Last month I talked about optimizing your site architecture for Panda. This is of course important, but it is only one component in optimizing for Panda. As everyone who read Mike Moran’s recent post on this blog knows, Panda rewards quality content. And Penguin punishes low-quality sites. So content quality is obviously very important for search effectiveness, perhaps more important than site architecture.
Saying “create quality content” is like saying “build beautiful websites.” Determining what is high quality or beautiful involves highly subjective and ambiguous judgments. Quantifying quality seems like a fools errand. Well, I’m here to tell you you can define content quality. More importantly, you can measure and track progress, giving you confidence that your content will be of sufficient quality to rank well for Panda. I explain this as briefly as I can in this post.
Defining content quality
Before I became the global search lead for IBM, I was editor in chief of ibm.com. My job was to set content quality standards and help web teams develop content to those standards. I built a new style guide and spent a few years educating teams around the globe on how to write and edit content that complies with the style guide. What follows is the functional definition for content quality from the style guide, called the Seven C’s of quality content.
- Client-Centric: Quality content is not about what you want to write. It’s about what your clients and prospects need to read. If you learn your audience’s needs and pain points, you have a chance at creating high-quality content for them.
- Compelling: How do you engage with your audience in a way that they will want to read your content? Web users typically scan the page and only read if they see words that are obviously relevant to them. If they come from search, they will look for the words they typed into their queries to determine relevance. If you bold these words within punchy prose, you will tend to be more compelling.
- Credible: Web readers are inherently skeptical and sensitive to hyperbole. Everything you write needs have solid evidence behind it. Think of it like writing a Wikipedia article, where you actually need to put references into every statement. Outside of Wikipedia, you don’t need to include the references, but keep them in your back pocket and refer to them to strengthen your position, if necessary.
- Concise: Web users are extremely time challenged. Give them the information they need in the smallest possible time. Notice I said time, not space because it’s really not about text length as much as how easy it is to process the information. A dense sentence can take longer to process than an accessible paragraph. Most importantly, conciseness is about providing all and only the information your audience needs on that page.
- Clear: Clarity is also in the eye of the beholder. And that’s really the most important point. If you think as your audience thinks, you will learn how to express yourself in ways that are clear to them. Use their language and simplify your sentences. If you do, your content will tend to be clear to them.
- Conversational: On the web, nothing turns an audience off faster than writing that sounds mechanical, like you’re trying too hard to increase your keyword density. Only slightly better is writing that is dry or impersonal, like academic writing. Imagine your audience sitting across from you in a café and engaging in a chat. Read your writing out loud before you publish it. If it sounds mechanical or stuffy, it doesn’t pass this simple litmus test.
- Clean: Like it or not, typos and grammatical errors turn off a portion of your audience. It’s the easiest C on the list to control and measure, but it is no less important than the other six.
We at ibm.com used various means to test the results of our efforts to educate the thousands of web writers and editors on the seven C’s. One was called the Web Listening Post, which was a survey that popped up at random on a percentage of users’ browsers when they came to ibm.com. The survey asked users if the site helped them achieve their goals or not. If they said “no,” it allowed them to select the reason why they did not achieve their goals. Content quality was one of the selections.
When we started the program, content quality was the second most-often selected reason users said they did not achieve their goals. Year after year in the program, we saw a significant decrease in the percentage of users who selected this option. When we shut the Web Listening Post down, the content quality option was the least-often selected option. This is strong evidence that the seven C’s really work as a functional definition of content quality on the web.
Measuring Content Quality
Surveys are not the only ways to measure content quality. We also use more traditional metrics to infer content quality from web analytics tools. Here are some handy ones:
- Bounce rates: The number of users who clicked away from a page in a short time. This metric is a relevance indicator. The higher the bounce rate, the lower the relevance. 20% bounces rates are excellent. 60% or above are poor. 80% or more means the content is functionally irrelevant to the audience.
- Engagement rates: If a user does not bounce, she engages with your content. But engagement rates are not just a litmus test. They are a diagnostic tool to help you discover the most and least relevant calls to action on your pages. If items are not getting clicked on pages with low bounce rates, consider changing the content around the links or the link anchor text. In this way, engagement rates can help you improve relevance–a proxy for content quality–over time.
- A/B tests: If you have pages in your environment that you want to update, consider running an A/B test. Edit the pages for the Seven C’s. Then randomly serve the updated pages to half of your users while serving the old pages to the other half. When you get a representative sample of data, check your engagement rates for the two groups. A few years ago, we did this to a cross section of our marketing pages and found 30% improvement in engagement rates on our key calls to action for the edited pages across a diverse set of user intents and target audiences. This helped us get funding for more programmatic work to improve content quality across the enterprise, using the Seven C’s.
The Seven C’s are a good approximation of the criteria Panda uses to assess content quality. In reality, it’s more complicated than a simple set of criteria, however. The quality testers who train the algorithm to look for content quality signals are as diverse as your users. But if you run education and measurement programs to improve content quality using the Seven C’s, you will see a corresponding improvement in your search results in Google.