We speculated that an engineer looking for a solution in 2025 would work through the following resources:
- Word of mouth.
- Customised knowledgebases.
- AI-based search engines (e.g. Perplexity.ai).
- Offline resources.
- General ChatGPT type applications.
- The World Wide Web.
Here, we consider the implications for marketers. How can they ensure their content is available to existing customers and prospects on both the WWW and its new competitors?
To clarify, what follows is an early personal opinion. I don’t (yet) know the answers. It will take time, experimentation and reflection to understand this new landscape. Nobody really knows how search engine optimisation works and that has been a thing for over twenty years.
The World Wide Web
We discuss the problems with traditional search engine optimisation elsewhere on this blog. Let’s discuss what’s new. That is Search Generative Experience (SGE) from Google, Bing AI and AI summarizer from Brave.
The technological details behind how each performs are different. As I want to give an overview rather than a detailed comparison, I only discuss Google SGE here.
Google states “SGE is an early step in transforming the Search experience with generative AI. When using SGE, people will notice their search results page with familiar web results, organised in a new way to help them get more from a single search.” They go on to list the advantages –
people are able to:
- Ask new kinds of questions that are more complex and more descriptive.
- Get the gist of a topic faster, with links to relevant results to explore further.
- Get started on something quickly, like writing drafts or generating imagery.
- Make progress by asking conversational follow-ups or trying suggested next steps.
In summary, SGE is much better than a standard search engine at understanding complex queries.
SGE is currently in a testing phase in the USA and many other countries. It will not be available in the UK until early 2024.
Google wants to provide users with quick answers to their queries. One solution is features snippets. These are text elements pulled from web pages and displayed at the top of the search results page.
These push the organic search results down the page and result in fewer click-throughs to the publisher’s web pages. What Rand Fishkin terms zero-click searches.
SGE takes this problem (for content producers) to a whole different level. This post from Authoritas summarises the issue.
What Google pulls into the SGE panel depends on the query – is it informational, learner or transactional? It also (I suggest) depends on the marketplace. Is it B2B, B2C or E-commerce?
Note that some conclusions of the Authoritas study, such as the match between the SGE panel results and the organic results, are contradicted by other studies. This, in my opinion, is because each study looks at a different marketplace.
So what’s the impact on Search engine optimisation (getting your content found) likely to be? This post from Animalz neatly sums it up.
What’s the solution for those trying to ensure their content is found organically in the traditional search results? I am not sure there is a solution.
We all need to accept the game has changed. One thing I do know is the funnel concept is dead.
Think of ChatGPT as a featured snippet generator, but without linking to the source material. Worse still, it does not provide any organic links to other potential sources of information.
Of course, you can ask (prompt) ChatGPT for more information, but the response depends on both your initial question and the content of your prompt. Using a search engine effectively and getting the most out of ChatGPT requires two different skill sets.
To get your content to show in ChatGPT, it must show in the training data. To use an SEO-related phrase, you need to get your content indexed. At present, the training data is out of date (late 2022 vintage), but that will change.
Unlike a click in internet search, the best you can hope for is ChatGPT will pull an element (snippet) from your content. Often, it will combine this snippet with information from other sources. Unlike a WWW search it will not show the full page or post.
How too get your content to show on ChatGPT? One answer could be to use the same approach SEOs use to get content to show in featured snippets.
ML Driven Search Engines
There are an increasing number of ML-driven search engines. Perplexity.ai and Komo.ai are two examples.
As with a traditional search engine, a user types in a search phrase (a prompt), but instead of a cluttered search results page, ML-driven search engines deliver a short block of text that specifically answers the question asked.
The block of text could come from a single web page (like a featured snippet) or other online resource. Sometimes, it is a compendium of several pages/resources.
ML search engines deliver a selection of links to pages should the user want to read more. They also suggest related search phrases the user may wish to explore.
The output of ML search engines is similar to the experience Google intends to offer with SGE (see above). One advantage ML search engines currently have over Google is there is none of the clutter (or Ads) found on a typical search results page.
Another is they are here now and are rapidly gaining traction in the marketplace. Google is still trialling SGE.
There are significant battles to be fought over the coming months. What if ChatGPT starts showing links to resources and related searches? What happens to Perplexity (and others) then? Given Google dominance in Web search, what happens if they get the SGE experience right (de-clutter)?
The marketer would be wise to learn how all options (ML search engines, LLMs like ChatGPT and SGE type search) pull in content. They can’t influence who will dominate, so it is prudent to cover all the bases.
It is difficult to understand the algorithms used by ML search engines to pull in content in response to a prompt. As discussed with SGE (above) it seems to depend on the market. All the marketer can do is experiment and learn over time.
The fundamental problem for marketers is a loss of control. It has become increasingly difficult to deliver content online, the rise of LLM’s and ML-based tools will take the problem to a whole new level.
Some businesses, particularly those in B2B markets, have already found solutions by reverting back to marketing practices that existed long before the Internet. This makes sense, but there needs to be a balance.
Those searching for information online cannot be ignored. Diversification in how information is sought and consumed requires marketers to adapt their tactics.
The emphasis has shifted from trying to rank higher in search engine results to understanding the intricate workings of AI algorithms and how they curate and present content. This shift poses a significant challenge, particularly as the algorithms behind these tools are complex and continually evolving.
For content creators, the increase in zero-click searches and the possibility of their content being summarised or paraphrased by AI poses a real threat. Marketers must consider how their content can be optimised for AI.