Using AI for SEO: Dos and don'ts

When the whole world (or at least the whole of your LinkedIn newsfeed) thinks they're an AI expert, it's hard to know which opinion to trust.

Yes, AI can fit into the content or marketing lifecycle at all stages, helping you work more effectively, productively, and creatively. But there are some things you can't do with AI; or some things you can't rely on it fully for.

Here, we're gonna discuss about using AI in SEO—the dos, the don'ts, and some of the exceptions:

Whether you're new to SEO or content, or your product is developing into new fields, it's important that you take part in topic and keyword research.

You can use AI to identify topic clusters and content gaps, and generate related keyword suggestions based on seed keywords. This helps bulk out a comprehensive content plan for your team of writers or freelancers.

You can dive a little deeper too, by using AI to analyze search intent patterns across each of your chosen topic areas to make sure you're ticking all the boxes of what answers or opinions your target audience is looking for.

But remember: Not all AI tools have access to Google search or keyword data to understand what will rank or what would make your content more competitive. More on that in the 'Don'ts' section.

Example prompt: "Taking the topic 'sustainable fashion', identify related subtopics and keyword clusters that would kick off a comprehensive content strategy"

Wherever you are on the spectrum of marketer—from a creative in content to technical in SEO—you'll know that doing content briefs can be (or, just is) a boring task. But you'll also know how necessary they are, especially when it comes to SEO or search-driven content.

Using AI, you can quickly create detailed content briefs for high-level topics, including multiple headline variations to choose from or to A/B test. You can ask the tool to expand on or add in specific sections where you see necessary, too.

Optimizely Opal is embedded in the content brief feature of Optimizely Content Marketing Platform to make this super easy for its users. This means no copy-pasting then clicking around to find the tab you're searching for. Instead, Opal will provide you with suggestions that you can apply there and then. Winning. 🏆

Example prompt: "Create me a brief for a piece of competitive SEO content that's targeting the primary keyword: 'eco-fashion' and secondary keywords: 'environmentally friendly fashion' and 'green fashion'"

When we say 'dishwasher'-type tasks, we mean the ones you need to do... you just don't want to. Why? Because there's zero fun in doing them.

While you can't automate unloading and loading the dishwasher (yet, anyway), you can use AI for SEO tasks like:

👉 Generating schema markup based on content type (eg. lists, recipes, product & offers)
👉 Creating meta descriptions (just tell it to incorporate your keyword or keywords naturally)
👉 Writing optimized image alt text at scale
👉 Spotting internal linking opportunities within your existing content

Example prompt: "Generate schema markup for this recipe page, including cooking time, ingredients, and nutritional information"

Content should never be 'one and done', especially when it comes to search-driven content. Target keywords change, competitors swoop in, #1 rankings come and go...then sometimes come back again.

This is all why content optimizations are a big part of content marketing and SEO. Here are the types of content optimization for SEO that AI can help with:

🤖 Suggesting semantic keywords to enrich your content
🤖 Identifying opps to enhance your existing content
🤖 Generating title tag variations, whilst still maintaining keyword intent
🤖 Creating FAQ sections based on related search (a personal fave)
🤖 Analyzing your website for suggestions to improve conversion rates

Oh, that last one? That might just be Opal. #humblebrag

Give Opal a URL and, as it's powered by Google search, it will give you a complete analysis for how to improve existing content for your target keywords.

Example prompt: "Analyze this content against the top 3 ranking pages and suggest optimization opportunities"

While most of us love being data-driven in 2025—and if you're not, it's probably on your resumé anyway—going through enormous amounts of data isn't always the most fun job. In fact, it's pretty time-consuming and headache-inducing.

🚨 Spoiler alert: Another way to use AI for SEO is to let it do the data trawling for you.

By inputting traffic, keyword, or any other search data you have from other platforms, you can get AI tools to get through the headache for you.

Whether you ask them to track traffic patterns from analytics data, identify trending topics within your dataset, spot any performance anomalies, or provide quick insights from large datasets, they will get the job done quick. And the best bit is, they won't even complain.

Example prompt: "Analyze these monthly traffic numbers and identify potential seasonal patterns"

So, before we set all our SEO hopes and dreams on AI completely, let's just think about this:

Where are the current search volumes coming from?
Where are the competitor metrics coming from?
Where are the ranking difficulties coming from?
Where are the market trend predictions coming from?

You can't rely on AI tools for these up-to-date numbers. Sure, AI might say that 'sustainable fashion' has a high search volume, but without current data, this is just a guess.

For example, keyword research. You can ask for semantic keywords around your chosen topics, but these will be collated from existing online content, with no keyword data behind them. Optimizely's Opal (your AI sidekick) is powered by Google search, meaning it will use the top ranking articles to collate a list of the competitive target keywords.

Solution: If you want to use AI for SEO data analysis, regularly input your data sources (eg. Semrush keyword, traffic, or SERP data)—the better the input, the better the output.

No one knows your brand as much as you do, including whichever AI bot you choose. This means you can't really let AI define your content strategy or set your brand positioning.

AI can't understand your brand's unique voice or market position without extensive training. And even with that training, you can't fully trust it's going to say exactly what you want it to say.

Solution: Well, extensive and continuous training to make sure it's keeping up with your product, business goals, and trends... or make the most of the new era of AI agents, with a brand control agent. Find out all about AI agents in our recent post.

Speaking of AI-generated content and the need of human intervention, it's important that all AI content is reviewed by a human for a number of reasons:

❌ AI-generated content summarizes what's already on the internet
❌ AI-generated content does not have unique opinions
❌ AI-generated content can sound a little... robotic sometimes
❌ AI-generated content doesn't necessarily contain the most up-to-date information

The moral of the story? Avoid using AI for thought-leadership pieces or industry-specific expertise. If we're keeping on the same example of 'sustainable fashion', AI might mention trends that are outdated or inaccurate for a fun, trend-based article.

After all, you need to provide unique spins on content if you want to create impact—not regurgitated content that already exists online.

Solution: Always edit content produced by AI to make sure it's accurate, valuable, and loved-by-Google.

If you're keen to hear more about how to best use AI as any kind of marketer, you should check out The AI Playbook: A practical guide to AI for marketers.

Full to the brim with tips, tricks, and insights, as well as covering the emerging era of AI agents, it's not one to miss.