You are currently viewing AI Chatbots Make Organic Search Marketing Harder, Not Easier

AI Chatbots Make Organic Search Marketing Harder, Not Easier

If your Twitter feed is anything like mine, then it’s been inundated with hype about AI chatbots and how they will make our lives as organic search marketers easier. GPT-4 just released, and Google is still assuring us their offering is coming, so I don’t anticipate the torrent of headlines to stop anytime soon.

Contrarily, I think AI chatbots will actually make our jobs harder for a few reasons:

  • AI chatbot results mean we’ll suffer more zero-click results.
  • ChatGPT won’t help Bing take significant search market share from Google
  • Writing content the hard way (by real people, with real experience) is more necessary than ever.

This isn’t to say that AI chatbots can’t do anything useful for us. Tactically, using AI chatbots for things like content outlines and writing non-essential text will save us time. But they can only do so much.

The question on everyone’s mind is, how will AI chatbots impact the search engine landscape and our subsequent strategies? While some people are worried that AI chatbots will result in the fundamentals of SEO and content marketing being thrown out, at a high level, they will actually make those core principles more important than ever.

What is ChatGPT & How Does it Work?

ChatGPT and its kin are conversational text generators, and they’re pretty good at it. They should be if it takes billions of sequences of text and millions of dollars to train. 

Fundamentally, AI chatbots like ChatGPT cannot think. They are just like other software that mechanically output data in response to some input. What is notable about them is how well they imitate a real person’s response. 

ChatGPT belongs to a family of neural networks called Large Language Models (LLM), and they are typically trained with two major steps – unsupervised training and supervised training.

In the unsupervised training step, an LLM will be given an incomplete sequence of text (sentences and code) and asked to predict the next word or symbol. This is how it learns what natural language looks like and how it “knows” anything. Since the text it’s trained on is from the web, it is very good at repeating text that has already been published.

Example of LLM training from Molly Ruby

In the supervised learning step, the language model is tasked with generating several answers to a prompt, and humans tell it which answers are the best. Also called a “fine-tuning” step, this human feedback specializes the model for certain tasks. For example, ChatGPT can respond to prompts for writing an email, but a model with a lot more of its human feedback geared toward email could perform better at the task.

Example of supervised training from ChatGPT

A large language model specialized for web search might be fine-tuned on prompts that are questions people type into the search bar, while the responses are good summarizations of the pages that appear in the top search results.

What AI Chatbot Results Look Like in The SERPs

Bing has already allowed some users to see chatbot results in SERP, and they look like this on desktop:

Screenshot of an AI-powered SERP from Bing
Example of a SERP from Bing

After expanding the box, we can see that it has a few other features: links to citations, prewritten follow-up questions, and a link to expand the feature to the whole screen:

Screenshot of the expanded information from an AI-powered SERP from Bing

Presumably, this feature will only show on search results where Bing thinks it will help users, instead of all of them when it’s finally opened up to everyone. A lot like a direct answer or featured snippet result.

If a search result doesn’t have the feature, Bing provides a link in their “Scope Bar” to take the user there directly:

Screenshot of Bing's Scope Bar with the Chat button highlighted

Google doesn’t have any public previews yet, so we’ll have to rely on what they’ve shared to get an idea of what’s coming:

Via Google’s Blog

Google’s preview isn’t from their final release, so it probably isn’t complete. However, we can expect that they want it to act like an extended featured snippet.

AI Chatbot Results Mean More Zero-click Keywords

Contrary to a lot of fears about featured snippets, I find that featured snippets bring in a lot of traffic. Why wouldn’t they? The snippet is large, at the top of the page, and includes a prominent link to the source page. With a little bit of investigation and copywriting, it’s totally doable to get featured snippet results.

Getting featured snippets like this one doubled the traffic to this page

The real zero-click results are the ones that come from the search engines themselves. Both Google and Bing have knowledge graphs made up of facts they’ve scraped from other websites. Knowledge graph results don’t link to sources and deprive websites that rank below them of valuable clicks. Search engines love them because it keeps users on their site because, hopefully, the user will search for something else that will generate ad results.

If search results like this hurt your bottom line, AI chatbots are going to make it worse for you

Will AI chatbot results be more like featured snippets or more like zero-click results? I’m betting on them being more like zero-click results, but I want to be wrong.

Organic Attribution is Important for Traffic

Wanting attribution in AI chatbot results makes sense for marketers. The search engines trained their large language models on our content, so they could at least cite their sources. 

Bing’s ChatGPT results do link to sources, including pages not already in the top ten positions, but they aren’t immediately visible on the SERP.

Screenshot of an AI-powered SERP from Bing
Users have to click “See more” to see attribution links

So I expect Bing to deliver some clicks to us but not many, provided they choose us as a source.

With Google, their preview screenshot didn’t show any attribution links, but they did make a vague allusion to continuing to give brands traffic from AI search features:

No promises and nothing specific. I think we’ll hear more about ad formats in Bard (Google’s upcoming AI chatbot) before organic attribution.

This is a big problem for organic search marketers because search engines are reneging on the informal arrangement they imposed on us.

The better content we produce, the better results search engines can display to users. If users are satisfied and keep coming back, then they will click on more ads. Search engines need marketers to enhance the value of their products. In return, the search engines give us some traffic, so we’ll continue to stay in business and keep delivering content to them.

If AI chatbots can summarize our content and cut us off from the traffic that sustains us, we’ll be in a very bad position.

ChatGPT Won’t Make Bing Competitive

Historically, Bing’s problem when competing with Google is that it tails whatever Google is doing. Whenever Google comes out with a new feature, Bing presents their own months to years later. Even their snippet design is like Google’s from a few years ago:

Google 2019:

Screenshot of a Google SERP result from 2019

Bing today:

Screenshot of a Google SERP result from 2023

With so much hype around Bing adding a ChatGPT-based feature to their search engine, we should expect to see more traffic coming from Bing to our websites. Many marketers hope this will happen because it might help dislodge Google from its monopoly position in web search.

Unfortunately, users aren’t rushing to Bing like it’s a Black Friday sale. Looking at a few of the websites I work on, Bing traffic has increased but not as much as expected from a revolutionary new feature.

Here are some screenshots from Universal Analytics from the last six months, year over year:

Screenshot of YoY users in Google Analytics
Organic Bing traffic from Portent, a regional HVAC & home services company, and a recipes site

Traffic has been up, but not to Google-killing levels. News about ChatGPT in Bing started in January 2023, but their big ChatGPT announcement didn’t come until March 5th, which is when search interest peaked. Bing also had a significant algorithm update beginning January 18th, so we might see the effects of that or both. Either way, traffic from Bing needs to be several times larger before it becomes an important traffic channel.

I think a major problem holding Bing back is its design. Not just because they are tailing Google but because they don’t take it as seriously as Google. Here are the organic results for “vacuum cleaner.” Where does Bing want us to focus our attention?

It’s a mess. Each feature on this SERP is screaming for the user’s attention, and it shows they follow a different design philosophy from Google.

Focus on the user and all else will follow. Since the beginning, we’ve focused on providing the best user experience possible. Whether we’re designing a new Internet browser or a new tweak to the look of the homepage, we take great care to ensure that they will ultimately serve you, rather than our own internal goal or bottom line.

It’s disappointing that Bing doesn’t seem like it’s going to take a much bigger chunk of the search engine share. At Portent, we’ve always said that diversifying your traffic sources is a good idea for managing risk. 

For many of my accounts, roughly 90% of their organic traffic comes from Google. What if Google wipes it out after an algorithm update? Where are they going to get their qualified traffic from instead? Bing’s 5% and DuckDuckGo’s 1% aren’t going to cut it. Not every brand can make video or social platforms their primary source of traffic, and Google’s dominance puts them in a risky situation.

The sooner Google’s monopoly ends, the easier we’ll all be able to sleep.

AI Chatbots Raise The Floor for Content Quality

Many brands are excited at the prospect of using AI chatbots to write their content because it’s fast, and their pricing is cheap compared to a human writer. 

The problem with adopting this content strategy is that their competitors can very easily do the same thing. Since AI chatbots only “know” what has been published before, they will all be drawing from the same pool of information to repeat. I’m expecting search results to flood with pages all saying the same things in various ways.

How will the brands who are using ChatGPT to write their content have a competitive edge over everyone else using the same writer? They won’t.

Additionally, if search engines are using the same class of technology to answer user queries directly, why would users read their AI-generated content instead? They’re also failing to differentiate themselves from search engine content. 

Even further, what incentive do search engines have to rank a brand’s AI-generated content over what the search engine produces? History has shown that Google prefers to rank its own products in organic search where it can. In a sea of undifferentiated content, Google will probably rank itself higher than anyone else on average.

As the capabilities of large language models advance, the content quality floor for viable content in organic search will rise. The next key question brands will need to answer before they publish a new page is, “is this content better than what an AI chatbot could produce?” If the answer is no, then that page is not likely to rank in the top 10 results anywhere.

Google Search is still meritocratic to some degree, so there must be a way to stand out from the crowd of AI-generated content. Fortunately, the answer has been around for years, and it’s known by many names but mostly as “10x content.”

Google basically agrees with the principles of 10x content, but they call it E-E-A-T. They have also anticipated the homogenizing effect AI-generated content will have in organic search and are already advising their search quality raters on how to differentiate superlative from average:

A table outlining Experience, Expertise and Authoritativeness

This table and other advice gleaned from Google’s Search Quality Rater Guidelines are going to help guide us out of the oncoming AI-generated content swamp.

The Winning SEO Strategy Hasn’t Changed

While AI chatbots are creating new challenges, the strategy for achieving success in organic search is the same as the old one: have real people provide new information based on real experience that goes beyond the core answer to a user’s query. 

Focusing on this fundamental strategy for high-quality content also addresses the three problems I listed at the beginning of the article:

  • Search engines won’t show zero-click results on queries where users clearly prefer our content.
  • We’ll be able to get more traffic from whatever search engines are available. Even video and social because the formula applies to them too.
  • Real people with real-world experience can outperform AI chatbots because only they can bring new information to the web.

The post AI Chatbots Make Organic Search Marketing Harder, Not Easier appeared first on Portent.