How to Do Keyword Research for a Kenyan Business

Keyword research goes wrong when you begin with a tool, export hundreds of phrases and call the spreadsheet a strategy.

A useful keyword map answers harder questions.

• Which searches describe something your business genuinely offers?

• What does the searcher want to do?

• Which page should answer?

• Does that page already exist?

• And can you create something useful enough to deserve the visit?

For a Kenyan business, geography adds another layer. A person may search with “Kenya,” a city, a county, an estate, “near me,” or no location at all.

Those variations do not automatically need separate pages.

This guide shows you how to move from one offer to an organised page plan. You can complete the process with customer information and free Google sources.

If your site already receives impressions, Google Search Console and Glarify can add first-party evidence from your own search performance.

If you need the broader foundation first, read What Is SEO?. For a wider list of tactics, use Amplify SEO's 15 SEO tips for Kenyan businesses.

This article stays focused on the keyword decision itself.

What useful keyword research produces

The final deliverable should not be a list of phrases ordered by search volume. It should be a map with five connected fields:

That structure prevents two common failures: publishing content that cannot lead to the business and creating several pages that compete for the same search.

1) Define the offer before you collect keywords

Choose one offer. Write down what the customer receives, who needs it, where the business provides it and what the visitor should do next.

For example, “cleaning” remains too broad. A more useful starting point would be:

Recurring office cleaning for organisations in Nairobi, with a site visit and quotation as the next action.

This sentence gives the research boundaries:

• Offer: recurring office cleaning

• Audience: organisations with offices

• Market: Nairobi

• Decision: whether to request a site visit or quote

• Page requirement: enough service, location, process and trust information to support that decision

Repeat this exercise for each major offer. Do not combine office cleaning, post-construction cleaning, home cleaning and pest control simply because one company happens to sell all four. Customers may describe and evaluate each service differently.

Your revenue also matters, but do not confuse revenue with relevance. A profitable service cannot turn an unrelated high-volume phrase into a good target.

2) Build seed phrases from real customer language

A seed phrase starts the research. It does not settle the final keyword.

Collect language from places where customers already explain their needs:

• sales calls, enquiry forms and WhatsApp questions;

• product or service names used by staff and customers;

• quotations and proposal headings;

• website navigation and internal site search;

• Google Business Profile questions and reviews;

• competitor navigation and service pages;

• Search Console queries for an established site; and

• industry documents, regulations or marketplaces where the terminology differs from everyday speech.

Keep the customer's phrase even when it differs from your internal label. A software company may sell a “human capital management platform” while buyers search for “payroll software Kenya.”

A property business may call a unit a “residential development,” while customers search for “apartments for sale in Kilimani.”

Google's SEO Starter Guide notes that beginners and knowledgeable users may use different query language. Your research should capture both when both groups matter to the business.

Start with nouns and problems rather than decorating every phrase with “best” or “affordable.” For the hypothetical cleaning company, the initial seeds might include:

• office cleaning;

• commercial cleaning;

• corporate cleaners;

• contract cleaning;

• workplace cleaning;

• office deep cleaning; and

• office cleaning quote.

You will test and refine these phrases later. At this stage, you want language coverage, not a verdict.

3) Add Kenyan context only where it changes the need

Localising a keyword does not mean attaching “Kenya” to every phrase. Add geography and local buying language when they affect who can fulfil the search or what answer the person expects.

Geography

Test relevant variations at the level where customers choose providers:

• country: Kenya;

• city or town: Nairobi, Mombasa, Nakuru or another served market;

• county: Kiambu, Kajiado or another genuine service area;

• neighbourhood or estate: Westlands, Kilimani, Upper Hill or another area with distinct demand; and

• proximity: “near me” or searches that trigger a map-based result.

Only research locations the business genuinely serves. A page for “office cleaning in Mombasa” needs a credible service operation in Mombasa, not a changed place name and copied Nairobi text.

Buying language

Add modifiers that reflect real decision stages:

• company, provider, agency or consultant;

• services;

• price, cost or charges;

• quote or quotation;

• buy, book, hire or order;

• supplier, installer or contractor;

• compare, best or alternatives; and

• near me.

Treat “affordable” and “best” cautiously. A searcher may use them, but the page must explain price or selection criteria honestly. Repeating the adjective does not prove the claim.

Local terminology

Test the words your Kenyan customers use for the product, payment method, location or institution. Do not assume that American or British wording always matches local speech. When English and Kiswahili variants both matter, record them separately first and then inspect whether they lead to the same need and page type.

4) Expand and compare the list with Google sources

No single source sees the whole market. Use each source for the question it can answer.

Use Keyword Planner for discovery, not an SEO forecast

Google Keyword Planner can generate ideas from seed phrases, a website or both. It also provides estimates of monthly searches and advertising cost.

Enter several closely related seeds rather than one broad word. Set the relevant location and language, then remove ideas the business cannot serve. Google currently says users must complete account setup, including billing information, to access basic keyword-discovery features, so confirm the current access requirements before you depend on it.

Keyword Planner supports Google Ads decisions. Its ranges and forecasts do not tell you whether an organic page will rank or convert.

Use Google Trends for comparisons and timing

Google Trends can compare terms or broader topics across locations and periods. Use it to test questions such as:

• Does one common wording show stronger relative interest than another?

• Does interest rise during a predictable season?

• Do regions differ enough to change the content or landing-page plan?

• Does a sudden spike look temporary rather than durable?

Google explains that Trends uses sampled, aggregated and normalised data. A score of 100 marks the highest relative interest in the selected comparison; it does not mean 100 searches. Google also cautions that Trends represents one data point rather than a scientific poll or complete measure of popularity.

Long local phrases may show little or no Trends data. Do not reject a commercially precise query solely because Trends cannot display it.

Inspect the result page before you assign a URL

Search the candidate phrase from the market you plan to serve. Record what the results suggest:

• Are the leading results service pages, product categories, articles, directories, videos or maps?

• Does Google emphasise prices, products, local providers, definitions or comparisons?

• Do several different phrases produce substantially the same results?

• Does the search require current local information?

• Can your business offer something accurate and useful that the existing results miss?

The result page does not command you to copy its structure. It helps you understand the task that successful pages appear to solve.

5) Use Search Console and Glarify for first-party evidence

Market-discovery tools estimate what people may search. Search Console records how Google Search already showed your site.

Google defines queries in the Search Console Performance report as search terms that led to the site. You can compare queries with their pages, clicks, impressions, click-through rate, country, device and time period.

Look for four patterns:

1. Relevant queries with impressions but few clicks. Check whether the page title, description and content match the need before assuming the snippet alone caused the problem.

2. Queries where one page appears for several useful variants. These may belong to one cluster rather than separate pages.

3. One query showing several pages. Inspect whether the pages serve different intents or compete for the same one.

4. Unexpected queries. They may reveal customer language, a useful adjacent need or an irrelevant association that the page should correct.

Search Console does not show every query. Google omits anonymised queries to protect privacy and limits the rows it stores and displays. Treat the report as strong evidence from your own site, not a complete list of Kenyan searches.

Where Glarify fits

Glarify can reduce the manual work of moving between query, page and planning views. Its public content-planning documentation describes a workflow that can surface:

• keywords that generated visibility but lack a dedicated page;

• high-impression, low-click queries;

• current pages associated with a keyword;

• potential content gaps;

• search-intent and content-type suggestions;

• underperforming pages that may need updates;

• possible keyword cannibalisation; and

• post-publication indexing, traffic, engagement and conversion signals.

Its content-planning feature page also advertises keyword-driven planning, volume and difficulty insights, competitor analysis and a content calendar.

Use those outputs as prompts for investigation. Verify the query, page, intent and business value before accepting a suggested article or score. The tool cannot know every operational constraint, margin, service boundary or editorial strength inside your business.

Glarify becomes most useful after a site has enough performance history to analyse. A new site still needs customer language, market discovery and live result review because it has little first-party query data.

6) Identify the searcher's task and choose a page type

Traditional intent labels help, but the actual task matters more than the label.

These examples remain hypotheses until you inspect current results and customer behaviour.

Do not force an informational article to target a provider query simply because the phrase has volume. Do not stretch a service page into a 3,000-word tutorial when the searcher wants scope, evidence, location, price factors and a quotation route.

7) Cluster phrases by the need one page can satisfy

A cluster groups different searches that one strong page can answer without becoming confused.

For the cleaning example, these phrases may belong together:

• office cleaning services Nairobi;

• commercial office cleaning Nairobi;

• corporate cleaners Nairobi; and

• office cleaning company Nairobi.

They vary in wording but appear to support the same provider-selection task. One service page can use the language naturally while explaining the offer.

Do not automatically combine “office cleaning prices Nairobi” with that group. If people expect a detailed price breakdown, cost factors, sample scope and procurement guidance, the topic may deserve a separate guide that links to the service page. If the service page can answer the pricing need fully, keep it there.

Use three checks before splitting a cluster:

1. Does the searcher want a materially different outcome?

2. Do current results favour a different page type?

3. Can the existing page answer the need without distracting from its main job?

Minor plural changes, word order and close synonyms rarely justify separate pages. Different services, locations, audiences or decision stages sometimes do.

8) Assign one page owner to every cluster

For each cluster, choose one action:

• Keep: an existing page already satisfies the need.

• Improve: the right page exists but lacks evidence, clarity, depth or alignment.

• Merge: several weak pages compete for one need.

• Create: no suitable page owns the intent.

• Leave: the phrase lacks business fit, evidence or a useful page opportunity.

Record the canonical URL even before publication. This small step exposes duplicate ideas early.

A working keyword map might use these columns:

The example is hypothetical. It illustrates the fields and decisions, not documented search demand.

9) Prioritise with five questions

Keyword volume can help compare opportunities, but it should not make the decision alone. Use this five-question filter:

1. Does the cluster match a real offer or customer problem?

2. Can you identify the searcher's task and the right page type?

3. Do you have credible evidence that customers use this language?

4. Can the business create a page that deserves the visit?

5. Can one page own the cluster without competing with another page?

Prioritise clusters that receive a strong answer across all five questions. Delay clusters that require unavailable expertise, unsupported locations, missing product data or a second page for an intent you already cover.

This filter deliberately avoids a universal score. Ten searches for a high-value specialist service can matter more than 1,000 visits from students, job seekers or people outside the service area. You still need real conversion data to learn the business value.

10) Work through one offer from seed to page map

Consider the hypothetical office-cleaning business again.

Input

• Offer: recurring office cleaning

• Market: Nairobi

• Audience: office managers, operations teams and business owners

• Conversion: request a site visit or quotation

• Customer language: office cleaners, commercial cleaning, cleaning contract, cost per month

• Existing assets: homepage, broad services page and Google Business Profile

Research operation

1. Collect seed phrases from enquiries, quotations, staff language and the current site.

2. Expand them with Keyword Planner and live search results.

3. Compare wording and seasonality where Trends has enough data.

4. Review Search Console queries and pages if the site already has visibility.

5. Use Glarify to organise query opportunities, possible gaps and competing pages.

6. Separate provider, price, local and informational tasks.

7. Group close variants that one page can satisfy.

8. Assign one URL and one action to every retained cluster.

Output

The first plan might contain:

• one improved Nairobi office-cleaning service page;

• one pricing guide if research confirms a distinct cost-information need;

• stronger Google Business Profile alignment for proximity searches;

• one practical checklist article for early-stage readers; and

• no thin page for every Nairobi neighbourhood until service coverage and evidence justify one.

The output remains a hypothesis. Publishing creates the next evidence: impressions, queries, clicks, enquiries and customer feedback.

Turn the map into the next three actions

Do not finish with 500 rows and no decision. Choose three actions:

1. Fix the page closest to revenue. Improve the service, category or product page that already owns the strongest relevant cluster.

2. Resolve overlap. Merge or refocus pages that compete for the same task before adding another URL.

3. Create one missing support page. Answer a question that helps the customer move towards the core offer and gives the service page a natural internal link.

After publication, review the page and query together. Google Search Console lets you see which pages appeared for a query, while Glarify can help organise that performance into content updates, gaps and tracking. Change the map when the evidence changes.

Keyword research works when it leads to fewer, clearer decisions: the right topic, the right page, the right audience and a measurable next step.

If your site has scattered service pages, unclear ownership or performance data you cannot turn into actions, request an SEO audit from Amplify SEO. The audit should show which pages to improve, merge or create before you commit to a larger publishing plan. You can also review the current SEO packages if you need ongoing implementation.