Why Does No One Click on Links?
I hope to see 100% of you at least choose one...
No one clicks on links, maybe ~25% of users click even one in a giant post.
Just how giant are the Zvi’s posts? When I put “On Bounded Distrust” into a word counter it was nearly 17000 words long—that’s 34 pages single-spaced! There are probably 40-50 links in there. And 3/4 of the voracious readers who actively consume Zvi’s content don’t click any links at all???
What is the deal?
Some possibilities…
1. Zvi is an outlier
Could be, but in which direction? My blog is specifically designed to appeal to infovores and my average “at least one click” numbers seem similar enough. I also contribute to Time Well Spent and there the ALOC’s are much lower.1
2. People don’t like links
I find this hard to square with the tremendous popularity of link-centric posts. You see them literally everywhere on Substack these days, and the people are still clamoring for more. After putting my “Corners of the Internet” series on hiatus, I had a reader reach out to me via Twitter DM requesting that I bring it back.2 Yet despite their popularity, link clicks are only somewhat higher for these kinds of posts.
[You could try to combine with #1 to save this hypothesis except Zvi also does link-centric posts.]
3. Mental Accounting
If most people set aside some roughly fixed quantity of time to spend reading blog posts (or a particular newsletter, post, etc.) then longer posts may generate less link-clicking engagement than shorter posts.
4. Mood Affiliation
Perhaps readers are not primarily interested in information per se, but in the way it is presented. In particular many people seem to care more about the status implications of a blog post than the intellectual content.
5. Sometimes Context is Less Scarce
What purpose do most links serve? While I personally try to make links as interesting as possible to reward digging deeper, in most cases links are simply an efficient way of bringing readers who may be unfamiliar with the basic background of the topic into the loop of what regular readers have seen before.
I find it a little hard to believe that a full 75% of Zvi’s subscriber list needs zero additional context on any given topic he covers, but it’s hard to tell whether I might be underrating this factor without being able to see link click data for non-subscriber visitors.3
6. The Zeroes Problem
Tyler Cowen has written about how psychologically painful it is to click on something and not be rewarded for it.4
The blogosphere, and many other forms of web consumption, keeps you interested by giving you pleasure from the process itself…
Usually a blog will fail if the blogger doesn’t post every day or at least every weekday. People don’t like the idea of visiting the blog and coming away empty-handed, so to speak. It only seems like a visit to the blog is costless; in reality we get a brief pang of pain from “coming up empty.” And once a blog disappoints I classify the site as a “NO.” The site is still only a click away, but for most practical purposes the cost of revisiting the site is now virtually infinite. In my emotional universe that site no longer exists for me and it holds a status lower than the proverbial needle in the haystack.
Internalizing this insight is one reason why Marginal Revolution has been such a successful web page and I take it very seriously indeed.5 Maybe most people’s experience with clicking links is too disappointing to risk being hurt again?
7. Many of you are not truly reading at all
This possibility depresses me.
Normative Implications
What do I infer about people and links from all this? Hard to say definitively without more data to discriminate between hypotheses, but I put some weight on the following possible takeaways…
1. Any link you post should serve one of two primary purposes—pleasing your biggest fans or attracting new subscribers. Avoid in-between stuff that does not truly satisfy anybody.6
2. Use links to make your points seem more authoritative. Hardly anyone is checking anyway…
3. Ignore #2 and become even more committed to providing strong supporting evidence in your links. Readers may take most of your claims on faith, but trust can erode over time in the presence of even infrequent checking.
4. Don’t over-index on Substack likes. Given the small percentage of people who click any links at all, they are not likely to be particularly representative.7
5. Be a little more persistent and obvious when doing self-promotion, erring on the side of “overdoing it”. The Bryan Caplan strategy of linking to your own previous work at every opportunity is underrated because most readers need to be propositioned many times before they will click through.
6. Use links to facilitate your own learning and recall. If a link is personally edifying, but not perfectly on the nose of the word attached to it, include it anyway.
7. If I may be just a little more normative, I think many of you should be clicking on more links. For the cost of a single motion, you may be missing out on something great.
Thanks for reading! If you enjoyed this, I would appreciate if you would share this article with a friend or leave a comment.
In terms of size, Zvi’s blog is the largest of the three and mine is the smallest. Hoping to hit 1000 subscribers soon!
When I checked their reader page, they were subscribed to more than 100 Substacks so I’m sure they’re getting lots of links already! [Also, very grateful for the feedback which I welcome anytime :) ]
This relates to Zvi’s point #6—"Metrics on individual posts are probably better than nothing, but it’s unclear." We need more data and analysis facilitating features!
is awesome at constantly improving things for the hard side of their network so I hope we will see more of these soon.Andrew Chen refers to moments where the user does not get what they wanted from an interaction with a network “zeroes”. Some basic background here, but I recommend reading his book if you are really interested in the broader topic.
I had the chance to ask Tyler once whether the every day rule still applied post-Substack and he said the email newsletter has made things totally different. People check their email habitually already, so posting once per week or every other week can work so long as you aren’t missing coverage people expect from you (e.g. if your beat is the Ukraine war, you better not miss an important Zelensky speech). But the general insight still applies because you want people to be rewarded for opening nearly every single email.
Perhaps choosing one of these two standards to apply to every link uniformly is better yet?
Like and Comment buttons also count toward link clicks for the Substack metric Zvi is referencing and tend to be a very small share of the engaged link-clickers.
I haven't (yet) clicked any links in this post, but I suspect that the explanation is at least partly due to the same dynamic you see with people who leave comments: very few people who read any given post will comment on it. 90%+ of people are lurkers, for better or worse, *even if* they consider themselves infovores. An infovore may like this post, but not click on any of its links, because she has dozens of other tabs open with *other* links that she's clicked from some other page.
Depends on how many read on mobile. I almost always click on links when the desktop browser is open but almost never on smartphone. It's load time+laxk of space, difficulty of getting back to the article