent; it provides guidance on what the customer is looking to buy. If we can build up enough of a history with transactional data, the patterns emerge.


First party is data collected through your brand based on transactional history and account preferences.

It’s your data; nobody else can use it unless you’re selling. It’s by far the best form of data to market against, but it is somewhat limited to re-targeting since the consumer has already purchased with your brand.

Within a strategic mindset, first-party data is going to be amazing for current customers – its value as a retargeting tool is unparalleled, but it’s not as useful with acquisition. Focus on up-selling and increasing return frequency. Use it to drive product newness and seasonal initiatives.


Second-party data is where things get really fun for marketers.  It’s the type of data that’s going to support all of the acquisition objectives for the season.

Second-party data is usually going to be transactional, as well. The difference between first and second-party, is that second-party data is collected from another brand.

In essence, we can take first-party data from other companies and market against it as if it were our own. This form of data is also available through an exchange and it avoids some of the filler you get from aggregators that don’t have as much clarity into what you’re buying.

So instead of focusing on whether or not potential customers live in the same city, or subscribe to the same publication, marketers should be focused on finding like-minded brands with a similar customer base.

In the next example, let’s use second-party data. Let’s say we are running acquisition for a high-end chain-store fashion retailer with an AUR between $70 and $90.

It makes much more sense to harness the transactional data of J. Crew, or even Pottery Barn, than it does to target 54-year-old men that live in the same geographic area. There are more similarities between the consumer base and we also know intent since they have purchasing power.

In the end, what marketers want to target is an actionable lifestyle, not passive demographics.

The rise of programmatic is one of the most exciting developments for the digital sphere. It opens a new window for advertisers and it is already shaking the foundations of how publishers operate.

If programmatic is to evolve and have continued success, it needs to follow the path email, CRM, loyalty and site-marketing have taken of utilizing actionable, reliable, first and second-party data to prove out acquisition and investment.

Marketers need to know what’s in their data and trim out the filler to provide continuous, data-driven ROI for their brands.

Finding intelligence to act on from big data: a five step approach

Every marketer has been sitting with his or her analytics team, reviewing an overwhelming spreadsheet of data points. It tends to hurt your eyes and you don’t know where to find the data that can help you.

Meanwhile, an analyst is droning on about a thousand different “insights” that mean little to your marketing channel, or don’t give you enough information to execute an actual strategy.

It’s a frustrating situation because it debilitates a marketer’s ability to drive the business.

We are short on time and we usually have a dozen teams breathing down our necks looking for insights, reporting and strategies. The worst possible scenario, given the circumstances, is if we go too far down the wrong rabbit hole with our analytics—it can be a massive setback, and not just a waste of time.

Most marketers would like to be omniscient; we want every data point, every piece of information. We need to be informed on every possible scenario to make sure our strategies and tactics are sound. But therein lies the issue. Just as a developer would spend months perfecting his or her code instead of releasing; marketers can spend months identifying data paths prior to executing.

Big data can be overwhelming and it’s our version of quicksand that produces inaction and pushes us deeper into the quagmire. Our own behavior and temperament can often be our own unravelling.

So how can we overcome our own mindset? How can we find the path forward and reduce our own gluttony given the over-abundance of data?

We need to put ourselves on a KPI diet.

Here’s a five-step approach that you can use to produce actionable intelligence from a deluge of data.

The goal is to be lean and efficient–to cut the fat wherever possible and avoid getting off track.

Step one: Identify your north star metric

Every brand should have one metric—above all others—that homogenizes each team to drive results.

A North Star KPI is founded on the basis of cross-functional unity and action. It’s one indicator, selected above all others, that is the driving force to meet objectives.

During every creative review, every strategy session, every brainstorm, every hindsight—this metric should be top of mind.

This mandate is executive-driven and can change as often as necessary, but it’s usually a seasonal goal. As an executive, it’s your greatest weapon to drive success quickly, since it harnesses the full momentum of each group to strive for results.

Step two: identify channel-specific KPIs 

While an over-arching metric will help the entire team to drive brand goals, your team will need specific KPI’s to drive as well.

The idea is to tier our priorities across the marketing teams to drive objectives as a well-oiled machine, instead of as a group of individuals competing for their own metrics and results. The individualist mindset often produces a chaff for the consumer and mismatched user experience.

This approach will also reduce the number of directions your team will be focusing on at the same time.

While it’s very tough to narrow scope, avoid having more than three channel-specific KPI’s at a given time. Remember that each metric is going to compound.

Your team is going to be building filters and segments on top of KPI objectives. The resulting impact can become unmanageable very quickly.

Step three: refine your audience & segments  

This is where each KPI is going to compound so marketers need to be very specific and goal-oriented on segments.

Segmentation is where marketers produce the most stagnation. The possibilities are endless, there’s so many ways we can target and we want to execute each and every one. So put the blinders on and cut down to the top group of segments with the most room for actionable results.

There are a million different breakdowns, so focus on three or four. The simpler the better upfront; you can dive deeper once you have a foundation to build on.

Here are some ideas:

  • Gender. It’s the most black-and-white segment and it’s very high up the funnel, so it’s a great foundation to build on. But there’s still a lot we need to solve. What about multi-gender buyersThis is something you’ll definitely need to keep in mind, especially if you are marketing for a kid’s brand.
  • Channel contribution (Social, Email, SMS). Channel identifiers are a great top-of-funnel strategy to drive engagement for site marketing. It will help get consumers to product quicker by providing a contextual experience. If they came through social, try to land them on a page with UGC to lessen the barrier to entry.If it’s through email, get them to the product quickly–they’re already an engaged consumer. Provide product recommendations since you most likely have their account information.
  • Purchase frequency. Transactions can be drilled down in a lot of different forms. To keep your data actionable, try to focus on the extremes.High frequency and near-lapsed—they’re the lowest hanging fruit with the most margin for incremental gains. Marketers can also go higher up the funnel with buyer / non-buyer, or loyalty / non-loyalty.Frequency is a great segment for email subscribers and display retargeting. If your team utilizes a lifecycle approach, there are great benefits to upselling through order confirmations and a cadenced email or digital program.

Step four: evangelize the approach 

Sit down with your analytics team and discuss these metrics and their importance to your channel. They can also help guide your channel-specific KPI’s in the right direction.

If you can help them understand why you’re implementing a narrowed scope, it will help them weed out extraneous information in the future. It means more efficiency for your team.

Step five: test, measure, refine

As any good marketer should be doing in this era, your execution should take place through a testing period.

Identify what’s lacking through your key performance indicators that you selected at the beginning, break it down by your audience types and find the weak spots. From there, test for incremental gains and execute.

Two common themes within this five-step process are unity and simplicity. Marketing teams are far too fragmented today. They’re also overwhelmed with objectives, initiatives and metrics.

Avoid paralyzing your team by streamlining your approach, producing unified objectives across function, and setting your brand up for long-term success.