Every so often, someone in SEO announces “f*** keywords” or a similar sentiment. It’s not really a “hot take.” Moreover, it’s something that we see time and again.
Sometimes it’s a fair reaction to how reluctant some people in the industry are to change. Other times it feels more like a meme response, or a flex. A way to show you’ve moved on from old-school SEO without saying much about what you do differently, exaggerating a tension between pro/anti keywords without understanding it.
I have some sympathy for the sentiment for sure. I’ve worked on too many campaigns where success came down to whether a handful of keywords went up or down. Once those rankings became THE GOAL, it was easy to spend more time explaining two dropped places than asking whether we met the primary business need. More damagingly, in the worst cases we were left making decisions just for promoting keywords, rather than the business goals they were supposed to enable.
I don’t know how many people still run campaigns that way, and I suspect fewer than the “f*** keywords” cry implies. What interests me is how quickly the argument goes from “this is a bad way to judge a campaign” to “why are we thinking about keywords at all?”
Which Page Is Catering To What Intent?
Take an example search like [hotels near Heathrow with a shuttle]. It’s a decent shorthand for what someone wants to know before they book. Is there a shuttle to the terminal? Does it run when my flight lands? Do I need to reserve a seat? It also gives you a clue about what language you need to use to gain recognition from the searcher.
If guests are looking for a shuttle, the page titled “ground transportation solutions” is making them do more work than they need to. Again, a fictitious example, but I have lost track of how many times in my career I’ve seen businesses wanting to target the ways they define their products/services, NOT how people search for them.
In this instance, what is more important is that the initial search query becomes emblematic of a series of questions that need answering, potentially across a series of pages. Whether we call this a hub and spoke model, or a content cluster, whatever, we have captured part of the customer journey with a query and then determined the potential related queries.
This is where a keyword map (or site map) earns its keep, long after the initial research is done and the content written and published. Imagine that six months later someone rewrites a particular location page. The shuttle timetable moves into a general FAQ and a new transfers page goes live. All three changes are totally logical, but taken together, they can leave the site without a clear place to answer the question the location page used to own. Without realizing it, you have sleepwalked into undoing the research you have done before.
If you’ve recorded which page serves that need, the next content editor or SEO has something to go back to. They can ensure that intent/search query is still considered, move content responsibility, update the internal links, and ensure the redirects are done. The keyword map preserves the reasoning behind those original decisions and helps you tell the difference between an intentional change and targeting that drifted over a few rounds of perfectly reasonable edits.
The keywords aren’t the single truth, nor are they to be treated as an immutable dogma, but throwing them in the “f***it bucket” could be missing something important.
Too cool for keywords now? Great, but how does governance work? How do you ensure you’re not wasting the money already spent on research? Yeah, I know – that sounds tedious, because it can be. And building a process to make sure this isn’t painfully manual IS hard.
But it is worth it.
What Will You Actually Track?
You can organize your work around topics, and you can use embeddings to help group the language – in fact you should! But at some point, you’ll at least need to benchmark overall visibility, and you still have to choose what to type into a search box or put in a prompt. You can aggregate your rank/prompt tracking for a content cluster – which is needed – but before the aggregation comes selections of individual prompts and terms.
Stakeholders need to buy into those prompts/terms/queries that make up that group OR you 10x the number of specific tracked elements to totally de-emphasize any individual one. Not a bad idea, but you increase your costs significantly and still need to consider the wording, terms and intents on individual pages when it comes to making optimizations.
Instead, I’d happily track a few representative searches for a given page, where I would also want to see which URL(s) appeared and whether they answered the question. Those observations can tell us whether the page is doing some of the work we intended, whether another page is potentially conflicting and where our own content is falling behind competitors.
“Wait, we still use keywords in our data, but we don’t report on the success of individual keywords anymore!” That is perfectly sensible, but it turns out that maybe, perhaps, the “f*** keywords” position is more clickbait than actual smart thinking – right?
These are SEO basics; I totally understand. Prompt tracking and AI search do ultimately de-emphasize the importance of exact keywords in favor of meaning. But consider that embedding text into vectors is so “machines” can understand it. Your team, colleagues and stakeholders still need the human version to understand how to optimize and tell the story of what is/isn’t working.
What Happened Between The Ranking And The Business Outcomes?
All of this helps us understand what we’re visible for, but it doesn’t tell us what happened next – and that is perhaps the biggest weakness.
In Search City, I wrote about the temptation to only optimize for the bits of the search journey we can see. Rankings were always one of the easier bits to observe – that is why we have had an almighty job replacing them with a more useful metric, and why the issues hitting prompt trackers are so disruptive to the industry! But now, even when someone makes a search and sees you for that query/keyword/prompt, the part that links that moment to a click and then an important business event can be almost totally disconnected.
Knowing whether we “rank” for a keyword or not is still an important data point. Fixating on that single data point is mad – but that goes for any single data point. We need to use it to build the picture, build a story and provide insight.
If I went back to any of those campaigns I ran where I felt single terms were prioritized (to a fault), I’d probably still keep some of the same keywords – as triggering as that may be.
I’d use them to agree which searches mattered the most, decide which pages should serve them and check whether those pages still do the job. When it came time to judge the campaign, I’d want to know what happened to the customers/clicks we were trying to win. That was the question the rankings were supposed to help us answer in the first place, and it is still something they can help answer today.
More Resources:
- AI Visibility Measurement: What To Track & What To Ignore
- 4 Datasets On AI Visibility: ChatGPT Sends 0.62% Of Google’s Organic Traffic
- How To Define & Report SEO KPIs That Actually Move The C-Suite
This post was originally published on Chris Green SEO.
Featured Image: Roman Samborskyi/Shutterstock
