Microsoft Bing: Verbs Matter More In AI
Verbs as a part of speech and generative artificial intelligence (GAI) have linked search engine optimization (SEO) and search advertising together. This happens frequently in contextual advertising and creative content.
Verbs have become more important in queries, Microsoft Bing Principal Program Manager Fabrice Canel says. But if the user does not specify a verb, then it is all about understanding the content of the query in a specific chat. Even adding the word “do” and related words like “done” and “does” can change a query.
When GAI became part of the equation, it required search professionals to spend much more of their time trying to keep up with changes Google and Bing made to the technology, although the platforms believe automation using this technology would free up time. It does, but there is a learning curve.
Lately, the challenge has become a balancing act between traditional search and the evolution of search into GAI.
Canel shares some of those answers in an interview with Jason Barnard, CEO of the software company Kalicube, where he explains how AI search is different from regular search, what SEOs should do now, and ways to prepare for the future.
Interactions with Bing Chat, similar to Bing Search, help the company learn more about human behavior through natural-language queries — information that can be integrated into Microsoft Advertising.
The complex user queries — questions like “I’m having a dinner party with six people and three are allergic to gluten, so what should I serve?” — provide insights — not so much with this specific query, but the words within the query. The chatbot will return a series of lengthy answers, allowing users to ask follow-up questions.
SEO professionals need to remember that sometimes the focus should be on queries to explore the web, and other times it’s about getting details about a specific topic, Canel said.
Clicks from AI search queries are still valuable to Bing. Barnard asked about something that Canel referred to as a “perfect click.” It’s intended to reference traffic to websites that originated from a chatbot query.
The context in the interaction between the AI search query and Bing results in more relevant answers — and should result in better optimization of ads through Microsoft Advertising.
The technology is not about keyword-term matching in the query to keywords on a webpage. It must understand the context.
The example given includes the word “book” — which can mean to book a flight on an airline or a book to read. The focus was always on context, but with longer queries, it’s time to move past keyword matching. More verbs give Bing the ability to better contextually understand the query.
Marketers also need to start thinking about these practices on brand websites, as more AI creative platforms begin to pull details from information already on websites to build ads, such as products recently announced separately from KERV Interactive and Teads.
When asked whether Bing Chat and Bing Search algorithms are the same, Canel said Bing benefits from the indexes, but it’s not a large language model store. The benefits come from having deep interaction with the index — which is dynamic, not static.
And while the technology is the same, the chat content is more complex in terms of retrieving and analyzing the content.
“Chat gives us access to more time to do a little bit more things, understanding, also interacting deeper with the user via the chat experience and session, where we can also not only suggest text — suggest verbs that the customer can do to continue the discussion with a search engine to retrieve the best content on the Internet,” Canel said.
He says there is still too great a focus on the search-engine results page and not enough on the search experience to create the “perfect click.”
Other suggestions from Canel include that SEOs adopt IndexNow to get indexed faster — sometimes in seconds — as well as to have content based on a basic template, to not forget the basic national-language processing form, and to use a template. Building an index on the LLM can take years, he added.
Canel suggests avoiding “crazy Ajax calls” to retrieve the content that the developers say will help, because “for a search engine it will be a disaster.”
Helping the search engine understood the content, he said, means adding HTML tags and HTML titles to differentiate the headings from the paragraph, adding structured data to the index and helping the large language models understand the content.
(4)
