The title of this post is a question I’ve heard a few times recently, including after my last presentation at the Engage conference in Portland. It’s a good question, and one worth thinking about.
Although I don’t have access to the algorithm, I do have experience and some ideas about how Google and their algorithm might interpret the problem.
First, let’s consider the two most important layers of Google’s algo (according to the source itself): content and links. If links are to matter less in this combinatory equation, content needs to matter more.
So, how is it possible content can matter more in one area, but not others?
Topics With Indistinguishable Results
On some query sets, it’s quite possible an answer to the searcher’s problem may not be immediately achievable, and to the end user, several examples may be close to equal.
For example, let’s take one search result quite close to my heart: [content marketing services]. On this search result, there are four agencies listed 1-4. Each is solid, and to the end user, clicking the first one wouldn’t do much to inform satisfaction.
Most people would want quotes from multiple agencies. They’d open 2-4 examples, browse around, and probably contact two or more of them, depending what they like or don’t like about each. Some people might like one because they specialize in social: another might like us because we specialize in search.
In the end, there wouldn’t be clear preferential engagement signals gained in Google’s eyes. There’s not much to differentiate from a title perspective, most users aren’t going to be immediately satisfied/will dwell on each business, and in the end, the measurable engagement margins will be thin.
As of today, Google can’t see signed contracts. Or at least I think they can’t.
Because of this, it’s quite possible links have a higher than normal impact. Engagement signals are likely to be neutral among all the results, or at best, not a clear differentiator.
On the opposite side, think of a search such as [premium electric SUVs], where we know Tesla will have a significant comparative clickthrough rate and dwell time engagement advantage compared to most other players, no matter how hard they try.
Here, links probably wouldn’t matter as much (not that other sites are likely to win there, either). Branding, CTR signals and content quality all predominantly favor Tesla.
On other search results without branding at play, this phenomenon can still occur where the preferred website can’t be clearly determined on surface level engagement metrics.
Looking at a search result like [adult coloring pages], Google can get a clearer sense for user satisfaction. One looks around, finds something they like, and prints it out. If a user doesn’t find it with the first result, they bounce and go somewhere else.
Because it’s possible to achieve satisfaction quickly with this query, there’s a bigger chance Google can more clearly differentiate the best from the worst using their surface-level metrics.
Some search results don’t have easily discernible winners from an engagement perspective because the answer takes a long time to resolve, or may be balanced between multiple options and user intents. For this reason, links may be a bigger lever.
Topics Without CTR Differentiation
Some markets don’t have identifiable brands or ways to differentiate reliably with title tags. This means one clear engagement signal, clickthrough rate, may not be reliable as a differentiator.
Take the topic [dental practice management software] as an example. Here, there are two big software leadgen players and software providers as well, but to 99% of the market, there are no recognizable brands available to click.
That, in combination with title tags that don’t really make any of the dental software companies (or leadgen players) stand out, means that the CTR/branding signals here may be almost null for this topic.
Because of that, Google only has dwell time/bounce back signals to rely on. When combined with the fact that this is a small market overall, it means that Google may need to rely on links more heavily to overcome the lack of differentiation from other signals.
It should be no surprise looking at the winners on this topic, then, that links will power the top results more often than not.
On some search results, CTR may be an easily relied-upon metric because some brands have clear preference, or qualities of that topic may make many title tags more clickable than others. In markets where this might not be possible, Google may have to rely on links more heavily to differentiate results.
Topics Without Significant Sample Size
In some industries, there simply isn’t enough demand for a given topic or set of topics to expeditiously make a conclusion as to what the best content is.
Google is of course using data to make their conclusions. So what happens when Google gets data back, but doesn’t have a clear lean in their results? They probably have to rely on more consistently reliably pieces of their algo — like links.
Take the search [VoIP software] for example. Wikipedia ranks first and according to SEMRush, generates around 655 searches per month. The average CPC is $22, so it clearly has some value to someone, even if in a slightly smaller market.
If the total topic size is around 1000-1500 searches, and every keyword is a bit different in terms of CTR and intent, it’s highly likely you’ll get lots of mixed signals.
If you’ve ever run an A/B test, you know that you need a pretty sizable sample to achieve statistical significance. The sample needed gets bigger when your normal conversion rate is on the low side — around 1-5%. Check out the below chart from Optimizely that shows how long to run a test, visualizing the sample needed to statistically hit each stanza.
If you have to see a 25% change to warrant switching designs, you may only need 13,000 impressions to make a decision to hold your current page or not (and you’ll most likely hold). But if you’re open to a 5% win on your 1% conversion rate, you’ll need 458,900 impressions to make that decision.
Think about that from Google’s point of view. They’re not only relying on one data point of course (unlike an A/B test), but if the two biggest factors are content and links, and we assume “content” basically means engagement in most competitive markets, they’d need a pretty big sample to achieve confidence — a sample they won’t be able to get to too frequently with only 1000 to 1500 monthly searches.
In areas with smaller average topic sizes, it’s possible links may be a bigger factor because statistical significance from an engagement perspective may be hard, if not impossible to achieve. Many B2B markets come to mind in this regard.
Where Do Links Matter Less?
If you take all of the above factors together, you might be able to come to a natural conclusion where links might matter less. Where brands are clearly differentiated, B2C industries, and in topics/areas where the best result is constantly getting better and better with time (review markets come to mind).
That said, it’s not to say that links don’t matter. They’re almost certainly a variable at play. When you’re in the car insurance vertical and every player that’s heavily advertising is pretty equal in terms of brand, you suddenly get closer to breakeven again.
Every market is a confluence of signals. It just so happens in some of them, one piece of it lends to market dominance for that brand or a set of brands. But in others, those same strong signals put them on even ground, and the other variants may differentiate.
Overall, I believe links still have a big place. And in some areas, it may be possible that they have an even bigger place than others.. because no engagement data point may be clear enough to clearly tell us the winner.