A Lie Agreed Upon: Taking Analytics Beyond Vanity Metrics

January 31, 2019

Over the holidays an article in NY Mag titled “How Much of the Internet Is Fake? Turns Out, a Lot of It, Actually.” and an accompanying Twitter thread from Aram Zucker-Scharff (the ad director for the Washington Post) received a good amount of attention in the media and advertising space. The article (and resulting thread) detailed how much of the metrics that are online, and those that drive advertising dollars in particular, are fake. Bots drive fake views of ads which appear on fake websites with inflated viewers.

analytics vanity metrics reporting fake data

While there’s a lot to unpack here, some of it comes from two sources. One is the “pivot to video” that many media companies made a few years ago. This led to companies laying off a lot of writers and ultimately finding that video wasn’t driving the actual views/traffic that was promised. Much of this was driven by Facebook, who encouraged this behavior and allegedly over-inflated video view metrics – some estimates by 150-900%.

This is all supported by companies and organizations that have no real reason to actively root out the issue because they benefit from the higher numbers. There’s a whole infrastructure that’s built on this, and very few people benefit to get to the bottom of it. While advertising and digital agencies are going to be the focus here, this will also impact PR – as much of the industry’s reporting is based on impressions.

As leaders of analytics, measurement and reporting within the industry, we need to think critically about this and have strong answers. Here are a few things to consider:

1. Are you reporting inflated impressions numbers? Almost certainly, yes. We are limited to what data is available and as the content above highlights, even the data we would think is most reliable (direct website traffic, video views) can be over-inflated. Further, web traffic data can also be higher than actual impressions based on what we report from sources like comScore.

What to do: Always provide the source of impressions and be consistent on how you report them. Also, be very clear that you’re reporting “potential” impressions based on the data that is available.

Additional Thought: Impressions have always been controversial. Before the internet had bots and fake sites and “video views” there have been questions about what is a real “view”. Nielsen has used a variety of techniques to get viewership with each having its own set of issues – small sample sizes, actual attention paid, dogs watching TV. Magazines and newspapers use circulation, which is a substantial assumption, and don’t even get me started on pass-along rates.

The internet was supposed to give us stronger ways to measure, but those have proven to be just as questionable, if not more so.

2. Focus on quality of placements. As we always note, it’s important to go beyond the straight number of impressions and focus on the quality of coverage. This still means focusing on the quality of individual placements, which you can actively track versus using tools to view total number of potential views.

What to do: Always report on the quality of earned coverage as well as any quantitative data.

3. Sales and audience outcomes are the most important data points. No matter the total impressions (or potential impressions) for a story or campaign or ad, none of it matters if it doesn’t actually have an impact on the company. You should be measuring business impact as often as possible. Program goals should align with business goals such as perception, specific behaviors, sales or stock price.

What to do: While there are ways to drive up impressions and, frankly, report very questionable results, you cannot drive fake sales or fake awareness. By relying on surveys to measure awareness and reputation and actual sales data, you can show the impact of your work without relying on highly suspect metrics.

These are real issues and we, as an industry, need to be honest in our reporting. When the industry enthusiastically shares outrageous impression numbers, we do a disservice to ourselves, our work and our own intellectual honesty. We must continue to push our teams and clients past impressions as a single metric and into actual numbers that can be proven, with quality-based context metrics included as well.We must also actively question any vendor, influencer or company that promises impressions and vet their data as closely as possible. Even then, the focus should remain on business goals, and not on vanity metrics.

Patrick is Vice President of Analytics in Ketchum’s Chicago office. He has been with Ketchum since 2006, where he started at the company headquarters in New York. Patrick has extensive experience in both primary and secondary research. He has specialized in survey research, including corporate reputation measurement, campaign analysis and surveys for publicity. He has assisted clients and account teams with online and traditional media measurement, crisis tracking, goal-setting and program measurement. He has deep interest and expertise in the food space, leads research and analytics for Ketchum’s Food group.