Keyword Research Services: What People Type, and What They Mean By It
Demand, intent and difficulty, priced honestly. A keyword you cannot win is not a target, it is a wish.
Researched from Ho Chi Minh City, in English, Vietnamese and whichever other languages your markets search in.
THE SERVICE
What keyword research actually is
Keyword research is the process of finding the terms and questions real buyers use, working out what they mean by them, and deciding which ones you can realistically win. Done properly it produces a prioritised map of terms assigned to specific pages, not a spreadsheet export sorted by search volume. Our service covers four things: mining your existing Search Console data for near-miss rankings, classifying every term by intent, pricing difficulty against the authority your domain actually has, and mapping the result to existing or new URLs. The output is a plan with owners and priorities attached, delivered in five to ten working days.
Volume is a vanity metric. Intent is the business.
Most keyword research is a data export with the rows sorted by search volume. That produces a list of terms your competitors already own and your customers were never going to convert on.
We start from the buying decision instead. What does someone type at the point they have a problem, what do they type when they are comparing options, and what do they type when they are ready to spend? Those three groups behave completely differently and deserve completely different pages.
Then we price it. A term is only a target if you can realistically win it with the authority you have, in a timescale you can afford to wait. We would rather give you fifteen winnable terms than four hundred aspirational ones.
OUR PROCESS
How we actually do it
Data first, opinions second.
Start from your own data
Search Console is the most honest keyword tool you own: it shows what you already rank for, on page two, with impressions and no clicks. Those near-misses are usually the fastest wins available.
Map demand to the buying journey
Every term gets classified by intent, informational, commercial, transactional, navigational, because the intent decides the page type, the content and whether it belongs in the blog or on a service page.
Price the difficulty against your authority
Difficulty scores are relative. We assess what is actually ranking, how strong those pages are and whether your domain can compete, then split the list into now, next and later.
Hand it over as a plan, not a spreadsheet
The output is a prioritised roadmap tied to pages and owners, which becomes the input for the topical map and the content programme.
Check nothing is competing with itself
Before a single term is assigned, we check whether you already have two or three pages chasing it. Keyword cannibalisation is one of the most common problems we find in a new account and one of the fastest to fix, because the traffic already exists, it is just split across pages that undercut each other.
What you get
Deliverables, not adjectives.
You get it as a working spreadsheet and a short written summary. The spreadsheet is yours to keep and to hand to another agency if you ever want to. We do not build the roadmap into a tool you have to keep paying us for.
Ask for a sample deliverableOUR SOURCES
The tools and data behind it
No single tool is right, and anyone who tells you otherwise is describing their subscription rather than their method. We cross-reference several sources and treat all of them as estimates.
Your own impression and position data. The only source that is actually about your site rather than a model of the market.
For volume, difficulty and competitor visibility.
What is actually ranking, and how strong it is. A difficulty score is a guess until someone looks at the page holding position three.
The comparisons and questions people put to ChatGPT, Perplexity and Gemini, which rarely look like a Google query.
The single most underused keyword source in any business. The words prospects use on calls are the words they use in search, and they are usually nowhere in the tool exports.
MULTILINGUAL
Keyword research in more than one language
Translated keywords are not researched keywords. Running an English list through translation gives you terms that are grammatically correct and commercially useless, because people in different markets do not describe the same problem the same way, and often do not use the same word for the product at all.
We research each market natively. That means separate demand analysis, separate intent classification and separate difficulty pricing per language, because the competitive landscape in Vietnamese bears no relation to the one in English. For La Petite Ecole this was the whole job: winning in two languages and two markets at the same time, where the parents searching in Vietnamese and the parents searching in French wanted different things from the same school.
Multi-market research takes longer than single-market, usually three to four weeks rather than one to two. Anyone quoting you the same timeline for both is not doing the work twice.
Who this works for
This is usually the right starting point in four situations.
You are about to invest in content and want to know where to point it.
Cheaper to research first than to publish forty pages against demand that was never there.
You inherited a keyword list and do not trust it.
Most inherited lists are volume-sorted and intent-blind. We will start from yours and tell you which parts survive.
You are entering a new market or a new language.
This is where guessing is most expensive and where translated lists do the most damage.
Traffic is flat and you cannot see why.
Often the terms are fine and the problem is that three of your own pages are fighting over them.
We do this most often for international schools, B2B and professional services firms, SaaS companies and considered-purchase e-commerce, for clients headquartered in the US, UK, Australia, Canada and the EU as well as businesses operating inside Vietnam.
Questions, answered straight
No fluff, as promised.
How many keywords should we target?
Fewer than you think. A focused set of winnable terms mapped to genuinely useful pages beats a list of hundreds you never build for. We would rather ship fifteen pages that rank than plan two hundred that do not.
Do you research keywords for AI search as well as Google?
Yes, and they differ. AI engines answer questions rather than match strings, so we research the questions and comparisons people put to them. That work sits in AI visibility and GEO.
We already have a keyword list. Can you just use it?
We will start from it, but expect us to challenge it. Most inherited lists are volume-sorted and intent-blind, and rebuilding the intent layer is usually where the wins are hiding.
How long does keyword research take?
Five to ten working days for a single market, once we have access to your Search Console and analytics. Three to four weeks if the work spans multiple markets or languages, because each one has to be researched natively rather than translated.
What does the deliverable actually look like?
A working spreadsheet with every term classified by intent, difficulty and priority and mapped to a specific URL, plus a short written summary for people who will never open the spreadsheet. It is yours to keep, including if you later take the work elsewhere.
Which keyword tools do you use?
Google Search Console for your own data, the paid keyword tools for volume and difficulty, and manual SERP analysis on every priority term, because a difficulty score is a guess until someone looks at the page currently holding position three. We treat all tool volumes as estimates.
How do you classify search intent?
Every term is tagged informational, commercial, transactional or navigational, based on what is actually ranking for it rather than on the wording alone. The intent decides the page type, which decides whether the term belongs on a service page, a comparison page or the blog.
Can you research keywords in Vietnamese or other languages?
Yes. Each language is researched natively, with its own demand analysis, intent classification and difficulty pricing. Translating an English keyword list produces terms that are grammatically correct and commercially useless, because markets rarely describe the same problem the same way.
What happens after the research is delivered?
It becomes the input for a topical map and a content plan. You can run that in-house, or we can build and publish against it. The research stands on its own either way, and there is no obligation to continue with us.
Related services
This never works in isolation.
Proof from real work
All case studies
Want to know what you can actually win?
Tell us what you're working on. You'll get an honest answer about whether we can help, and exactly how.