A week or so ago Barry Schwartz, aka @RustyBrick, summed the current mood among many SEO’s with nice parody: “Code Red…Google’s dead…what’re we gonna do…everybody’s panicking.”
(Quick tip: if you’re serious about SEO and are short on time, I always recommend you make time — if for nothing else — for Barry’s weekly updates, either at SE Roundtable or on his YouTube channel)
As an aside: I’d love to hear your thoughts on this article and the impact of ChatGPT: Is it an earthquake, the rumble of a train going by on its tracks, or something in between?
According to a 12/21/22 article published by the New York Times (paywalled content, btw), the “code red” phrase was used by management at Google and has now made its way into the media ecosystem with the ferocity of a tsunami. No wonder publications like the Motley Fool are wondering whether we should be concerned about Google’s stock value (events since then seem to have confirmed their worries). The ‘Fools’ advise us “you’d likely need to spend a lot more time and effort getting information on some topics using Google than you would with ChatGPT. Depending on the topic, ChatGPT is potentially a much better tool for research.”
So it’s not just many SEO pros who are already writing out their Google eulogies. The headlines in mainstream media, the investing press, and elsewhere seem also to be mourning (or celebrating) the passing of Google.
There’s no doubt that generative AI, namely the “Artificial Intelligence” that can generate seemingly original images, text, programming, and soon even music, is changing the Internet landscape dramatically. It can be very, very good at what it does. In the case of ChatGPT, what it does is answer questions, basically. A few sessions with it can have you concluding that it would have a good shot at fooling Alan Turing. Just about everyone thinks it is…cue the time-worn cliché…a “game changer.” My question is, what game is it changing? After all, the possibilities are — if not endless — multiple.
The chat enhancement of ChatGPT introduced by Microsoft recently, which now seems to be running under the name “Sydney” in the popular press, does go beyond answering questions. But not necessarily in a good way, as it’s been observed speculating, pontificating, and sometimes arguing with and gaslighting users.
Whose face has more egg on it, Google’s or Bing’s?
Since I wrote the first draft of this post, even more has happened to draw the attention of the mainstream press to these developments in the world of Search. Microsoft did a very slick demo of their integration of ChatGPT-adjacent technology (at least using the same partner, OpenAI) and the world was understandably impressed with how slick and (seemingly) useful it was — yeah, the “world” includes me. (You can watch the presentation for yourself here.)
Hard on its heels, Google demonstrated its answer to Bing’s “Copilot” with “Apprentice Bard.” It didn’t go so well, because the fact-checkers were busy and saw how Google’s chatbot stumbled and made a factual error. The mistake was high profile and led to a steep drop in Google’s — or rather parent company Alphabet’s — share price, losing the company $100,000,000,000 in value in just 48 hours following the demo (I think it looks more impressive if you put in all the zeros, but that’s $100 billion with a “b”).
In the meantime, now that numerous people in the privileged press are playing around with the OpenAI-powered chat and search capabilities in Bing search. And journalists, being the inquisitive creatures that they are, have now revealed that Bing’s chatbot can be far more embarrassing than Google’s. In fact, Bing’s initial headline-grabbing demo had more factual errors in it than Google’s, but of course since Google’s caught the attention of the press, few people are aware of how riddled with errors Microsoft’s demo was.
As an aside, why didn’t the press catch on to Bing Chat’s factual errors as quickly as Google’s? The only thing that comes to mind is that quote from the 2002 movie “Catch me if You Can”:
Frank Abagnale Sr.:
You know why the Yankees always win, Frank?
Frank Abagnale, Jr.:
‘Cause they have Mickey Mantle?
Frank Abagnale Sr.:
No, it’s ’cause the other teams can’t stop staring at those damn pinstripes.
The fact is, Microsoft had a much smoother, more polished presentation. Marketing matters. At least for a time. But MS’s time ran out rather quickly (I guess that’s a commentary on the limitations of pinstripes — i.e. marketing.)
It’s truly amazing that a pair of weeks that started with unprecedented admiration of the scale of Microsoft’s Bing PR triumph, and which witnessed a very rare opportunity for perennial butt-of-jokes Bing to humiliate and steal a march on Google, ended with all sorts of people having second thoughts about the search-AI alliance at its core. If you want to see the depth of the unease, just check out this, or this, or this. Or especially this.
On the other hand, don’t dismiss AI’s influence on search
It’s easy to be bounced all over the beach with the huge waves of AI innovation crashing down on the relatively predictable world of search. First we think search as we know it is dead, then we think that AI is “obviously” a novelty, and a dangerous one at that, which will go nowhere.
I think the answer is somewhere in between, and so you do need to be paying attention, but how it affects you depends — and here I go back to my starting premise — on what “game” you’re talking about.
If your game is getting a question answered on line
All of us are guilty of it: “googling” a question to get an answer. It didn’t require Open AI for Google and Bing to figure out that people who ask something like “how do I get my cat to stop scratching the couch” are not interested in pouring over a dozen web pages, crafting their research, comparing their sources, evaluating the material, and then getting busy on the solution they’ve chosen. So for years search engines have been enriching their search engine results pages with suggested answers in the form of snippets; hence the term “rich snippets.”
In fact, the SEO business has been decrying the increase of so-called zero click results. Zero click results are searches that don’t result in a click because the searcher gets their answer without having to click on anything. In other words, a zero-click search result will not drive a visitor to your website, even if you rank at an enviable number one for the search query, and whether or not the result was generated by so-called “AI” or simply a well-tuned search engine in the traditional sense.
In fact over a year ago Rand Fishkin over at Spark Toro published a study that found more than half of all searches ended in such “zero click” results. I’ve seen different numbers elsewhere, but they are always frustratingly high for websites that have slaved over content that intend to attract those searchers.
Essentially, ChatGPT or Apprentice Bard (the currently floated name for Google’s answer to it) are zero click results on steroids. So if half of searches are being answered without a click already, a big portion of the “answered on the page” hit has already been absorbed. AI might not do much more than what’s already been done. The next question is, “what’s left?”
If you’re a searcher intending on something other than getting a question answered on line
“It is difficult to determine the exact percentage of searches on Google that are questions as the company does not release this information publicly” — ChatGPT
Thank you for that amazing insight, ChatGPT! (Ugh. Can we come up with another way to refer to it? I’m tired of typing that out, but I can’t think of a clever replacement that people would recognize.)
Yeah, I tried asking ChatGPT about the percentage of searches on Google that are questions. No, really — I was really hoping Chat-gippity (ok, you think of something better to call it) would be able to help. I could not get an answer from conventional search, and the subsequent tedious “leafing” through web pages. Actually, once Google gave me a rich snippet that said 8%, but when I followed the link, I found it was a very suspicious “statistic.” As you see above, I could not get an answer from Chat-gippity. So obviously the fact that a Chatbot is AI neither makes it all-knowing or even more-knowing, at least in this instance.
Regardless, I think it’s safe to point out that if you are like most websites, most of the queries you’ll find in your Google Search Console data are not questions. So then it comes down to the age-old game that Google has been training itself for forever: Searcher Intent. Most searcher intents, IMO, would not be met well by AI.
If a searcher wants information, then yes, ChatGPT or BingGPT or Apprentice Bard or whatever flavor of AI chat is currently top of mind is probably a good option. But not always.
If your search is a how-to, then you likely are going to favor a video. Maybe we’ll get generative AI onto how-to videos, but I’m not holding my breath. Illustrations and near-infinite variations of tasks are a hallmark of how-to, so these don’t play to AI’s strong suits at all. (I now wait to be proven wrong and made a fool of. I’m breathless. It’s suspenseful.) Just look at how Bing bungled the question about pet vacuum cleaners in their demo. Although that search was transactional in nature, i.e. it had a commercial intent, it shows how badly AI can misjudge between similar products.
If your game is helping a business website get exposure (in other words, if you’re an SEO pro)
There are a number of reasons I think you shouldn’t stop planning a large part of your online marketing efforts around Google and SEO just yet.
The most important reason relates to what’s stated above, namely the vast number of searches are not purely informational. If, for example, you are an SEO consultant helping an ecommerce site, then perhaps you’ll need to adjust your “top of the funnel” strategy (that’s the place where you would connect with searchers starting their purchase journey with research) in order to make a connection. However, if you want to connect with your users along the rest of the funnel, where searchers are researching vendors, prices, delivery options, etc., then traditional SEO, as long as you use contemporary ways of accomplishing it, still is the best play.
Naturally you need to evaluate your client’s business model in order to make these strategic decisions, along with the motivations (i.e. searcher intent) of people they need to connect with when those potential customers are online.
If your game is SEO training (wait, that’s OUR game!)
For Search Engine Academy, we definitely will be adjusting what we train to based on the current developments in generative AI. But heck, that’s just par for the course. When a person or organization undertakes to teach “SEO” or even more broadly “digital marketing,” they must resign themselves to constantly changing the training up.
One clear example: Google no doubt is playing the AI game as fiercely as Microsoft and partner OpenAI (for example, while Microsoft has invested over $10,000,000,000 — gotta love all those zeros! —witness Google’s outright ownership of a more enduring and perhaps more sophisticated AI company, DeepMind). But at the same time Google seems to be hedging its bets by emphasizing the attribution of Internet content to trusted content creators with their E-E-A-T framework.
The fact that Google added another “E” to their framework, this time for “experience,” is very telling, since no AI is capable of having “experience” in any meaningful, three-dimensional sense. And until “I Robot” becomes reality, I don’t see that changing.
I mention this because we at Search Engine Academy do pay attention to such things and train to them. Right now in our content development instruction, for example, we are elevating the same things that Google is.
It’s hard to say whether Google, Bing, or some unheralded newcomer will execute AI in search with greater skill and better marketing, but regardless, Search Engine Academy will be following the progression of this closely, with an eye to giving you the best advice possible in protecting and promoting your website and online businesses (ok, so I did a little advertising there at the end, but hey, I think I kept it pretty painless).
What are your thoughts? Or better, how would you like us to help you through our classes and training? Please let us know in the comments below.
Great read. I, for one, am excited to see how the search engine world will adapt to this new development.
Thanks for the feedback, Nick!
As always, thorough breakdown and good insight.
I do think these language models are a game changer for a lot of industries, but search as a whole is relatively safe for now.
Thanks for the comment, Eli. I tend to agree, but I do think it will modify how we interact with search.
Great insights Ross! And possibly the biggest change in the arena for many years, if not ever.
Thanks for the response, Sue. It’s the biggest change…until we find out what they have in store for us next year 😉
Such a great read. So interesting witnessing all of this new technology firsthand & how we all will adapt to it. Thanks so much for sharing!
You’re welcome, Amber. Thanks for reading.
Loving what I’m seeing so far from the various AI programs. Even with the limited experience I’ve had with it so far, it’s exciting (and terrifying) at the same time. Thanks for the insightful and humorous take on it so far.
Steve, you’re welcome. It’s interesting how “exciting” and “terrifying” can feel very similar (which is why I avoid rollercoasters!)
What an excellent and informative article! Thank you…. I am researching this whole area as a potential career change and your links to other articles were so helpful!
Glad this came through at a good time for you, Trish. Thanks for the feedback.
I think people are just using the headline “Google is dead” to gain attention. They know very well as every other person that Google and SEO are not going anywhere. They will only revolve.
Ali, I tend to agree about the attention-grabbing bit, but there are a few people who actually do believe it. Like this guy: https://fortune.com/2023/02/01/gmail-creator-predicts-ai-bots-chatgpt-will-destroy-search-engines-minimum-two-years/