10 Do’s and Don’ts of Data-Driven Content Strategy
- Read original article here
Tech companies often rely on data to make decisions. Like, every decision, from which employees get promoted to which shade of blue a sign-up button should be.
Content, however, stubbornly resists data-driven decision-making, despite the variety of analytics tools that have emerged to help strategists.
SEO software, for instance, helps content teams penetrate the mysteries of the world’s most popular search engine: Google.
Some tools surface search volume data around specific keywords, to help strategists ensure they’re “speaking the same language [their] audience is speaking,” Deborah Carver, a digital strategy consultant and the writer behind the newsletter The Content Technologist, told Built In. Other SEO tools, she said, are “really very specifically about decoding Google’s algorithm and making sure your content gets seen.”
Content analytics platforms and user experience tracking, meanwhile, measure content’s impact. Tools in these areas help strategists quantify how many people read a piece of content, how far into the story they scroll and how they behave after reading.
But over-reliance on data from any of these tools creates problems. There’s a persistent gap between what has worked and what will work.
“You just never know what’s going to be a hit,” Carver said. “It’s so humbling.”
So when a blog post goes viral, it’s understandable but ultimately ineffective to build all future content strategy around mimicking that outlier, runaway hit.
“You can’t just do the same thing over and over again, and expect it to [prompt] the same exact reaction,” Carver said.
Sure, revisiting formats and broad topics can work, but readers, and Google’s algorithm, reward novelty. So, how can content strategists leverage data while still making space for new ideas and the humblingly unpredictable workings of the internet?
Carver gave us some of her top tips for balancing the quantitative and the creative in content strategy.
Browsing actual search engine results pages yourself helps contextualize data on keyword search volume, Carver said.
Say you want to write a blog post about rock-climbing shoes, and the term “rock-climbing shoes for women” gets huge search volume. The situation seems auspicious! But numerical data alone shouldn’t get a blog post greenlit. Especially in this case.
“If someone is searching for ‘[rock-climbing] shoes for women,’ they probably want tobuy… shoes,” Carver said. Not read about them.
That’s the kind of thing an actual Google search can clarify in a way data alone won’t — the intent behind a search. (Google results prioritize content users actually clicked on, and clicks on e-commerce pages, say, reveal an intent to shop.)
It might turn out that the optimal blog post keyword has a lower overall search volume, but indicates stronger intent to actually read about shoes, Carver said.
When you’re building out a content strategy, it’s important to consider not just what your audience wants to read about, but also what level of expertise they bring to the table.
“You need to know your audience,” Carver said. (Or, if you’re just starting out, you need to know the audience you’re seeking.)
Novices and experts might be interested in similar topics, but they’ll search different terms. Rock-climbing experts, for instance, will have ultra-specific, low-volume queries about the subtle differences between different kinds of ropes and carabiners. Novices will be looking for more basic information, like the difference between top-roping and bouldering.
To avoid writing novice-level stories for experts, or vice versa, Carver often breaks a given topic’s popular search terms into expertise categories during her SEO research.
It’s a common practice, which Carver calls “fine,” to A/B test two creative alternatives against each other, to see which one performs better based on one particular metric.
When it comes to email subject lines, though, she has found that the results are always the same: “The email subject lines that are more direct always win.” In other words, they get the most click-through from a mass audience.
This frustrates pun-loving creatives, but clarity makes these subject lines “technically a little bit more accessible,” Carver said. “The thing about some creativity is you don’t always know exactly wha’'s going on.”
“Never look at one data point as a source of truth,” Carver said. Some strategists, for instance, focus on pageviews, but pageviews are just one dimension of content performance. Rich analysis includes other dimensions, too. These might include:
Visibility: Usually measured by reach or impressions, visibility data answers the question, “How many people saw this piece of content?”
Engagement: Engagement data encompasses pageviews, but it also tracks how many people liked, shared or commented on a piece of content in their social feeds, and how readers behaved on the page during their pageview. How long did they stay on the page? How far down did they scroll?
“People always want to get more in depth on [this] part,” Carver said. “Especially with content, when the goal is for someone to consume it — well, they did.”
In other words, readers don’t need to necessarily “convert”— make a purchase, or read a second article — for a session to count as a success.
Loyalty: Loyalty data — say, data on how many users revisit your site, and how often — helps differentiate the “people who casually engage with you” from “that deeper level, people who pay for your subscriptions, or pay for your product and are actively reading your content marketing about it,” Carver said.
Looking at multiple data dimensions at once can illuminate subtleties pageviews can’t, Carver said. One piece of content might have been shared by a celebrity on Twitter and racked up tons of pageviews from first-time readers; another might have gotten fewer pageviews but resonated deeply with a company’s most loyal readers. Those are both success stories, in different ways; it’s more strategically helpful to recognize that than to treat the less-viewed one as “worse.”
The most successful content strategies reliably resonate with loyal, engaged readers, Carver said. So most worthwhile analytics clarify who they are, what channels they rely on and what subjects they care about.
“It’s always more successful when people [say], ‘Your audience is made up of people and these people like to do these things,’ rather than ‘You have a million page views and a 4.5 percent conversion rate,’” Carver said.
Ultimately, not all pageviews are equal. It’s not that valuable to get a piece of content read by people — even a lot of people — who don’t care about your core product and never return to your site.
Another way to get to know your audience: Literally ask them to provide data about themselves through a survey. You might, for instance, ask readers to select the terms they find most interesting from a list of popular search terms, or ask them to rate their level of expertise on a given topic.
That way, “you’re actually triangulating and validating” insights from SEO research and performance analytics, Carver said.
Even the best content strategies take time to pay off — up to two years, Carver said. To make the slog bearable for higher-ups, Carver builds digestible dashboards of key content metrics for her clients, “so they can see how things change month over month” and watch the small successes add up.
A recommendation algorithm can help keep readers on your site, referring them to a new blog post they’ll love once they finish the one they’re reading. To build one, though, you need metadata.
At Netflix — renowned for its algorithm — “metadata is very organized,” Carver said. In fact the company, as of 2018, has paid people to watch Netflix full-time and tag video content with relevant metadata that covers everything from subject matter to goriness and “plot conclusiveness,” according to The Atlantic.
Not every recommendation algorithm requires such granular tagging, Carver said, but it does require sometagging, done by either humans or AI. Investing in this tagging can be a great move — leveraging metadata via algorithm can lead to longer sessions, and increased loyalty.
Ultimately, SEO research and performance analytics should inform content strategy, but all that data should mostly serve as a starting point. Human judgment is key.
If you’re assigning a story tied to a popular search term, for instance, an editor or a writer should agree that it’s story-worthy and of interest to your audience, Carver said.
And no search analysis, or even natural language generation, can replace human creativity. If you want good content, “you just have to hire a good writer,” Carver said.
More on MarketingNoise Is Other Marketers
