Search & Content · Workflow
You do not need a $249 subscription to find out what people are searching for. You need about $30, a spreadsheet, and the discipline to read a search results page properly.
There is a particular kind of paralysis that hits people starting a website. They read that keyword research is the foundation of everything, they open a comparison article, they discover that the “recommended” tool costs more per month than their hosting costs per year, and they quietly decide to just write about whatever they feel like. Which, ironically, sometimes works better than what the people with expensive subscriptions do, because at least it produces content.
I want to make a narrower argument than “free tools are enough.” They aren’t, always. What I want to argue is that the expensive part of keyword research was never the data. It’s the judgment. And the judgment is free, though it takes a while to acquire.
What follows is the actual sequence I use when I’m researching a topic for a site that has no budget for a platform subscription. It costs somewhere between zero and about $30 a month depending on how much you already have lying around. I’ve laid it out in the order you should do it, because the order matters more than most guides admit — step one dramatically reduces the amount of work in steps two through five, and almost everyone skips it.
First, Understand What You’re Actually Measuring The part everyone rushes past
Keyword research pretends to be one activity. It’s three, and they use different data sources with wildly different reliability.
Demand is how many people search a phrase. This number comes, ultimately, from Google, and every tool that shows it is either buying it, modelling it, or scraping an approximation of it. Nobody outside Google has the real figure. Treat every volume number, from every tool, as an order-of-magnitude estimate.
Difficulty is how hard it would be for you specifically to rank. Every tool sells you a number for this, and every one of those numbers is computed almost entirely from backlink counts of the current top ten. That’s a genuinely useful proxy and a genuinely incomplete one, because it knows nothing about your site, your topical authority, or whether the current results are actually good.
Intent is what the searcher wants. No tool measures this. You read the results page and work it out. This is the step that separates people who rank from people who publish.
Notice which of the three is the hardest to buy. The paid platforms are excellent at demand, decent at difficulty, and functionally useless at intent. So the free workflow is weakest exactly where paid tools are strongest, and equally strong where paid tools are weakest. That trade is more even than the pricing pages suggest.
Step One: Mine the Data You Already Own Cost: $0 · Skipped by nearly everyone
If your site has any traffic at all — even a few hundred visits a month — Google Search Console is sitting on better keyword data than any paid tool will sell you, because it’s not an estimate. It is a record of queries where Google actually showed your pages to actual humans.
Go to Performance, set the date range to the last three months, and switch on both Average Position and Impressions. Now sort by impressions and look for queries where you rank somewhere between position 8 and position 25.
These are your striking-distance keywords, and they are the highest-return work available to you. Google has already decided your page is relevant enough to show. You aren’t asking it to reconsider the whole page from scratch; you’re asking it to nudge you up a few spots. A page that moves from position 14 to position 6 can double or triple its clicks without a single new backlink, purely from tightening the content around the query Google is already matching you to.
Two limits worth knowing. The Search Console interface caps most exports at 1,000 rows, which sounds like a lot until you have a few hundred pages. And the data window is 16 months, rolling — whatever falls off the back end is gone permanently, which is an argument for exporting a snapshot every quarter even if you don’t look at it. If you outgrow the 1,000-row ceiling, the free Looker Studio connector pulls considerably more, and Search Console’s Bulk Data Export streams the raw daily data into BigQuery for anyone comfortable writing SQL. Both are free. Neither requires a subscription to anything.
Step Two: Get Volume Numbers Cost: $0 to $7/month
Now you need demand data for terms you don’t rank for yet. Three routes, in ascending order of usefulness.
Google Keyword Planner — free, with a catch
Keyword Planner is free and comes straight from the source, but Google restricts the data for accounts that aren’t spending money on ads. Without active spend, the Avg. monthly searches column shows one of roughly seven very wide logarithmic bands — 1K–10K, 10K–100K, and so on. A keyword getting 1,200 searches and one getting 9,400 land in the same bucket. That’s not a rounding issue; it’s a nine-fold difference presented as identical.
Run a live campaign with real spend and the buckets narrow dramatically, into something like sixty finer bands. Google has never published the spend threshold, but practitioners report that even a modest campaign running for a few weeks changes what the interface shows you. There’s also a well-known partial workaround: enter your keyword in square brackets under “Get search volume and forecasts,” then open the Forecast tab and set an unrealistically high maximum CPC. The impression forecast it returns is a much tighter estimate of monthly demand than the bucket. It’s fiddly and it only works one keyword at a time, but it’s free and it’s Google’s own number.
One thing Keyword Planner is genuinely good at, and which SEOs undervalue: the Competition and Top of page bid columns. Those measure how hard advertisers are fighting for a term. High commercial competition is a reliable signal that a keyword converts, which matters far more than volume if you’re monetising with ads or affiliate links.
Bing Webmaster Tools — free, and quietly excellent
The keyword research module inside Bing Webmaster Tools gives real numbers rather than buckets, has no daily lookup ceiling, and shows a rolling history. Bing’s search share is small, so treat the absolute numbers as a fraction rather than a total — but the relative ranking between keywords is broadly consistent with Google, which is what you actually need when deciding between two topics. It also matters more than it used to, since Bing’s index feeds several AI search products.
Keywords Everywhere — the cheapest real data
Keywords Everywhere is a browser extension that overlays volume, CPC and competition onto whatever page you’re already looking at — Google, YouTube, Amazon, Bing, Etsy. Pricing is annual-only and credit-based: Bronze is $84 a year for 100,000 credits and one seat, Silver $168 for 400,000 credits and three seats, Gold $480 for two million, Platinum $1,440 for eight million. There is no monthly billing, and unused credits expire a year after purchase.
One credit buys the data for one keyword. That sounds generous at 100,000, and it is, provided you turn the extension off when you’re not researching. Left running, a single Google search that renders related and long-tail suggestions can consume fifty credits in one page load. I burned through a third of a year’s Bronze allowance in about three weeks the first time I used it, purely because I forgot it was on while browsing normally. Toggle it. Seriously.
At roughly $7 a month, Bronze is the single best value purchase in this entire workflow, and the fact that it’s a browser overlay rather than a dashboard means the research habit forms much more naturally than it does with a platform you have to remember to open.
Step Three: Expand the Question Space Cost: $0 to $12/month
Volume tools give you keywords. What you actually want is the shape of a topic — the questions people ask around it, in the order they ask them.
Google autocomplete and People Also Ask remain the purest free source, because they’re live Google behaviour rather than a model of it. Type your seed term and cycle the alphabet after it. Tedious, effective. Expand every PAA box; each expansion loads new ones, and the tree goes deeper than people expect.
AlsoAsked automates exactly that, mapping People Also Ask data into a nested tree up to four levels deep. The free tier gives three searches a day with no signup at all, which is genuinely enough for a small site. Paid plans run $12/month for 100 credits, $23 for 300, and $47 for 1,000, with one credit per search. The hierarchical view is the point — it shows you which question is the parent and which are follow-ups, and that structure translates directly into an article outline.
AnswerThePublic covers similar ground from autocomplete rather than PAA, and offers a free tier of three searches a day. Its paid pricing has been restructured more than once since NP Digital acquired it, and the tiers I see quoted around the web disagree with each other, so check the live page rather than trusting any article — including this one — on the number.
Google Trends does something none of the above do: it tells you direction. Absolute volume is useless if a topic has been declining for three years. Trends is relative-only, which frustrates people, but relative is exactly the right unit for the question “is this topic growing or dying?”
Reddit, forums, and the comment sections of the pages already ranking are where the phrasing lives. Search tools give you the terms people type into a box. Communities give you the way people describe the problem before they know the term. Some of the highest-converting content I’ve written came from a single frustrated forum post that used no recognised keyword at all.
“Every tool sells you a difficulty score. Not one of them has seen your site, your topic history, or whether the current top ten is any good.”
Step Four: Judge Difficulty By Reading, Not By Score Cost: $0 · The actual skill
This is the step where the free workflow stops being a compromise and starts being an advantage, because it forces you to do the thing that people with expensive tools skip.
Open an incognito window. Search the keyword. Then read the first page properly, asking five questions.
Who is ranking? If the top ten is Wikipedia, Healthline, Forbes, and three government domains, walk away. If it’s a mix of mid-sized sites, forums, a YouTube video and one page that clearly hasn’t been updated since 2019, you have an opening. The presence of Reddit or Quora high in the results is one of the most reliable green flags in SEO — it means Google couldn’t find a good dedicated page and settled for a discussion thread.
Is the intent consistent? If all ten results are the same format, the intent is settled and you must match it. If results are split between product pages, tutorials and definitions, Google is uncertain — and uncertainty is opportunity, because a page that resolves the ambiguity cleanly can take the whole set.
How good is the worst result on page one? Not the best. The worst. You aren’t competing with the leader initially; you’re competing for the eighth slot. If position eight is a thin 600-word page with a stock photo, you can beat it this month.
What’s above the organic results? Four ads, a shopping carousel and an AI Overview means position one might get a fraction of the clicks you’d expect. This doesn’t disqualify a keyword, but it changes the maths considerably.
Does the page answer a question that ends the search? Or does answering it create the next question? Keywords that generate follow-up searches are worth far more than their volume implies, because they let you build an internal linking structure that keeps the visitor on your site.
You can approximate the link-strength half of difficulty for free, too. Ahrefs Webmaster Tools gives verified owners real backlink data on their own properties, and Majestic’s public Site Explorer will show Trust Flow and Citation Flow for any domain at limited depth without a subscription. That’s enough to sanity-check whether a competitor is genuinely strong or just old.
Step Five: Cluster Before You Write Cost: $0 · One spreadsheet
The single most common mistake made by people doing free keyword research is treating every keyword as a separate article. You end up with forty thin pages competing with each other for variations of the same query, and Google picks one more or less at random.
The paid platforms sell clustering as a feature. The manual version takes twenty minutes and works fine: search two keywords, compare the top ten results for each, and if five or more URLs appear in both, they’re the same page. Google has already told you it considers them one topic. Merge them.
Build a sheet with columns for keyword, estimated volume, the parent cluster it belongs to, the intent you read off the SERP, and a difficulty note in plain English rather than a score. Then add a column called “why would anyone link to this,” and be honest. Rows where the answer is “they wouldn’t” are still worth writing — not everything needs links — but they need to be supported by pages that do.
Prioritise by a rough product of achievability and commercial value, not by volume. A 200-searches-a-month keyword with a $12 advertiser bid and a weak page one is worth more than a 20,000-searches-a-month keyword owned by Wikipedia. This is obvious written down and almost universally ignored in practice, because big numbers are seductive.
The Stack, Priced Out
| Tool | Cost | What it does | Honest limit |
|---|---|---|---|
| Google Search Console | Free | Real queries, positions, CTR for your own pages | 1,000-row UI export; 16-month window; your site only |
| Keyword Planner | Free | Google’s own demand data, advertiser bids | Wide volume buckets without live ad spend |
| Bing Webmaster Tools | Free | Uncapped keyword lookups, real numbers | Bing-scale volumes; use ratios, not absolutes |
| Google Trends | Free | Direction of travel, seasonality | Relative index only, never absolute volume |
| AlsoAsked | Free (3/day) or $12–$47/mo | Nested People Also Ask trees | Credit-metered; one credit per search |
| Keywords Everywhere | $84/yr (Bronze) | Inline volume, CPC, competition everywhere you browse | Annual billing only; credits drain fast if left on |
| Manual SERP reading | Free | Intent, real difficulty, content gaps | Doesn’t scale past a few dozen keywords a week |
Total for the full stack: $84 a year, plus $144 a year if you want AlsoAsked Basic. Call it $19 a month. That is roughly one seventh of the cheapest “professional” platform plan, and for a site publishing four to eight articles a month it will not meaningfully hold you back.
Where This Workflow Genuinely Breaks Down
I’d rather tell you the limits than have you discover them in month four.
It doesn’t scale. Manual SERP reading is maybe thirty keywords an hour if you’re quick. At agency volumes, that’s not a workflow, it’s a hostage situation. The moment you’re researching for multiple clients, the platform subscription pays for itself in hours saved during the first week.
Competitor keyword gaps are the real hole. Nothing free tells you the full list of terms a competitor ranks for that you don’t. That single report is, more than anything else, what the paid tools are actually selling, and there’s no honest substitute. Ahrefs’ $29 Starter plan exists precisely to sell you a rationed taste of it.
Rank tracking is separate. Search Console shows average position across everything, which smears local, mobile and personalised results together into a number that moves for reasons you can’t diagnose. If you need to know your position for twenty specific terms on a specific device in a specific city, you need a tracker.
Non-English and small niches suffer most. Free tools thin out fast outside major English markets. If you work in Korean, Portuguese or Polish, the gap between free and paid data widens considerably.
The Order of Operations, Compressed
If you take one thing from this, take the sequence. Export Search Console and fix what’s already ranking between positions 8 and 25. Only then look for new keywords. Get rough volume from Keyword Planner or Bing, or pay the $84 for real numbers inline. Expand the topic with autocomplete, PAA and AlsoAsked. Read the results page for every keyword you’re serious about, and judge difficulty from what you see rather than from a number. Cluster by SERP overlap before you write anything. Prioritise by commercial intent and achievability, never by volume alone.
That’s the whole method. The tooling is a rounding error in it. I’ve watched people with $500 monthly stacks produce worse keyword strategy than a careful beginner with a spreadsheet, because the expensive tool answered the question “what could I write about” and never once answered “should I.”
Buy the platform when the constraint you’re hitting is genuinely time. Until then, the constraint is judgment, and judgment doesn’t come with a monthly plan.
Tool limits and prices verified August 2026 against vendors’ own pricing and documentation pages; SEO tool pricing has been volatile this year, so confirm before subscribing. This article contains no affiliate links — every link points to the vendor’s own site or a free Google/Microsoft property.