Local news publishers are often able to pull together a decent share of pageviews in their communities — but many can’t wring enough revenue out of them. For practical solutions to this quandary, I went to two experts at Ezoic, a Google Certified Publishing Partner whose artificial-intelligence platform aims at helping its thousands of publisher clients to deliver better user experiences, resulting in more engagement and leading to higher ad revenue.
As the company’s Head of Marketing, Tyler Bishop has helped dozens of sites increase their organic traffic — and former Googler Ohad Tzur, who is VP of Global Partnerships, specializes in bringing machine-learning technology to publishers and technology providers. Here’s my Q&A with Bishop and Tzur:
What should news publishers be doing — especially at the local level — that they too often ignore or overlook?
Publishers could be doing a lot more to enhance their businesses by looking closer at user-tailored experiences. The reality today is that the vast majority of publishers are taking a “storefront” approach, providing the same experiences to different users with different intents.
Many publishers — of all shapes and sizes — are still measuring their progress by looking at distorted metrics like cost-per-thousand ad impressions (CPMs) and ad revenue per thousand impressions (RPMs). These advertiser metrics fail to provide a true north to publishers who are trying to measure the success of their efforts to improve revenue.
Explain the shortcomings of CPMs and RPMs.
What’s misleading about RPM and CPM rates is that they could be going up but overall ad revenue could actually be going down. The math behind this is something we talk about a lot and is actually pretty simple (see double chart below).
The reason is that CPMs and RPMs do not account for critical user-experience metrics like pageviews per visit, “bounce rate”[percentage of times inbound visitors view only the linked page] and session duration. You can read here about how CPMs and other advertiser metrics can make this overly complex.
You stress “revenue per visit.” How do you calculate that?
The best metric is “earnings per thousand visitors” (EPMV). This is simple to calculate. You measure your total earnings over a period of time and then divide that by per thousand website visitors.
This calculation will give you a true north of whether you are making progress in your publishing business. Unfortunately, many publishers still do not fully understand how or why user experiences play such a big role in the optimization of a web property.
This is why providing individualized user experiences is such a valuable opportunity for publishers. A user who comes to a news site from Facebook on mobile may interact differently compared to a user who comes from search on tablet.
At Ezoic we’ve seen noticeable differences in results when tailoring user experiences to visitor profiles based around testing powered by machine learning [a subset of artificial intelligence designed to learn from user data or behavior without being explicitly programmed to do so]. Any publisher can benefit from these learnings, no matter how big or small, but it requires two fundamental realizations:
- Adapt to the concept of automated testing as a driver for decision making (avoiding subjective/emotional judgment (e.g., “I’m placing an ad unit here because I think it looks great”).
- Measure a user’s value on a per-session basis – EPMV instead of CPM.
What is significant about the chart?
It shows changes in “bounce rate.” The “bounce” is the process of a user session originating and then ending on a single page. An increase in pageviews per session is likely to result in more time on site and more value (earnings) for the publisher. These are the types of data-driven learnings that we find many publishers failing to implement into their practices.
Overall, we see that improving user experiences by extending the session leads to more ad impressions and better overall session revenue (EPMV). This is why it is critical for publishers to know which landing pages produce their overall EPMV. In many cases, they may have a very high earning page, but it may also have a high bounce rate. A page with lower earnings but a much lower bounce rate may actually produce more revenue because it may lead to more pageviews and overall greater session earnings.
Ultimately, user experiences and earnings are inexorably linked. Almost every publisher can improve his/her existing business by thinking more critically about how to provide the best possible journey to each individual use, based on real data.
When you talk about revenue here, you’re basically talking about programmatic ads. Can programmatic work for smaller as well as large publishers?
Programmatic can be the driver forward for smaller local publishers as well as larger ones. Many big-name publishers are starting to focus on smaller audiences rather than big ones. Publishers will never have the reach that Facebook or Google has, so trying to compete with them for advertising dollars is difficult.
Concentrating on small, quality audience and driving more targeting strategies at them do leave the door open for direct deals, but not nearly on the scale as it once was. The days of getting big paydays from homepage takeovers is going away. That begs the question, then, what do we do with our broader, less-frequent audience (the bulk of traffic). Better programmatic is the answer, and that starts with better data on user demographics, attributes and behavior on the “consumer journey.”
To enhance user experiences and extend the length of sessions, would it help if news publishers started deploying chat apps – for example, getting reactions from the community as the Trump White House issues executive orders on immigration, Obamacare and other hot-button issues that would directly impact millions of people?
Do testing first. Chats may be a great idea for some people but terrible for others. How do they affect user experiences? Do they have a positive correlation with EPMV, bounce rate, session duration, etc.? If so, they are likely good. However, they will inevitably be liked/disliked by some more than others. This is why individualized experiences are so important. It’s also important that local publishers don’t rely on tests that others perform. All audiences are different. Do you own testing.
What, specifically, should news publishers, especially at the local-community level, be doing to engage social media inbounds more deeply and extend sessions?
The answers depend on publishers’ core capabilities and resources and what benefits would they see short and long term? What uplift do they expect on key performance indicators? What degree of granularity would they take with their data?
Small to medium-size publishers are not likely not to have the technical savvy and analytical capabilities to close this gap on their own. Their competitive advantage is their content, and they should continue to home in on that to increase user engagement.
Publishers can use tracking and analytics tools to understand their basic user engagement (e.g., traffic sources, user path). A/B testing can yield some good results.
Paradigm shifts for smaller publishers will require technology providers to step in and close gaps on infrastructure, UX improvement and monetization. Are publishers up to date on the concepts of server-side header bidding [where they offer inventory to multiple ad exchanges simultaneously before making calls to their ad servers]? Do they know the granularities of programmatic?
To conclude, as we enter the future of online publishing in which content, experiences and monetization are tailored to individuals, publishers should automate testing procedures to enrich decision making and embrace the concept of user journey (session) value, leaving behind legacy concepts such as CPM.
Tom Grubisich (@TomGrubisich) writes “The New News” column for Street Fight. He is editorial director of hyperlocal news network Local America, and is also working on a book about the history, present, and future of Charleston, S.C.