Anyone who has done keyword research knows the problem. Finding keywords is easy. Finding the right ones is much harder.
You start with a topic, export hundreds or thousands of keyword ideas, and begin filtering. Some terms are genuinely relevant. Others are loosely related. Some look like perfect opportunities until you look more closely at what the person searching is actually looking for.
You can exclude terms, filter by search intent and use the categorisation provided by SEO platforms. These features are useful, but they don't always get it right. A keyword can look relevant on paper and still have very little to do with the product, service or audience you're trying to reach.
And when you're dealing with thousands of keywords, manually identifying every false positive becomes a considerable task.
How AI helps us choose keywords
We've been using AI to help with keyword research for a while. General-purpose models such as ChatGPT and Claude can help generate ideas, classify search queries and evaluate their relevance.
But we kept running into a problem. The initial results could be useful, yet they still required substantial checking, editing and correction before we could use them confidently in a content strategy.
Recently, we started using Jev, a new AI model developed by TypeSafe AI, to tackle a different part of the process: evaluating keywords at scale.
Jev is designed to make structured decisions rather than generate paragraphs of text. That makes it particularly interesting for repetitive tasks where we need a model to assess large amounts of information against clearly defined criteria.
Instead of asking an AI to produce yet another list of keywords, we can give Jev a list of existing keywords, define what makes a keyword relevant to a particular health brand, and have the model evaluate the list against those rules.
We can establish criteria around the product or therapeutic area, the intended audience, search intent, topical relevance and whether a query fits the website's scope. We can also distinguish between keywords that deserve further investigation and those that should be excluded.
Each keyword receives a relevance assessment that we can use to filter and prioritise the list.
The result? We can process thousands of keywords and spend much less time cleaning up the output.
Why this matters for health brands
For many health businesses, keyword research isn't simply about finding the largest possible search opportunity.
You need to stay in your lane.
A health brand generally needs a much more focused content strategy than a publisher covering health as a whole. Its content should remain relevant to its products, services, therapeutic areas and intended audience. Straying too far from that scope can result in a website that becomes less focused, even when individual articles target keywords with substantial search demand.
This matters for SEO, and it matters when planning content intended to be useful in generative AI search experiences (GEO), too. The aim isn't to cover every imaginable topic related to health. It's to develop a comprehensive body of relevant, credible content around the areas a brand is genuinely equipped to address.
The challenge is that the most relevant opportunities aren't always the obvious ones.
Broad keywords tend to generate plenty of suggestions, but those suggestions can quickly become noisy. More specific, niche queries may have less obvious search demand, yet they can reveal valuable questions and content gaps that would otherwise be overlooked.
By evaluating keywords against the same set of rules, we can filter out irrelevant terms while keeping the more specific opportunities worth investigating.
We can then combine that assessment with search demand, competition, existing website content and SERP analysis to build a more complete picture of where the brand could realistically compete.
The objective isn't to find more keywords. It's to find more of the right keywords, including opportunities that are easy to miss when working through a spreadsheet manually.
What changed for us
What has made the difference for us is not simply that we're using AI. We've been doing that already.
It's that we now have a more efficient way to apply a consistent evaluation process across a very large set of keywords. In our workflow, Jev has significantly reduced the amount of post-processing and correction required compared with the general-purpose models we've previously used for this task.
That means less time cleaning up keyword lists and more time developing comprehensive content strategies: mapping topics, identifying gaps, understanding audiences and prioritising opportunities that fit the business.
It doesn't mean we blindly accept every score. We still need to validate the criteria, review ambiguous cases, check the actual search results and make sure the recommendations make sense for the client. Particularly in health, human judgement and appropriate content review remain essential.
But we can now approach the research with a much more complete set of relevant opportunities, without having to spend the same amount of time manually sorting through the noise.
That's what makes this interesting to us at Vine Digital.
AI doesn't have to write the content to make a meaningful difference to content marketing. Sometimes, its greatest value is helping us make better decisions about what we should be writing in the first place.
Related Posts
Week 47 Health Digital Marketing Newsletter
This week's Digital Health Observer covers the Adelaide University researchers who have given 'subscription medicine' a formal name, commercial digital health platforms, and are calling...
Read MoreSEO for healthcare businesses: why we fix every...
As a digital marketing agency for healthcare businesses, we get our clients ranking on Google and in AI tools through a lot of small, unglamorous...
Read MoreYour tracking cookies could be breaking privacy law
Did you know an IVF clinic broke privacy law just by using a tracking cookie? Like almost every health website does. Source: OAIC media release,...
Read MoreJoin the Newsletter
Want quick and digestible insights, delivered fortnightly to your inbox?
Our newsletter will keep you up to speed with the latest updates and opportunities in your industry.