Assuming you have no other “macro” drivers on your site – for example, no e-commerce facility, lead generation request from, store finder information, or advertisement click-throughs – how can you measure content engagement?

Here is my list of 10 tangible goals:

  1. Show a snippet/summary first and then require a click to expand for more information
  2. Use ratings e.g. rate this page/article, did this answer you question (y/n)?
  3. Click to read more deeper information e.g. page 2
  4. Social share buttons – Tweet, Like, +1 etc.
  5. Add a comment/review (e.g. a blog comment), or click to go to a 3rd-party site and add there (e.g. product review site).
  6. Click to print, or download the pdf version
  7. Connect in some way to the author e.g. Twitter, LinkedIn etc.
  8. Click to see larger images
  9. Read related articles – good content is always appreciated
  10. Subscribe to similar articles e.g. email, RSS etc.

All of these are reliable ways to track that a reader has *engaged* with your content – when all you have is your own content. It doesn’t mean you should bombard your visitor with all of them. Some will work for you and some will not. Cherry pick the best for your situation.

What to Avoid

Scroll-Depth is a weak metric and quite  simply misleading. What does it tell you if a visitor scrolls to the bottom of the page?

It is true to say that a visitor cannot read to the end of a page without scrolling to the bottom of it (assuming content is below the fold). But the converse is not true. That is, is it not true to say that because someone has scrolled down a page they actually read any of it.

Modified Bounce Rate is the setting of a trigger (event or virtual pageview) after say 30 seconds are spent on a page. Similar to scroll-depth, it tells you nothing about intent. The visitor could have gone for a break, opened a different tab and viewed something completely different, picked up the phone, walked away form their computer, or simply found themselves lost and confused on your pages. Triggering an event when you do not know anything about what happened is misleading.

In either case, forcing a correlation is scientifically unsafe. Its the same point as made here: https://plus.google.com/+avinash/posts/9nisfaKWSww - which shows the folly of false correlations (US murder rate v Internet Explorer market share!)

Note: a forced correlation is not the same as an inaccuracy.