AEO Consultant, Kenya

Collecting Kenyan questions from calls, chats and search suggestions

The richest source of questions for a Kenyan business is rarely a keyword tool. It is the WhatsApp Business inbox, the call center log and the branch staff who hear the same handful of questions every morning. Customers ask whether you accept M-Pesa or bank transfer, whether delivery reaches their estate, what happens if a product arrives damaged, whether a service needs a site visit first.

I gather these from several places:

• exported chat and ticket histories, with names, phone numbers and account details removed by your team first
• short interviews with sales, support and branch staff about the questions they answer repeatedly
• search suggestions and People Also Ask boxes viewed with a Kenyan location setting
• questions asked in Swahili or a mix of Swahili and English, recorded exactly as customers wrote them

Chat logs and call notes contain personal data. Kenya's Data Protection Act applies to how they are processed, so I work only with redacted exports and suggest you confirm the process with your data protection officer or legal advisor.

The result is a question map: each question grouped by topic and buying stage, assigned to one page, with the format its answer needs. This applies the answer engine optimization method to how Kenyan customers really phrase things, rather than to an imported keyword list.

Writing answers that work on a small screen

Many Kenyan customers will read your answer on a phone, sometimes in a data-saving browser, often between other tasks. An answer that takes three sentences to reach the point loses them, and it also loses the snippet or assistant that was looking for a clean passage to lift.

Each answer follows the same pattern. The heading asks the question in customer language. The first sentence replies directly: yes, no, a definition or the first step. The second sentence adds the main condition. Detail, examples and exceptions follow. If the question is about a process, such as registering a warranty or booking a technician, the steps are numbered. If it is about choosing between options, a small table with consistent criteria does the job.

Before publishing, every answer is checked against a few questions:

• Would the first paragraph make sense if quoted alone?
• Does it use the terms customers use, such as Paybill, county or estate names?
• Has someone in operations confirmed the facts?
• Does it avoid shilling amounts that will date, unless your team commits to updating them?

Answers touching tax, KRA requirements or employment rules stay general and suggest the reader speak to their own advisor. A confident short answer on a regulatory point is exactly the kind of passage that gets quoted and can mislead.

FAQ, HowTo, QAPage and speakable markup, applied honestly

Structured data labels a passage as a question and answer, or a set of steps, so machines can read it with less guessing. It is not a switch for extra space in results. The set of sites that get FAQ and HowTo enhancements has been cut back, and the criteria still shift, so any enhanced display is a hope rather than a plan.

The rules I apply:

• FAQPage on pages where your business wrote both question and answer and both are visible
• HowTo where a page gives real instructions, such as preparing a roof for a solar installation, with every marked step present in the text
• QAPage only for genuine community question pages, where a user asks and others reply
• Speakable-style markup, where a search engine supports it, for brief passages that read well aloud, treating that support as narrow and changeable

Every block is validated and compared with the visible page before release, and templates are checked so markup does not duplicate across pages. Wider structured data work for products and branches sits under schema markup consulting, and template or rendering problems go to the technical SEO work in Kenya.

Voice queries, featured snippets and People Also Ask in Kenya

Voice questions tend to be longer and more conversational: which pharmacy near Westlands is open late, how do I reverse an M-Pesa payment sent to the wrong number, does this school offer transport from Ruaka. Support for spoken Swahili varies by assistant and device, so I test spoken-style phrasings in both languages where your customers use them and note where an assistant struggles. Assistants rely on short passages and consistent business details, and opening hours or branch information that differ between your site and your profiles produce wrong answers; those conflicts are flagged for your local SEO in Kenya work rather than fixed twice.

For featured snippets and People Also Ask, I record which site holds the answer for each priority question, its format and why it wins. Often the current holder is simply more direct. Rewriting a section as a crisp definition, a numbered list or a compact table can be enough to compete.

None of this is permanent. A snippet can disappear behind an AI summary or move to another site. I follow the question set as a trend, use Search Console data for question-style queries and add a field on forms and the first WhatsApp reply asking what prompted the inquiry, stored in your CRM, so effort goes to questions that lead to customers.

Swahili and Sheng answers, and updates when things change

Where customers ask in Swahili, the answer should read like natural Kenyan Swahili, not a translation of an English sentence. Sheng is a judgment call: it appears in how younger customers phrase questions and may suit a consumer brand's chat replies or social answers, but it changes quickly and can sound forced on a formal page. The engagement runs in English; Swahili and any Sheng wording come from first-language speakers you trust, whether staff or a Kenyan writer, using the phrasing captured in the question log.

Local-language answers are worth adding where the log shows demand and your team can continue the conversation in that language. Otherwise a clear English answer is the better choice.

Answers go stale faster than many other pages, because they deal in specifics. Mobile money charges, delivery zones, opening hours, product ranges and KRA requirements all change. An assistant may keep repeating an old reply long after you have moved on. The answer register prevents that by recording, for each answer:

• who owns it in the business
• which facts it depends on
• what change should trigger a rewrite
• which language versions exist

For the separate question of which sources assistants cite, see generative engine optimization in Kenya.