For digital news publishers, artificial intelligence currently occupies an odd space — there's broad agreement that it is a transformative technology, but discerning exactly how it can help right now can be difficult.
What many organizations aren't aware of is that while some of the most ambitions applications of AI are still a ways off, there are some effective artificial intelligence-powered solutions in the marketplace today. As a recent overview of AI in the news industry showed, forward-looking publishers such as Toutiao are already utilizing the technology to accomplish tasks such as improving distribution.
One area in particular where AI has been incorporated in useful and powerful ways is as part of analytics suites.
What does that mean? What is the role of AI in news publisher analytics? How do these tools work? What fresh insights can they provide?
Here's what publishers need to know:
What is the role of AI in Analytics?
The fundamental difference between AI-powered analytics tools and non-AI technologies is that the former are able to not just ingest and present, but also to intelligently analyze, understand, and predict.
As with other platforms, AI analytics start with publishers' data. From there they take the next step: the systems are able to autonomously glean insights, anticipate/predict future actions, and identify solutions.
Crucially, AI analytics platforms are also able to improve their own performance and the quality of their analysis over time. This is the root of AI "machine learning"; the computers, like humans, take in information, learn from it, and develop.
What's so great about AI in Analytics?
One of the major challenges facing content creators when it comes to analytics is scale. Traditional analytics platforms can collate and clarify information, but the work of turning this into useful insights has to be done by humans.
As the amount of content being created has grown, and as the nature of online publishing has become more complex, this has become increasingly untenable; the difficulty and expense of having staff tackle the full-range of analytics needs has become impossible for many organizations.
Artificial intelligence and machine learning-powered analytics solve the problem of scale. These platforms are able to ingest massive amounts information; perform sophisticated analysis quickly; develop prioritized, highly-specific recommendations; and predict a wide-range of future actions. In other words, they expand both the number of, and quality of, actionable insights while reducing costs.
What Types of Insights can AI-powered Analytics provide?
The scope of insights that AI analytics platforms can provide is extensive. Depending on the system and the goals of the specific publisher, it's possible to apply the technology to tackle all sorts of needs.
That said, AI analytics insights tend to fall into these three broad categories:
- Diagnostic: AI analytics can continually scan and examine a publishers' offerings to detect anomalies and flag patterns. For example, they can tackle tasks such a identifying if pieces by specific authors tend to overperform/underperform or alerting teams if particular content offerings are lagging in reach and are in need of extra promotion.
- Predictive: AI analytics platforms aren't limited to current or past behavior; they can also predict future actions. These capabilities include identifying the visitors who are most likely to subscribe or click on certain ads, recommending if topics are the right fit for certain channels, and determining the expected reach of particular pieces of content.
- Applications: AI analytics tools can work hand-in-hand with other technologies to continually accomplish real-time tasks for publishers. For example, they can power content/channel/topic recommendations for visitors, determine paywall lock ratios, and optimize the delivery of elements such as subscription prompts and specific adverting units.
What does an AI-enabled Analytics Solution look like?
In the realm of content analytics, we would expect it to do what's described in the animation below.
We're biased, of course, but we believe our NativeAI solution is a good example.
The platform utilizes natural language processing (NLP), semantic analysis, and machine learning algorithms to catalog each piece of content — including headlines, authors, tags, and media. It then places them within an interconnected ontology that is continuously learning and growing.
Essentially, it is able to understand the context of content: it places each piece within a taxonomy of more than 1 million topics and measures it against content both on-site and across the web. Publishers are thereby given the ability to connect the "what" with the "how"; they can see the value audiences place on a piece of content relative to other content and can optimize for future engagement.
This approach enables publishers to get a range of metrics and applications beyond standard analytics, such as:
- Automatic advanced topic classification and content tagging
- Suggested related topics to form new interest categories
- Connections between content, topics, authors, and pieces of content
- Sophisticated audience behavior and segmentation insights
- Real-time recommendations for increasing engagement
- Optimizations for subscription/paywall triggers and ad targeting
This is just the tip of the iceberg; AI-powered analytics can provide a host of additional specific insights. Perhaps more importantly, they can also help publishers tackle broad, vital questions that previously were beyond the reach of analytics tools, such as: Which audiences are likely to engage with a piece of content? Which visitors are most primed to upgrade to a paid subscription? How can optimizing for locations, channels, demographics, and audience interests improve ROI? Which authors/creators are sparking the most engagement? Which content topics are being underserved?
Again, what's important to note is that these capabilities are not coming in some far off future. Publishers can utilize AI analytics right now to gain these insights, boost subscriptions, enhance ad targeting, and improve their content.