How to Use Google Autocomplete for Keywords: A Step-by-Step Guide

How to use Google autocomplete for keywords: the method nobody bothers to explain

Type "how to clean" into Google and watch what happens. Before you finish the third word, the dropdown has already offered you a dozen endings: how to clean a washing machine, how to clean silver, how to clean grout, how to clean a burnt pan. You didn't ask for any of those. Google volunteered them, because thousands of people typed them before you.

That dropdown is a keyword tool. A free one. No login, no API key, no dashboard. And most people, including me for the first two years I worked in SEO, treat it as an afterthought — a small convenience feature rather than a research instrument.

Which is a mistake. When I started building content for a niche e-commerce site in 2023, Google autocomplete was the only tool in my stack that gave me actual queries typed by actual humans, not modelled estimates. I built two content clusters off it, and one of them still pulls around 40% of that site's organic traffic. Nothing fancy. Just a browser and a spreadsheet.

Here's how to use it properly.

How does Google autocomplete work?

Google's suggestions come from a blend of sources: what other users have searched, how often they search it, trending topics, and — crucially — your own context. Language, country, even the device you're on all nudge what appears.

Three filters matter for anyone doing keyword research:

• Popularity — the suggestion has to be a query real people type, at real volume. Nonsense strings never appear.
• Freshness — trending queries can surface before they show up in any paid tool. I once spotted a seasonal term in the dropdown three weeks before it appeared in Keyword Planner.
• Policy filtering — Google strips out violent, hateful, sexually explicit and some privacy-sensitive completions. So if your niche touches any of those areas, your suggestion list will look thin. That's not a data problem, it's a filter.

What is the difference between predictive search and autocomplete?

Honestly, most people use the two interchangeably, and in everyday speech that's fine. But there is a real distinction, and it matters when you're reading documentation or buying a tool.

Autocomplete is the mechanism: as you type, the system completes what it thinks you're typing. It completes the string. That's what the browser's URL bar does when it finishes a domain you've visited before.

Predictive search is the broader behaviour: the system anticipates what you're trying to achieve, not just which characters you're entering. It might offer a product directly, a category page, a question, or a correction. Google's search box does both at once, which is why the line gets blurry.

The practical upshot for keyword work: autocomplete gives you query fragments. Predictive behaviour gives you intent signals — and if a suggestion comes with a question mark, a "how" or a "vs", it's telling you something about what the searcher actually wants.

The alphabet method, step by step

This is the core technique. It's tedious. It works.

1. Open a private or incognito window. This strips out most of your personalisation — you want the generic version, not the version shaped by your own browsing.
2. Type your seed keyword, then a space, then the letter a. Note every suggestion.
3. Repeat for b through z. Yes, all of them.
4. Now do it again with prepositions: for, with, without, near, vs, like.
5. And once more with question words: how, why, what, when, can, should.

For a single seed keyword, that's roughly 40 passes. Twenty minutes if you're efficient, an hour if you keep getting distracted by the suggestions themselves (I do, every time).

The output is a raw list of 200 to 400 real queries. Not all of them are good. Some are junk. But somewhere in there you'll find phrasings you would never have invented — "how to clean a burnt pan without baking soda" being one I'd never have guessed, and which became a page that ranked in the top three for its term within four months.

Using it on Android and mobile

Mobile matters more than desktop for a lot of local and transactional queries, and the suggestions differ. On Android, the Google app and the search widget both pull suggestions, but they lean harder on your location and recent activity.

Two adjustments I make on mobile:

• Use the Google app in a fresh account or signed out, so your location bias is at least predictable.
• Test near me and city-name modifiers directly. The mobile dropdown surfaces local variants aggressively — "plumber near me", "plumber [city]" — that desktop sometimes buries.

That said, if you're doing bulk research, mobile is a worse interface. The suggestions truncate, and you can't easily copy the full list. Desktop with a spreadsheet open is still the right setup.

Turning autocomplete on and off

You'd be surprised how often this comes up. People who want clean, unpersonalised results need to know how to switch it off. People who had it switched off by accident want it back.

How do I turn on Google autocomplete?

Autocomplete is on by default — there's nothing to enable. If suggestions have stopped appearing, the cause is almost always one of these four things:

• Browser settings. In Chrome, go to Settings → You and Google → Sync and Google services, and check that "Autocomplete searches and URLs" is enabled. In Firefox, it's under Search → "Provide search suggestions".
• SafeSearch or a family filter. A strict filter can suppress completions.
• A managed device or account. School and work profiles often disable suggestions by policy, and you can't override that from your side.
• An extension. Ad blockers and privacy tools sometimes kill the dropdown as collateral damage.

One thing worth knowing: the suggestions are generated by Google's servers, not by your browser. So if a different browser on the same network shows suggestions and yours doesn't, the problem is local — your settings, your extensions, your profile.

Turning it off

The reverse is straightforward. In Chrome: Settings → You and Google → Sync and Google services → turn off "Autocomplete searches and URLs". In your Google Account, you can also clear your web and app activity, which reduces personalisation even if you leave suggestions on.

For research purposes, I'd argue you shouldn't turn it off permanently. Instead, use incognito windows — you get the clean data without losing the feature for daily browsing.

Autocomplete vs Keyword Planner: which do you actually need?

Here's where I'll take a side. If you have to pick one, pick autocomplete — because it gives you queries, and Keyword Planner gives you estimates. But you shouldn't have to pick one, and the two answer different questions.

	Google Autocomplete	Google Keyword Planner
Cost	Free, no account needed	Free with a Google Ads account
Data type	Actual query strings	Modelled volume ranges
Volume figures	None	Yes, but often as broad ranges
Competition data	None	Yes (with caveats for low-volume terms)
Best for	Discovering phrasings and intent	Prioritising and sizing
Weakness	No numbers, personalised by default	Values are bucketed, so precision is illusory

Is Google Keyword Planner still free?

Yes. You need a Google Ads account, but you don't need to run a campaign or spend anything — you can access the planning tools with an account that has no active ads. Google has tightened the volume ranges over the years, so low-volume keywords often show as broad buckets rather than precise figures. Free, still useful, less precise than it once was.

What Keyword Planner can't do is tell you how people phrase things. Autocomplete can. That's why my workflow runs both: autocomplete to generate the candidate list, Keyword Planner to sort it by likely value.

Where this method breaks down

I'll be honest about the limits, because I've wasted time ignoring them.

Autocomplete tells you nothing about volume. A suggestion appearing in the dropdown means people search it — it doesn't mean many people do. I once built a full content piece around a suggestion that felt perfect, published it, and watched it collect 12 sessions in three months. The phrasing was real. The demand wasn't.

Personalisation is the other trap. Even in incognito, Google uses your IP and language settings. Run the same alphabet sweep from two different countries and you'll get two different lists. If your audience is somewhere you aren't, you need a location override — or a VPN, though honestly the cleaner route is a rank-tracking tool with a location parameter.

And the suggestions are cyclical. They shift with news, seasons, and whatever's trending that week. A list you built in January may look noticeably different by June. If you're working on evergreen content, that's fine. If you're chasing trends, you need to re-run the sweep more often than you'd like.

Turning the sweep into a habit

The reason most people don't get value out of autocomplete is that they do it once, badly, and conclude there's nothing there. The value is in the repetition — the same seed keyword run across letters, prepositions and question words, on a schedule, logged somewhere you'll actually look at again.

My own setup is unglamorous: a spreadsheet with one tab per seed keyword, a column for the raw suggestion, a column for the modifier that produced it, and a rough intent label. It takes twenty minutes a week. Over a year, that's a content backlog most teams would pay a subscription for.

The dropdown looks like a convenience. It's an inventory. And it's sitting there, free, every time you type.