Google, the Rise of Junk Articles and Why it has Become Harder to Find Info on the Web

You’re googling something. The first result seems promising. You open it. It’s a really long, messy article, and you can’t find your needle in this haystack. Know the feeling? It is becoming increasingly difficult to find the information you’re looking for through a simple Google search. Why is that? Read more to find out what we can do about it. TL;DR [Part 1] As we see throughout the article, there is a phenomenon of increasing difficulty in finding the answer to the specific question you’re querying on Google Search. [Part 2] The problem is twofold: : the user does not find what they’re looking for 1. Practical problem : the user is deceived into thinking that they have found what they’re looking for, or may find it by opening source pages. 2. Perception problem Various factors may cause these problems, from Google Search’s algorithm falling short to content marketers using dark patterns to maximize Search Engine Optimization. [Part 3] Regarding the perception problem, I propose some product modifications that may decrease the deception feeling for the user, such as: Allowing the user to expand the length of excerpts in featured snippets Showing the excerpt’s title in featured snippets Landing directly on the right passage of the source page on a Mobile browser. 1. The : it’s harder to find a piece of information through Google what 1.1 Step 1: Google Search Last Sunday, I had just got a tray full of fresh meat from the butcher, but I couldn’t get to my place for several hours. When I eventually did, I was left wondering: “can I freeze the meat and cook it later, or did it go bad?” Without further ado, I took my phone out of my pocket and, my mother tongue being French, I googled some keywords, the French equivalent of: “how long meat left out outside fridge”: Thanks to Google’s fabulous preview results (a.k.a ), I got the answer to my question at first glance. Awesome, right? featured snippets Well, no. Because the displayed answer mentioned “coun[ting] 5 to 6 hours”, but it might have been for a use case different than mine, such as how long meat can sit out after having been taken out of the freezer. So I as a user did not feel confident enough to fully trust the preview result. Therefore, the to my question was too short. It lacked context. featured snippet This was the first failure in my search process. After this, I decided to take the search a step further by opening the source page of the . featured snippet 1.2 Step 2: Opening the first result of my Google Search Here’s how I landed on the web page: Urgh. I landed at the top of the page. So I needed to scroll through the whole thing to find the answer to my question. But of course, the answer was nowhere to be found in the first few lines, I was welcomed instead by a lengthy introduction. Thus, I resorted to scanning through the text to pinpoint the paragraph that was the source for the . My eyes were in scanning mode for the string "5 to 6 hours”. After scanning through the piece the first time, I couldn’t find the right passage. featured snippet You could always retort that once the user lands on the page, they can simply use the ‘search on this page” feature (ctrl+f equivalent). However: It is too burdensome to do it on Safari iOS in my opinion. This feature has become decreasingly effective. Since some web pages are configured to load their content progressively , if you perform a ‘search on this page’ when you’ve just landed on the page, you risk getting 0 results returned, although the content is actually on the page. as you’re scrolling In my case, I persisted and after performing a second eye-scanning, in a slower fashion this time… Bingo: I found the passage. But two problems arose at this point. First, this information was buried deep, at around the 3/4th of the article. I wasted a lot of time trying to find the right passage within the article. Second, although I did find the right passage, I realized the paragraph was talking about how long it takes to unfreeze meat, not how long fresh meat may sit out. , I would’ve used the answer to a question that I wasn’t asking. This is what we call deception. In other words, had I blindly trusted Google’s featured snippet This was the second failure in my search process. So I decided I’d go back to Google and try my luck with the other search results. 1.3 Step 3: Trying the other results from Google Search Back to square one, Google Search’s result page: I tried to open the « people also ask » section, but it didn’t really provide a relevant answer. The next result was from a forum, not a reliable source in my opinion, so I skipped it altogether. The third result’s title mentioned « time of food preservation », and the preview contained « smoked meat », plus it was from a renowned French journal, Le Figaro, so I figured it would be a reliable and relevant option for me. And indeed, I found on this page a table with data on storage time for each type of fresh food, on a cupboard, in a fridge, or in a freezer. However, when I scrolled through the table… … Well, the cupboard column was left empty for the line that was relevant to my use case. This was the third, and last failure in my search process. That’s it for the storytelling part of this article, now let’s see what we can learn and induce from the experiment I went through. 2. The : why didn’t I find the right answer, and why was I repeatedly deceived? why I think there really were 2 problems in what I experienced: : I didn’t find the information I was looking for. Practical one : deception: I repeatedly thought I had found the information, only to be disappointed each time I opened and read the source pages – which added to my frustration. Perception one finally 2.1 Why didn’t I find the information I was looking for? I believe Google Search is at fault here. Of course, Google is often formidably accurate. But not this time. And dare I say, increasingly not. But it would be stopping in the middle to just point fingers at Google. How come the mighty Google is not good enough? 2.1.1 A theory of why Google Search’s performance is declining What I’m writing here is a theory of mine that stems from recent observations, but I feel like more and more websites are trying to fool Google Search. Getting found via Google Search is such a powerful customer acquisition channel that websites resort to all sorts of dark patterns to make it to Google Search’s first organic result – or better still – to make it to becoming the selected page for Google’s , the holy grail of Search Engine Optimization (or SEO in short). featured snippet Let’s take the example of the featured snippet I obtained in my initial search: Well, the first section of the article, “The cold chain,” is pure filler. You need to scroll further down the page to start finding substance, i.e., actual information: I am increasingly experiencing this problem. To the point that when I now open content-marketing-y articles on the web, I instinctively start scrolling to skip a few sections, because I now presume the substance is more likely to be found around the middle/end of the article. At this point, you’re probably all wondering the same thing, the same question you’d like to ask the content marketing cowboys who use these dark patterns. Allow me, in unison, to quote Jeff Bezos here: dear marketing cowboys, Although I am no SEO expert, I can only presume that metrics such as or are important to 1/ show users as many ads as possible and thus squeeze the maximum ad money from each visit, and 2/ get your web page a good ranking on Google Search’s results. And as a consequence, marketers do everything to maximize these metrics. average-time-spent-on-page average-session-duration Another typical dark pattern that is rampant, is reading the excerpt of a , then opening the source page, only to be unable to find the excerpt on the webpage. This is the kind of thing that drives me crazy, leading me to do 10 back-and-forths between the featured snippet and the source page to make sure I am not dumb to have misread the excerpt. featured snippet I could list many more dark patterns, but to put it in a nutshell, it seems that we end-users are stuck in the middle of a perpetual war between content marketing cowboys and Google, the former continuously trying to outsmart the latter. Whatever the reason why Google Search’s performance is declining, let us now consider the second problem, the perception one. 2.2 Why was I repeatedly deceived? A perception problem may seem minor. But being repeatedly misled into thinking that you are presented with the right answer to your question in Google’s previews, only to discover it was not an answer to your specific question when you check out the source page… It’s the feeling of . A sickening feeling. being bullshitted Why did I feel bullshitted? Let’s go back to the first search result I obtained (Google’s so-called ): featured snippets Now, if we analyze the , and particularly its text excerpt: featured snippet “Indeed, count 5 to 6 hours for small pieces of meat and 12 to 24 hours for larger ones.” I think the problem here is that this answer like the answer to my question. But at the same time, the excerpt is too short for me to be 100% convinced that this is an answer to my specific question: For meat conservation in a fridge, on a shelf, …? seems ”Count” for what indeed? This is where ambiguity arises. Therefore, I end up checking out the source page, and – lo and behold – the answer does not match my question. That’s when the betrayal feeling hits. It feels like Google did lie to me by omission. In the end, the root cause of the deception feeling seems to be the shortness of the text excerpt presented by Google. Can we do anything about either the practical or the perception problem? 3. The : how could I have found the right information, how could I have not felt deceived? how Regarding the first problem, I wouldn’t presume to give recommendations to Google, since, on a technical level, I don’t know the first thing about search engine algorithms. On the other hand, I have a few ideas concerning the perception problem. 3.1 Solutions to the perception problem: how could I have not felt deceived? Here are three solutions I came up with that would have eased my deception feeling: Featured Snippet Make the excerpts’ length expandable Show the title of the excerpts Landing on the source page Land directly on the right passage that matches your query and that was shown in Google Search results’ preview 3.1.1 Solution #1: Expandable length of featured snippets Since the reason why I got deceived into thinking that the page answered my question is because the excerpt was too short Why: : A useful feature would be the ability to have a longer excerpt. What : Because it is hard to know whether it would be more useful to extend the excerpt by showing text from or the webpage’s most relevant line, it would make sense to add text from both before and after the “most relevant line” ( ). How before after a.k.a “the needle” : As to should be added before and after the most relevant line, I will settle in my example for [most relevant line] ± [2 lines], but of course, that choice needs to be refined. How how much text : Also, since Google's current excerpts’ lengths are often sufficient to find the answer one is looking for, I think it would be better to offer the user a longer version of the excerpt in an optional fashion, just like Google does it with songs’ lyrics. How All in all, here’s what something that would’ve been useful to me would have looked like: That would be a first way to do it, with the virtue of getting additional clues about whether the answer is the right one for the user’s question without the user having to leave the search page. Also, since Google is doing something similar with songs’ lyrics, they shouldn’t have much problem implementing this change. 3.1.2 Solution #2: Show the excerpt’s title in featured snippets Another way to have more context on the excerpt is to take, from the source page, the parent title in which the excerpt is embedded. You know, text in web pages is structured in such a manner: #Title level 1 #Title level 2 n°1 #Title level 2 n°2 blahblah #Title level 2 n°3 Well, getting the innermost title in which the text is embedded would provide a good clue as to whether the excerpt answers the user’s question: After running the test with the example of my own research, I realize that, sure, I do get more context, but it just adds confusion to the excerpt. Because in this case, the excerpt actually talks about how long it takes to meat, whereas, with my version of the feature, it looks like the excerpt talks about how long it takes to meat. unfreeze freeze Therefore, this feature looks unpromising based on the first test I ran. It does not mean that solution n°2 needs to be rejected altogether, but rather that it would take some trial and error to generalize this solution for all sorts of source pages. Note that solution n°2 is, of course, not incompatible with solution 1. Previews could include excerpts of expandable length + the title in which the excerpt is initially embedded. 3.1.3 Solution #3: Land directly on the right passage of the source page Finally, in the event that the user actually decides to open the source page to check it out, it would be useful for them not to have to go through the whole page with their eyes going into scanning mode to find where the original excerpt is located in the page. Directly landing on the right passage of the text in the source page when the user clicks on the search result is a feature that already exists on Chrome desktop, but for some reason, it didn’t work for me on Safari iOS. This probably means that there is a technical obstacle preventing Google from implementing this feature on Safari iOS. But regardless of technical feasibility, it would allow the user to waste less time and quickly identify whether the page contains the answer to their question. As I mentioned earlier, I don’t believe ctrl+f can be a satisfying workaround anymore; therefore, the “land on the right passage” feature would be more than welcome on Safari iOS. 4. Final considerations on finding information on the web My article takes place in the context of growing discontent toward Google Search. Google Search was a revolutionary tool in the late 90s. Since then, it has improved, and Google (and its parent company, Alphabet) has become the big tech behemoth we’ve all come to know. However, 20+ years later, people are done simply questioning Google’s hegemony, and a wave of new, opinionated search products are surfacing to offer alternative ways to search the web, often by focusing on a single vertical they serve. I happen to currently be a Product Manager at such a company: Doctrine’s mission is to make legal information accessible and understandable. I started being aware of Google Search’s limitations by reading an article by a fellow named DKB. If you’ve enjoyed my article, you should try his: . Google Search Is Dying (Feb. 2022) Thank you, dear readers, for accompanying me through this search journey. If you’ve enjoyed this piece, you can find more on . my Substack PS: To free you guys from the suspense, I did cook the meat for lunch a few days ago… As it turned out, it tasted fantastic, and I’m still well and alive :). Special thanks to Oriane Sarrouy, and Othman Bensouda for reading drafts of this. Also published here.