Playbook 2: Keyword research that survives the results page
A keyword list checked against every results page, split into terms to commit to and terms to work toward
- For
- Anyone choosing keywords for a service website
- You get
- Seven steps, shown on a real project's data
A keyword list is only trustworthy after someone has opened the Google results page for every term on it. That check reshapes the list, and it decides which terms we are willing to put in a contract. The examples below come from one project: a letting agency in Amsterdam that places homes for owners and finds homes for expat tenants.
1. Test the client's own terms first
Clients arrive with the phrases they would type themselves. This agency gave us seven. Five had no measurable search volume. The intent behind them was right, but buyers phrase the same need differently, and mostly in Dutch.
Client's term | Monthly searches | What we target instead |
|---|---|---|
how to rent out my house in amsterdam | 0 | huis verhuren amsterdam (210), the same question asked in Dutch |
expat rental agency amsterdam | 10 | expat housing amsterdam (320) |
letting agent amsterdam | 0 | Dropped: British English the Dutch market does not use |
trusted estate agent in amsterdam | 0 | Dropped: trust is built on the page, nobody searches for it |
amsterdam rental agency | 390 | Kept as proposed |
2. Separate people looking for listings from people looking for you
The biggest numbers in a market usually belong to someone else. Here, huurwoning amsterdam (5,400 a month) and apartment for rent amsterdam (4,400) are people looking for listings, and the property portals own those results. The smaller terms, amsterdam rental agency (390) and housing agency amsterdam (140), are people looking for an agency. We target the second group and say plainly in the strategy report why the big numbers are left out.
3. Check where the searches come from before choosing a language
The proposal assumed an English-first site. We measured the same terms per searcher country before building anything.
Listing searches come from abroad and go to the portals. Agency searches come from people already in the Netherlands. Owner-side demand was about 4,500 searches a month in Dutch against about 50 in English. The decision followed: Dutch at the root of the domain carrying the owner keywords, a complete English version under /en/ for tenants.
4. Open every results page and give each term a verdict
For each candidate term we look at who actually ranks (portals, directories, competitors with a dedicated page, or homepages that only mention the term), whether a map pack sits above the organic results, and which positions are weak enough to take. Nine of the ten terms we checked for this agency showed a map pack. On those terms the Google Business Profile competes for the click before the website does, so profile work goes into the project scope.
Keyword | Monthly searches | What the results page showed | Verdict |
|---|---|---|---|
huis verhuren amsterdam | 210 | City council, an info site, thin agency pages at 8 to 10 | Commit |
woning verhuren amsterdam | 110 | Near-identical results, no map pack | Commit |
verhuurmakelaar amsterdam | 720 | Ten agencies, six of them in the map pack | Commit, paired with the business profile |
amsterdam rental agency | 390 | Reddit at 1, an exact-match domain at 2 | Stretch |
expat housing amsterdam | 320 | Two brand-name domains own the phrase | Stretch |
vastgoedbeheer amsterdam | 110 | Ten dedicated property managers | Page stays, no commitment |
A verdict of Commit means positions 8 to 10 are held by pages we can beat with one dedicated page. A Stretch is worth building for but depends on factors a page alone does not control.
5. When nobody searches, write for the answer
Some valuable phrases have no search volume at all. For the leadership coach from Playbook 1, problem language such as conflict in a management team measured zero searches in Belgium and the Netherlands. People still ask an AI assistant exactly that question, and they still forward a good article to a colleague. Those articles are written to be quoted in an answer. The one term in her field with real volume, toxic leadership in Dutch at 720 a month, became the first article, with a workplace self-assessment attached as the lead magnet.
6. Split the list into what we commit to and what we work toward
Every target term lands in one of two groups. Group A holds the terms we judge realistic for the Google top ten within six months. Group B holds the longer-horizon terms. Brand terms are excluded from both. The client receives this split as a short appendix, confirms it by email, and it becomes part of the agreement: we target a top-ten position for a clear majority of Group A within six months of go-live. If that is not reached, we keep doing the on-page, content and technical work at no extra monthly cost until it is. The commitment assumes agreed pages get published and inputs arrive on time, and it excludes major algorithm changes.
The client sees the filtered appendix. The working sheet, with difficulty judgements, competitor notes and every rejected term, stays internal, because once it is handed over every row becomes negotiable.
7. Calibrate the next project with this one
At the end of each term we copy the Group A rows and their actual outcome into one benchmark sheet across all projects. After a handful of projects the line between A and B stops being a judgement call and starts reflecting what we have actually achieved.
Do you have a project in mind?
Let's talk about how to apply this to your own site.
