Last month Google demoed a new technology called “MUM” at their Google I/O conference which uses AI to provide answers to complex user queries, and already articles have started to appear asking Will Google’s MUM Kill SEO?
As someone who has been doing SEO for the past 15+ years, this question obviously interests me. So in this post I will explore the implications of MUM and whether it really marks the end of SEO. Spoiler alert: SEO is not dead yet.
What is MUM?
I won’t spend a ton of time explaining what MUM is (see Google’s blog post on MUM), but in a nutshell MUM (“Multitask Unified Model”) is an AI model that will help Google better understand search queries and provide better answers.
Think of it like GPT-3 on steroids. MUM is multimodal, so in addition to text, it incorporates and can make sense of images and other media. MUM is also multilingual, incorporating content from 75+ languages and presenting answers in different languages as well.
In short, MUM can synthesize content of different types from around the web, and provide answers to user queries. Like GPT-3, MUM can also use this synthesized knowledge to generate content of its own, hence the fears that it will provide answers to searchers directly and eliminate the need for SEOs.
Betteridge’s Law of Headlines
A good starting point for evaluating any sensational claim on the Internet is Betteridge’s Law of Headlines which states:
“Any headline that ends in a question mark can be answered by the word no.”
This is almost always true. Content writers create dramatic titles for their articles to generate clicks, but the answers to their sensational questions, 99% of the time, is simply “no”.
I’ve lost count how many times SEO has been declared “dead” over the years, and yet there are more SEOs working today than ever before.
In the rare cases that a the answer to a question in a headline is not “no”, the answer is usually “it depends”.
John Mueller, the head SEO liaison at Google, had this to say about MUM:
“I don’t really see how this [MUM] would reduce the need for SEO. Things always evolve. Remember the SEO joke about changing the lightbulb? None of that’s been necessary for a while now, which is due to developments like these, and yet, people still have enough to do as SEO.”
So straight from the horse’s mouth, Google does not believe that MUM will put SEOs out of work. But Google isn’t always completely honest with their comments, so let’s dig deeper.
It’s important to note that MUM isn’t live and won’t be fully for some time. Pandu Nayak, Google’s VP of Search, states that “We’ll bring MUM-powered features and improvements to our products in the coming months and years.”
So this isn’t something that is coming out tomorrow and completely changing the Google search engine. This is something that is rolling out over the coming years, and being incorporated into Google’s algorithm, much like previous AI-powered enhancements such as BERT and Rank Brain.
10 Blue Links
Although Google’s SERPs have evolved quite a bit over the years from its initial 10 blue links, it is still essentially a “search engine”, providing search results to content on the web for queries that users search for.
MUM can provide answers to complex user queries, in much the same way that GPT-3 can, but at the end of the day it is still an AI model which will help power Google’s search engine. It is not a front-end interface that is completely replacing Google search, at least for the time being.
As long as the essential function of the Google search engine does not change, SEO will still be relevant. As long as Google continues to return a set of ordered links to sites, there will be ranking factors that SEOs can influence to help one site rank higher than another.
Google is also far from the only search engines these days. YouTube is the second largest search engine in the world, and SEO is still relevant there. Amazon has a search function as well, and is quickly surpassing Google as a search destination for ecommerce queries. App stores for iOS and Android also have search results and factors that can be used to influence them.
As long as “search” exists as a medium of interaction, there will be a role for SEO professionals to optimize for those search systems.
Commodification of Knowledge
With all that being said, there has been a commodification of knowledge over the years, as Google increasingly presents more and more answers to simple queries directly in the search results. This is especially true with voice search, which is growing in search share as more and more devices opt for voice as a medium of interaction.
This means that knowledge is increasingly a commodity with decreasing value. If all your website does is provide simple answers to questions, then you can expect to see the volume of search traffic to your site decline over time, especially with algorithms like MUM which incorporate data from multiple sources and mediums to generate a single answer.
This does not mean that SEO is dead, however.
The commodification of data just means that your website needs to provide added value, in addition to just information. Supplementing information with value-added functionality and additional context will give users a reason to click-through to your site even if an answer is available on Google.
The same is true for editorial content, personality driven content, unique app functionality, etc. There are many types of search queries where Google cannot just provide a simple one-line answer on their SERPs. Google users are not just searching for answers, after all– they are searching the web for relevant content.
To provide a concrete example, I run a website called Salsa Vida SF which provides information about salsa events going on in the SF Bay Area.
If all I did was provide information, it would be very easy for Google to just provide the data in their search results. But the site also includes user reviews of events, a filterable calendar, profiles of local instructors & dance teams, articles about dancing, etc. This is all unique content that is only available on my site, and not something Google can simply display as an answer box in their SERPs.
In short, don’t be a commodity. If you offer unique value on your sites, users will have a reason to visit no matter how advanced Google’s algorithm becomes.
Credit Where Credit is Due
It will be interesting to see exactly how MUM is implemented when it does go live and how it gives credit to the original content creators when it provides answers.
If MUM synthesizes content from multiple sites (in multiple languages) and multiple images to provide a single result, which sites does Google link out to?
AI systems such as MUM and GPT-3 are able generate unique content based on their synthesis of all the content on the web, but because neural networks are in many ways a black box, it may not even be possible to know exactly how such a system arrives at a given answer or for the system to provide proper attribution to all the content creators who contributed to training the AI.
Copyright law likely hasn’t yet evolved to address the issue of AI generated content, but on a certain level this seems like a form of AI-powered plagiarism.
Search engines such as Google depend and make use of the content created by millions of people around the world, but if they don’t link out to the sources of their answers, they will essentially be taking all the value from content creators and providing nothing in return.
This could harm the entire web content ecosystem in the long run, by disincentivizing people from creating content or sharing it freely with Google.
Wikipedia recently started charging big companies who make use of its data through a new service called Wikimedia Enterprise, so clearly they realized how much free value Google was getting from using their content. And yet it won’t be possible for smaller publishers to negotiate similar deals for their content.
Google depends on content created by others since they are largely not content creators themselves, so they will need to figure out a way to return value back to the creators who empower them. Perhaps micropayments and cryptocurrencies could play a part here. They already follow a similar model on YouTube, so perhaps a similar model could be generalized for the web.
Ultimately it is in Google’s best interest to keep publishers and content creators happy, as it will negatively affect their search results if they aren’t able to feed their AI with new and relevant content.
We are still months or years away from MUM being rolled into Google’s algorithm, so all we can do for now is speculate based on Google’s demo and blog post on what the technology will eventually look like.
There is a possibility that web publishers could see their organic traffic erode further as Google provides more answers on their SERPs. But content creators can combat this by focusing on creating content that can’t easily be replicated by a robot.
Hopefully we will see Google evolve to provide even more relevant answers and search results, with proper attribution, and help fulfill their mission to “organize the world’s information and make it universally accessible and useful”.
In any case, it is not likely to spell the end of SEO anytime soon. As long as search results exist, there will be SEOs there to figure out how to get their results higher on the list.