What Marketers Need to Know About AI in SEO Strategies

Explore AI in SEO and learn how tools like ChatGPT and Bard are transforming SEO — from keyword optimization to on-page elements.

The Gist

• AI impact. AI in SEO streamlines keyword matching and placement, enhancing overall strategy.
• Tech limits. AI tools offer non-deterministic results, meaning technical answers can be imprecise.
• Strategy shift. Utilizing AI prompts can eliminate the need for multiple tools, boosting productivity.

People have started to turn to AI tools as frequently as they do Google for search queries. Much of the online buzz focuses on content, a key component of SEO strategy. However, marketers are beginning to ask whether they can also use AI in SEO planning. 

The short answer is yes — with an asterisk.

Each of the current AI tools contain some limitations with how they rely on web data, but they can offer ways to organize the inspection of on-page SEO elements and tasks. The end result is a faster productivity because some of the guesswork in SEO performance management has been removed. 

How AI in SEO Fits Within Strategies

Large language models (LLMs) power ChatGPT, Bard and other AI tools. They have surprised people in the number of ways prompt requests are answered and summarized. But they return results that are not deterministic. Each response to a prompt is an amalgamation of what a given model sees as a best match to the prompt. The quality of the given model’s training data sets the tone for that match. The deterministic quality is why technical answers are imprecise at times.

Current AI tools can identify the best match between keywords and a model's interpretation of a website or app. These frameworks can also be trained to suggest optimal keyword placement within web page or app elements. Utilizing prompts can streamline SEO tasks, condense results, and eliminate the need for a bevy tools to manage keyword options.

Here are a few ideas to start.

Matching A Variety of Keywords to a Page

You can use a prompt to request keywords that best match a specific page of your website. For example, let’s say you have a website for a bakery with one page for apple desserts and one for peach desserts. 

Two factors should be considered. First, limit the initial query to a small set of core keywords. This prevents an overload of keyword choices, allowing for focused page performance and enabling analysts to make logical adjustments as search patterns emerge. While there's no specific number of keywords to target, starting with a range of five to 10 offers a balanced approach without overwhelming analysts.

The second consideration is to use questions in prompts, followed by subsequent prompts to refine the generated keyword options. This approach offers a concise overview of various keyword suggestions, from synonyms that diversify content to keyword phrases frequently mentioned on social media. 

Related Article: State of Search: Top SEO Strategies for 2023

Enhance Pillar Content Choices

In SEO strategy, marketers have long understood the significance of phrases and contexts surrounding product keywords. For instance, a bakery specializing in desserts would build an SEO strategy focused on topics like cakes, pies and specialty desserts. This methodology is central to crafting a pillar content strategy. Pillar strategies aim to organize multiple pages under an overarching topic, allowing a single page to appeal to various queries related to a target keyword.

However, brainstorming a list among a small group of people in the business, even those with considerable topic experience, can miss opportunities for optimally presenting topics on a website page.

To help with brainstorming, marketers can use AI tools like ChatGPT, Bard, Claude or Propensity to identify keywords by topics that support each product or service. The results can indicate how topics and subtopics would fit as pillars that would best organize the product and service pages to work naturally yet appeal to multiple search engine queries. 

Related Article: Sustainable Search Engine Optimization: Build a Lasting SEO Strategy

Identify Frequently Asked Questions by Keyword

An extension of pillar content involves creating question-and-answer pages designed to align with search engine queries seeking specific answers. This web content feature evolved in response to search engines improving their ability to understand and categorize queries to deliver the most relevant answers. As a result, analysts must prioritize which keywords and associated questions are most likely to attract visits and, subsequently, conversions on the website.

One effective approach is to use an AI prompt to simulate a potential customer asking a question, using the keyword phrasing as a guideline. The response then becomes an imagined answer to those questions, shaping page content that serves as a potential reply to a customer's query.

Learning Opportunities

The above serves as a quick example in Claude.AI, outlining the top questions likely to be asked in an online search query.

Generate Blog Post Titles Based on Keywords

Everyone knows that search engines look for keywords in a page's title. However, crafting the right phrases that resonate with readers while also boosting search rank remains a challenge, despite advancements in search engines and tactics over the years. This can make coming up with the right blog title feel like writer's block when trying to match it to your keywords.

With AI, you can generate a series of blog titles based on your keywords. This streamlines the process, providing a starter list of compelling titles designed to capture an audience's attention.

Related Article: ChatGPT: Unleashing Efficiency in Marketing & a New Era for SEO

Plan Prompts to Address HTML Elements on The Website Pages

A crucial on-page SEO task involves determining the optimal placement of keywords within a page, often based on their location within HTML elements. By using an AI tool, you can create a series of prompts to guide keyword placement at each stage of a given website page.

The first prompt could address the meta description. Here is an example I used in Google Bard: 

“Write or create a meta description for this website page. The metadata descriptions must be no more than 160 characters in length. Use the aforementioned keywords from the list above.”

Here is what Bard returned:

Note that in the prompt, I'm instructing Bard to stay within 160 characters, as 150 to 160 characters is the optimal length for a metadata tag to be viewable in a search engine query. The AI tool will craft a description using your chosen keywords while adhering to the set character limit. Well-designed meta tags guide browsers and search engines on how the meta description should appear, aiding users in better understanding the page content.

AI in SEO assists in the design process. The character range I provide sets a context for the prompt to consider when generating an answer, effectively placing guardrails on the responses needed for the tag.

A follow-up prompt can be created, instructing the AI to use the keywords in crafting a page title tag within a character limit. (The goal is to have a title tag that both describes the page's purpose and helps users understand the content.) A third prompt could focus on H1 tags, also providing a specific context. The fourth prompt would generate an outline for the given website page, incorporating suggestions for H2 and H3 tags into the outline. 

Each of these prompts can save time in discovering where keywords can be inserted within a page and can help again in developing a pillar strategy. 

What to Keep in Mind When Using AI in SEO

Marketers are becoming increasingly aware of the mechanics behind AI-generated results. Language models like LLMs can craft fluent responses to prompts, but they do so based on statistical calculations of underlying tokens to estimate the best answer. An LLM gauges what it believes will make sense to the reader, relying on a probability comparison between the prompt text and the data it has been trained on.

For instance, ChatGPT Plus integrates search results, while the free-tier version doesn't compare against current internet traffic and only displays information trained up to 2021. Therefore, users who require more current data for their prompts often opt for a subscription.

Bard returns multiple variations of the responses, so users can gain nuanced choices and select what worked best for their needs.

This implies that you may want to validate the search volume for each keyword to gauge current search activity. You can't assume that phrases popular before 2021 are still as frequently used in 2023, particularly given the significant shifts in global behavior due to the pandemic. To verify activity, use Google Trends, which displays search volume as an index. A combination of Google Analytics and Google Search Console can also provide valuable insights into your site's current performance.

There are more strategies possible in SEO, but guesswork can leave an analyst's wheels spinning in misdirection. The concepts here give a good framework to get on-page SEO tasks started in the right direction.