Counting Content: Using Numbers to Turn Research Into Content Strategy

I had a miserable time showing my work in math class. Usually, I used “common sense,” context, process of elimination, or guessed. I preferred English. If there was no real way to quantify an argument, how could you be wrong?

So I latched onto information architecture (IA), a soft science, rather than, say, architecture, which lives on mathematical precision that literally makes or breaks a project. But the more I work with data, the more we measure, and the more we test, I’ve suspected that maybe information can be quantified, too.

Is content data? Not really. A few years ago, I would have eye-rolled at calculating content in order to find the “right” answer. But with IA-heavy projects such as redesigns for thousand-page websites, the scale of information begs for process. I decided to scope out an approach where we pretend content is data. Information can be corralled into topics, and I started counting how often each topic showed up in auditing and research. If certain topics leap out of the data, that’s where your content strategy should start.

Who should count content?

The basic answer would be content strategists. The real answer is anyone wrangling incredible amounts of web-based information. For the purposes of this article, it’s UX designers who already have a solid grip on user research, decent exposure to IA, and intermediate spreadsheet skills.

As an exercise, what follows is a sample set of rationale and steps for a standard .edu website redesign.

Why count content?

Actually, we’re counting “concepts” — hypothetical content which we can infer from research and auditing. Each concept is a common theme, topic, object, or subject that succinctly represents user-facing information.

Let’s say surveys indicate college students want to “decide what to major in.” Relevant concepts on a site you’re planning might include “courses” and “faculty.” To count the global relative importance of faculty-related content versus course content, you’ll want to pin up all the research you have, then count how often “courses” and “faculty” are suggested. From this example, you might resolve to feature faculty profiles higher on the “Academics” landing page, or justify pulling detailed course descriptions into the CMS.

The goal is to inform content strategy through data, rather than pre-existing assumptions, opinions, and beliefs about content.


1. Read your content

  • Pore over the research.
  • Skim the current site to get a sense of what’s there — the overall structure.
  • Read the current site. Every single word.
  • Read the research again.
  • Skim the current site again.

2. Identify your concepts

  • Turn it over in your mind a few times. Visualize what the core concepts appear to be.
  • Start to mind map. Don’t get so low-level that your list is unmanageable.
  • Avoid being redundant. Beware of synonym bias. If you count “programs” and “academics” as separate items, are you fudging the data by doubling the presence of one key concept?

3. Log concepts per research type

Note: I use Airtable. It lets you dynamically link columns to other tables using the “Link to Another Record” field type. As I add concepts I identify to each research table, they’re automatically logged in my master “Content Concepts” table. (Where the math happens.)
  • Grab your spreadsheet software.
  • Create one sheet for each type of research and another sheet for “All Concepts.”
  • Input research findings phrased as use cases, one per row.
  • Pair each use case with related content concepts.
  • Make sure you don’t miss or duplicate anything, or break consistency from the labeling conventions you’ve used throughout all sheets. You might realize a better approach halfway through the process. You might have to backtrack and relabel pages to match what you just realized.

4. Link your sheets

  • Row by row, I’m using formulas to tally up the instances of each concept in each research type, totaled on the “All Concepts” sheet. But wait! To calculate the grand total…
  • Maybe this is dangerously subjective, opening the door to human intervention in an already abstract system. But not all research has equal value. Think honestly about the quality, integrity, and importance of each source. You’re free to tweak the recipe. If I felt brazen with confidence in all our research, I might weigh it all something like this:

5. Weigh your research

  • Business goals — Weight: 5. Even now that user-centered design reigns, you still have KPIs. Stakeholders usually know what sells. Their concerns need to feel threaded through your data, or they won’t buy-in later. The higher weight is the only fighting chance for brand new or neglected content the client now wants to showcase.
  • Interview responses — Weight: 1. The goal of user interviews is to find out what you don’t know. You aren’t conducting a study. You’re exploring possibilities. Although you posed consistent, open-ended questions, the responses are probably too varied to look like data. Still, count the topics and concepts you notice. You’ll find insightful outliers that deserve more exploration.
  • Survey responses — Weight: 2 or 3. Writing clear, non-leading, useful questions and answer choices is tough. If you nail it, give yourself more points. You’ve got a good idea as to how many people are how likely to do what.
  • Contact form submissions / Customer service insights — Weight: 2. When users submit their complaints, questions, and feedback, that’s your chance to parse and count their unfulfilled needs. Often, users contact customer service when they can’t find content. That’s a gold mine of new wins.
  • Competitor content — Weight: 1. You might perform a competitive analysis for inspiration more than ideas, but if 7 out of 10 similar eCommerce sites use testimonials, it’s probably a pretty good idea.
  • Online reviews — Weight: 3. Closely reading Amazon reviews can show, feature-by-feature, what people genuinely care about, what they hate, and a full emotional rant about why. You might learn how to present your brand better, explain a feature, or alleviate common concerns.
  • Page traffic — Weight: 2. Page visits suggest whether specific content is important or useful. However, that doesn’t mean low trafficked pages aren’t valuable. With poor IA, some worthwhile pages might have just been hard to find.
  • Site search terms — Weight: 4. When the user needs content badly enough that they’ll go so far as to type it out and search, rather than just giving up and bouncing out, that’s valuable content. Internal site search shows what content is currently too difficult to find, informing better IA. And, it’s a free list of what users expect their content to be named, which leads to clearer page titles.

6. Total it up

Note: Airtable lets you create formulas that apply to whole columns (which it calls “Fields”). Here, on my “Content Concepts” table, I make a new column for each research table to count the number of concepts per row, then write a weighted formula to total them together.

7. Work that data

Note: I sorted by the “Total Count Weighted” field, highest to lowest.
  • What can you do to the data? The possibilities are awesome. Here are a few suggestions:
  • Create a table of all use cases, pair them with likely audience types, then filter by each audience type: instant personas.
  • Add dimension by weighing each business goal or product review source individually to hunt for patterns.
  • Tie in content audit spreadsheets to compare what already exists with what needs to be created.

8. Inform your content strategy

  • Information hierarchy: What concepts should be nested, like “Chess” inside “Clubs” inside “Organizations” inside “Student Life”?
  • Priority of navigation items: What major buckets float to the surface? Which funnels should be primary, front and center? Which are better kept in footer territory?
  • Amount of content on each topic: Is there enough interest in this concept to warrant a whole section? Is interest so low that you can nix it?
  • Topics to blog about: Would any of these topics boost your SEO or perform well on social?
  • Content to funnel people to: If information resolves important business goals, but user awareness is low, use internal linking and related content to present content they didn’t even know they needed. How can we balance content we want to feature with content users care about — without getting lost?
  • Define page templates: In CMS development, your unique “templates” (post types, content types, or whatever you want to call them) are your project. Use patterns in your data to pick a set of pages to design that’s flexible, but not too flexible, and covers what the client needs. Outline each template to clearly show how content fits in.
  • Define relationships between templates: Besides content native to each template, think about how the CMS might relate content types to each other by defining reference fields or contextual navigation. Use taxonomies to specify further how content objects within templates work together.
  • Work your content strategy into a sitemap: Too often, people start here, simply rearranging pages that already exist because they feel a couple card sorting exercises are reasoning enough.
  • Define research-driven page goals: I used to summarize everything I learned from users into one purpose statement per page. Now, I filter use cases by page, then view each page entry sorted by most common concept.
  • Wireframing and copywriting: I consider these the same phase, even for two people. If you don’t work side-by-side with your copywriter to outline what content fits the elements defined above, you might negate the whole process by letting copy dictate IA. (Also, letting IA fully dictate copy — or design — harmfully limits the creative input of the team, but that’s another article.)

Obviously, these deliverables are nothing new. I just synthesized them into one deftly contrived spreadsheet.

If you aren’t quantifying content in some way, you’ll notice patterns you wanted to find. Your biases will lead you straight to takeaways that validate what you already comped. When you start from a vision you’re emotionally attached to, too much is at stake to prove yourself wrong. Hopefully, this approach makes the data-to-content translation process less daunting and more manageable. It definitely worked for me.

At worst, we’ll stop spending weeks collecting great, quantifiable, objective numbers, only to undermine the sterile purity of the data by letting our visions cherry-pick stats into stories.

At best, we’ll weave content decisions from numbers into words, like a tiny chunk of a Rosetta Stone between math and English.

Look Under the Hood

For those who want to see or duplicate the example Airtable spreadsheet from above, you can check it out here and make an account. If you feel I’ve glossed over any key background knowledge or would like more resources, get at me!

If you enjoyed this article written for Orlando digital creative agency Purple, Rock, Scissors, please hit the heart below to recommend it to others!