How to Choose the Best Keywords for SEO (Stop Optimizing for Volume)
TL;DR
- Search volume is a lagging indicator of keyword quality. The better first filter is Revenue Signal: how closely the search intent matches your buyer’s decision stage. High-volume informational keywords convert at 1% or less, while comparison and alternatives keywords average 8.43% conversion rate in a Grow and Convert study of 95 articles and 123,000 pageviews.
- Google AI Overviews dropped organic click-through rates by 61% for informational queries in a Seer Interactive study of 3,119 queries across 42 organizations. But websites cited inside those AI Overviews saw 35% higher organic CTR. Your keyword selection now needs an AI Visibility filter, not just volume and difficulty.
- The Three-Lens Keyword Filter (Traffic Potential, Revenue Signal, AI Visibility) replaces the outdated volume-plus-difficulty approach with criteria that reflect how organic search actually works right now. Apply Lens 2 first, before you ever open a keyword tool.
Last year, I handed a client a keyword list I was proud of. High monthly volumes, manageable difficulty scores, clear informational intent across every topic. Textbook. Twelve months later, those articles ranked. Traffic showed up. Conversions? Nearly zero. The problem wasn’t the execution or the content quality. The problem was that I’d been selecting keywords with the wrong criteria for a search environment that had already moved on.
This was 2024. Google was just starting to roll AI Overviews out aggressively. The informational keyword category I’d leaned into was quietly getting hollowed out by AI-generated summaries that answered the query without sending a single click anywhere. I learned this the hard way.
Here’s what this article gives you: a three-lens framework for evaluating keywords before you commit to writing a single word. You’ll use it to filter out high-volume traffic traps, find conversion-signal keywords your competitors are sleeping on, and choose content angles that get cited by AI engines instead of summarized away by them.
The Problem Isn’t Your Keyword Tool. It’s Your Evaluation Criteria.
Somewhere on your hard drive, there’s a keyword spreadsheet. It’s sorted by monthly search volume, descending. The first 20 rows are dominated by head terms with difficulty scores in the 70s. You scrolled down, found a few with “manageable” difficulty, and those became your targets. That’s the playbook. Everyone runs it.
And it explains why so much SEO content produces traffic but not revenue.
Search volume measures how many people type a query. It says nothing about whether those people want what you sell. A keyword getting 40,000 searches a month might be owned by a completely different audience than your customer. Meanwhile, a 300-search-per-month keyword aimed at someone who’s already comparing your category can convert at 8% or better.
Grow and Convert ran the numbers on this explicitly. Across 95 articles targeting different keyword types, generating over 123,000 organic pageviews, they found that typical informational blog posts convert at 1% or less. Comparison and alternatives keywords (think “[Competitor] alternative” or “[Tool A] vs [Tool B]”) averaged 8.43% conversion rate. That’s not a rounding error. That’s a fundamentally different asset category.
Think of it this way: a high-volume broad keyword is like a mall entrance. Thousands of people walk through. Most aren’t looking for your store, and many aren’t even shoppers. A precise, high-intent keyword is like someone who walked directly to your storefront and asked, “Do you carry the thing I already know I need?” Volume tells you about foot traffic. Intent tells you about buyers.
The framework below fixes this. Three filters. Applied in order.
The Three-Lens Keyword Filter
A keyword that clears all three lenses is worth building content around. A keyword that fails Lens 2 isn’t worth your time no matter how attractive the volume looks. And a keyword that fails Lens 3 might be a trap: you’d earn the ranking but not the clicks.
Lens 1: Traffic Potential (Not the Number Your Tool Shows You)
Traffic Potential is the realistic number of monthly visits a keyword could send you at your expected rank position, accounting for actual click-through rates, not just raw search volume.
Here’s what the standard guides skim past: 92% of all keywords get 10 or fewer searches per month. And over 70% of all search queries are long-tail terms. The implication is that the “high-volume keyword” category you’re probably targeting represents a small slice of how people actually search, and it’s the most contested slice at that.
The smarter play is evaluating keyword clusters instead of individual terms. Twenty related keywords averaging 500 monthly searches each is a better opportunity than one 10,000-search keyword where you’ll realistically land on page two. The cluster builds topical depth. The home-run term builds frustration.
On keyword difficulty scores: treat them as a rough starting signal, not a ranking verdict. A KD of 45 sounds manageable, but if the top three results are government pages or Wikipedia entries with tens of thousands of backlinks, that score is lying to you about the actual competitive bar. Always open the actual SERP and look at who’s ranking. Check referring domain counts directly. Two of the top five results having fewer than 30 referring domains is the real green light for a small or mid-sized site.
Pro Tip: Before you tag any keyword as “achievable,” search it in incognito mode and count how many of the top 5 results are from sites in your domain authority range. A KD score of 30 on a SERP full of Forbes, HubSpot, and G2 is not a KD 30 opportunity for your site.
Lens 2: The Revenue Signal (Where Intent Meets Your Actual Product)
This is the lens that most keyword guides nod at and then immediately ignore.
Revenue Signal is the intersection of two things: the searcher’s intent (informational, commercial, transactional) and how directly their goal connects to what you specifically sell. A keyword can be transactional and still have zero revenue signal for your business if the buyer is looking for a competitor’s product category.
“The #1 thing dictating your conversion rate is the search intent of the person searching and how well the query aligns with what your product does.”
— Benji Hyam and Devesh Khanal, Co-Founders at Grow and Convert (Source)
To score a keyword’s Revenue Signal quickly, ask three questions in this order:
- Who’s searching? Is this person doing casual research, building a shortlist, or actively comparing options to buy?
- Does solving their problem require your product? Not “something in your category.” Your specific product, or close enough that you’re a direct answer.
- Are they aware they need a solution? The closer they are to that realization, the stronger the conversion signal.
Long-tail keywords score well on all three, which is why the conversion data is so lopsided in their favor. According to data cited by the Revenue Marketing Alliance, long-tail keywords average a 36% conversion rate. Compare that to short-tail head terms. The specificity is the signal. When someone types “project management software for remote engineering teams under 20 people,” they’ve already done the research. You just have to show up and be the best answer.
And don’t skip “Jobs to Be Done” keywords just because they look informational. Queries like “how to collect video testimonials from customers” look like how-to traffic until you realize that the searcher needs software to do the job. If your product solves the job in the query, the revenue signal is strong regardless of the informational framing.
Lens 3: AI Visibility (The Filter Almost Nobody Is Running)
Let’s talk about the data that should be making every content team rethink their keyword strategy right now.
Between June 2024 and September 2025, Seer Interactive tracked 3,119 informational queries across 42 organizations, analyzing 25.1 million organic impressions. Their finding: organic click-through rates fell 61% (from 1.76% to 0.61%) for queries where Google AI Overviews appeared. The broad informational keyword category, the one most “how to choose keywords” guides tell you to target as a traffic funnel, just got its click value cut by more than half.
But here’s the part that doesn’t get talked about enough. Websites cited inside those AI Overviews saw 35% higher organic CTRs and 91% higher paid CTRs compared to non-cited pages in the same search results. Being the cited source is now a better outcome than ranking organically at position three.
So the question your keyword list needs to answer isn’t just “can I rank for this?” It’s also: “if an AI Overview fires on this query, will my content be the one it pulls from, or will I rank and get zero clicks?”
Three keyword characteristics that correlate with getting cited in AI Overviews and AI engines like Perplexity and ChatGPT:
- Question-format queries. “What is,” “How to,” and “Why does” style keywords trigger AI Overviews heavily. If you target these, your content needs to answer the question in the first two sentences, not the first two paragraphs. AI systems extract the most direct answer they can find.
- Topical cluster depth. A Graphite research study found that sites with high topical authority gain organic traffic 57% faster than sites with low authority. AI engines pull citations from sources they’ve seen demonstrate sustained coverage of a topic, not sites with one standout article.
- Specificity and sourced facts. Vague overview articles get paraphrased and summarized. Specific, data-backed, entity-named pieces get cited directly. A keyword that lets you write something precise and well-sourced beats a broader keyword that pushes you toward generic coverage.
The strategic implication is this: for every informational keyword you target, you need a plan for being the citation, not just the ranking. If the keyword will trigger an AI Overview and your content isn’t structured to be cited, you’re investing in an asset that earns impressions and sends no one anywhere.
How to Score and Prioritize Your Keyword List
Pull up your keyword tool. Before you filter by anything, here’s the sequence that actually works.
Start by listing the keywords your buyer type searches, not the keywords that describe your product category. Talk to your sales team. Check your support tickets. Look at what questions customers ask before they buy. That’s your real seed list, not what a tool auto-suggests when you type your product name.
Then run every candidate through this table:
| Keyword | Lens 1: Traffic Potential | Lens 2: Revenue Signal | Lens 3: AI Visibility Risk | Priority |
|---|---|---|---|---|
| ”best accounting software” | High volume, very high competition | Commercial intent, broad audience fit | AI Overview likely, need topical cluster | Medium |
| ”accounting software for freelancers” | Low-mid volume, low competition | Strong: specific buyer, direct fit | AI Overview possible, great citation angle | High |
| ”what is double-entry accounting” | Mid volume, low competition | Informational, minimal buying signal | High AI Overview risk, click value low | Low |
| ”FreshBooks vs QuickBooks freelancers” | Low volume, very low competition | Extremely high: active comparison stage | AI Overview rare, direct SERP click likely | High |
Notice what happens. The highest-priority keywords aren’t the highest-volume ones. They’re the ones with the strongest buying signal and the clearest path to being useful to a reader who’s already close to a decision.
Here’s the four-step prioritization process:
- Pull 30 to 50 keyword candidates without filtering by volume. Let everything in. Weird long-tails, zero-volume queries you’ve heard from customers, comparison terms for your competitors. All of it.
- Apply Lens 2 first. Cut any keyword where you can’t clearly identify a buying trigger or a direct connection to what your product does. This step alone removes most junk from a typical keyword list.
- Apply Lens 1 to what’s left. For each remaining keyword, open the actual SERP. Check the referring domain counts on the top-ranking pages. Filter out anything where you’re fighting domain authorities ten times your own for a broad informational query.
- Tag Lens 3 risk and format accordingly. For any keyword where AI Overviews are likely to appear, flag the content brief to include a direct answer in the opening paragraph, clear H2 structure, named entities, and linked data sources throughout. That’s how you shift from “ranked but ignored” to “cited and clicked.”
One more thing that usually gets treated as a publishing problem but is really a selection problem: keyword cannibalization. Before any keyword makes your final list, search your own site for existing content targeting the same intent. Two pages competing for the same query dilute each other’s ranking signals. Decide which page owns that keyword before you write, not after you publish.
Frequently Asked Questions About Choosing SEO Keywords
How many keywords should I target per page?
Target one primary keyword and two to four semantically related secondary keywords per page. The primary keyword anchors the page’s topic and should appear in the H1, the opening paragraph, and the meta description. Secondary keywords strengthen topical depth naturally when your content covers the topic well, but forcing them in mechanically doesn’t help rankings and makes the writing worse. One page targeting twelve unrelated terms will rank strongly for none of them.
Is search volume still worth looking at in keyword research?
Yes, but treat it as a third-tier signal, not the starting point. Search volume estimates how much potential traffic a keyword carries, but it says nothing about conversion likelihood or whether AI Overviews will intercept those clicks before they reach your page. Evaluate Revenue Signal and competitive reality first. Use volume as a tiebreaker between two equally strong options, not as the primary sort column.
What’s the difference between keyword difficulty and actual ranking difficulty?
Keyword difficulty (KD) is a tool-generated score, typically on a 0-100 scale, estimating how hard it is to rank based on the backlink profiles of the pages currently sitting on page one. Actual ranking difficulty depends on your site’s topical authority, the quality of your content versus what’s already ranking, the real composition of the SERP (a government page and a Wikipedia article are different obstacles than two thin affiliate blogs), and whether AI Overviews have already reduced the click reward for ranking on that term at all. Always check the SERP manually. KD is a starting filter, not a verdict.
Should I target keywords that show zero search volume in tools?
Sometimes, yes. Ahrefs, Semrush, and Google Keyword Planner all sample from limited data pools. A keyword showing zero doesn’t mean zero searches exist. It often means the query isn’t in their dataset yet, which is common for specific product comparisons, new technology terms, and niche industry phrases. If a zero-volume keyword has a strong Revenue Signal and directly mirrors questions your customers ask during the sales process, write for it. The traffic ceiling is low, but the conversion rate can be exceptional precisely because the audience is so specific.
How often should I revisit and update my keyword strategy?
Every six months at minimum for your core topic clusters, and immediately following any major algorithm update or change in your product offering. Google AI Overviews have grown 115% since the March 2025 core update according to WordStream’s tracking, which means the click value of certain keyword types is shifting continuously. A keyword strategy built in early 2025 may already be sending you toward queries where AI Overviews now absorb most of the clicks. Static keyword lists have a shorter shelf life than they used to.
Conclusion
The old keyword selection model, sort by volume, filter by difficulty, check intent, repeat, was built for a search environment that no longer exists in the same form. AI Overviews are reshaping which keywords send clicks. Conversion rate data has been screaming for years that most informational traffic doesn’t pay. The gap between “we rank for it” and “it grows our business” is wider than ever.
The Three-Lens Filter doesn’t require new tools. It requires a different sequence: Revenue Signal first, Traffic Potential second, AI Visibility as the final check before you commit. Start there and you’ll build a keyword list worth acting on.
If you’d rather have a team run this process against your actual site and competitors, LoudScale builds keyword strategies grounded in exactly this framework for growing B2B and e-commerce brands.