- Qualitative review can uncover hypotheses that can then be tested broadly quantitatively.
- Quantitative approaches can allow you to more objectively sample content for qualitative review.
- You may be able to drastically reduce the qualitative review effort by quantitatively eliminating items that do not need deeper review.
There are two types of content analysis:
- Quantitative. Automatically analyze web content, for instance via crawling a website, scraping information off each page, and filtering or grouping that information to look for patterns.
- Qualitative. Use judgement to review content, for instance via evaluating tone, persuasiveness, alignment with business needs, and deciding how content should be transformed.
But doing just one or the other is not very effective. Practically speaking, they are rarely done alone. That said, these are some typical approaches of using both, each of which is not effective:
- Doing each one separately. This simply does not allow synergies that would enable even more powerful and efficient analysis.
- Doing the quantitative analysis to get a list of content, then moving to a line-by-line qualitative review of that list. The problem with this approach is that once you start filling out qualitative values (like columns for tone, alignment with business needs, etc) it is difficult to redefine how your standards for how to apply them (like if you are halfway into a review of 10,000 pages you decide the business need is slightly different).
Instead, we should be combining quantitative and qualitative analysis, to allow a dance between the two. In specific, we can:
- Test content hypotheses. In your qualitative review, you may discover what appears to be a pattern. This may happen at any point in the qualitative review. There are many times where you could test that hypothesis qualitatively. For instance, one user of Content Chimera had noticed some extremely long walls of text on some key pages, and assumed that this was a pervasive issue. But we could test that hypothesis by a quantitative review of all the pages (it turned out not being pervasive).
- Objectively sample a site for qualitative analysis. Instead of blindly doing a qualitative review line-by-line, you can use patterns in the quantitative analysis to ensure you are reviewing a wide representative of the content. One of the simplest things you can do is to randomly sample the entire inventory, but you can also be more thorough: sampling different publishing years, sections of the site, types of content, popularity of content, etc (preferably ensuring you combine multiple factors, like ensuring you sample different years of content per site section).
- Prioritize qualitative review. One of the biggest content analysis fallacies is the believe that you have to look at a piece of content to make a decision about it (see Decide, Don't Inspect). But it doesn't need to be this way. You can use rules to cull through your content. In particular, some content only needs sampling of content to determine what should be done with an entire related set of content. For instance, if you look at a sample of press releases that are at least two years old, you may quickly be able to declare "we can delete all these." That allows you to save that qualitative review horsepower for what really deserves the extra attention.
Another way to look at it is that you can alternate between qualitative and quantitative. You qualitatively discover issues, then you quantitatively review them. After the quantitative review, you are able to better drive the qualitative review.