Twelve weeks, live. Real sites, real audits, real work you can show. Built for people doing this alongside a full-time job.
Kenya & Nigeria · 6–8 hrs a week
Your juniors were handed SEO with no training, so they are learning on client sites and the clients are paying for the mistakes. Train them properly in one quarter.
You have been told to do SEO with no budget to hire for it. A hire is a salary forever. A retainer is 30 to 80k a month and the knowledge leaves with the contract. This stays in the building.
You run content, brand or social well. Search is the one channel you cannot quote on with confidence, and it is the one that compounds.
Someone types a question. A machine writes its own searches, reads the results, and answers without sending anyone to your site. The rules for being the source it quotes are not the rules most people learned. You can either understand that or keep guessing why the traffic moved.
Most training covers this with one idea and a statistic. There are at least six separate mechanisms at work, each one demands something different from a page, and the evidence moves fast enough that you need to be able to read it yourself.
A prompt is not a search. The model works out what it needs to know, writes its own queries, runs them, reads what comes back, and only then answers. Three or four hidden searches, none of which the user sees and none of which you can see.
Look at the third one. That is the model going straight to a company's own website to check a fact. In the largest public study of this behaviour, roughly three quarters of the generated queries looked like that.
"Who should I hire to clean my sofa in Nairobi?"
The prompt is the doorway. These are what walk through your site.
Some prompts never trigger a search at all, because the model already knows the answer and the answer does not change. Content sitting behind those prompts cannot earn a citation no matter how good it is. Knowing which is which decides what you bother writing.
How one prompt becomes several hidden queries, how shallow that actually is, and how to work out which queries a given prompt would generate for a business you are auditing.
Models arrive with specific companies in mind before retrieval even starts. Getting into that candidate set is brand recognition, third-party mentions and classic organic visibility. A completely different job from the one below, and the half most people skip.
Once you are a candidate, the model comes to your own site for prices, specifications, terms and policies. Your service pages are search content, not sales collateral. Vague copy is where you lose the citation to a competitor who stated the number.
Search engines match things, not strings. Which entities a page has to cover to read as genuinely expert, and how to write so a machine can lift a clean, self-contained answer out of it. Answer-first structure, question-shaped headings, and facts that are not buried in a slider.
Whether you are being cited, where, and for which prompts. Which prompts are worth tracking and which are vanity. The same business-value discipline you already apply to keywords, applied to prompts.
This field is full of people quoting a percentage long after the thing that produced it changed. You will learn to separate the durable mechanism from the dated number, so when a client asks where a claim comes from you have a real answer. That habit outlasts every specific statistic in the module.
Each one ends in something you can put in front of a client or an employer. Not a quiz score.
Rank keywords by what they are worth to the business, and read a competitor's entire strategy from their site structure alone.
Say what query it serves, whether it matches the results page, which entities it is missing, and what you would change first.
Crawl a site, read what search engines actually see, and separate real problems from noise a plugin flagged for no reason.
Write pages that get quoted rather than skipped, and explain exactly why a competitor is being cited and your client is not.
Use Search Console as a diagnostic instrument. Say what changed, why it changed, and what it was worth, with the data open.
Scope the work, write the proposal, price it, and answer the questions a buyer asks in the moment before they say yes.
Live sessions with assignments on real sites between them. You are not watching, you are producing work that gets reviewed and sent back.
Apply it under supervision. Three weeks, and it is the half that produces the portfolio and costs us the most to run.
Each module assumes the one before it. Click through to see what a week contains and what you produce in it.
Crawling, rendering, indexing and ranking, in that order, because every problem you will ever diagnose sits in one of those four stages. Set up your toolkit and run a first pass on a live site.
Find what people actually search for, then rank it by what it is worth to the business rather than by volume. Read a competitor entire strategy from their site structure.
Map keywords to pages by intent. Decide what to create, what to optimise, and what to leave alone. Build the topic cluster that everything after this sits inside.
Judge whether any page serves its query, then say what to change and why. Entities, information gain, cannibalisation, and reading titles and metas as a diagnostic instrument rather than a writing task.
A cluster is not a cluster because you drew it. It is a cluster because the links exist. Anchor text, hub and spoke direction, click depth, and finding the orphan pages nothing points at.
Crawl budget, JavaScript rendering, canonicals, redirects and index bloat. Separate the problems worth fixing from the ones a tool flagged for no reason.
Core Web Vitals, schema as a machine-readable restatement of what a page already proves, and how to debug something when two tools disagree with each other.
Retrieval, query fan-out, the shortlist problem, first-party verification, entities and extractability. The full module, taught from evidence rather than from headlines.
Google Business Profile, local signals and the location page trap. Product and category structure for stores. Then Search Console and GA4 as diagnostic instruments rather than screenshot factories.
Three weeks applying all of it under supervision. One-to-one sessions, group coaching, portfolio review, and the proposal and pricing frameworks you need to charge for this work.
Founder, OnMedia · Co-founder, Search Safari
Connect on LinkedInDaniel built OnMedia's search practice from nothing. He ranks Kenyan brands on live sites, not in theory, and he co-founded Search Safari, the first dedicated community of SEOs in Kenya.
He serves as the lead trainer for this cohort, joined by two other active industry experts. This bootcamp is the method they use on client work, cut into twelve weeks you can do around a job.
"After 12 years in the SEO game, Daniel beats literally everyone else I have ever worked with. Skills, ideas, implementation and a sharp eye for tech that moves our business forward."
"Daniel has been a solid go-to guy for all things content strategy and delivery. His unique approach, utmost professionalism and attention to detail have never once let me down."
Work gets done. The knowledge leaves when the contract does, and you are back where you started.
The skill is in the building until they leave. Then you hire again and pay to train the next one.
Someone already on your team does the work, or briefs an agency properly. Either way the skill stays.
Paid in full. Closes 31 August 2026, then the price goes up. Admission is strictly by application.
One seat, paid in full. Admission is strictly by application.
Two seats, KES 25,000 each. One person trained alone tends to leave with the skill.
KES 16,500 at enrolment and KES 16,500 before week 10, which totals KES 33,000 against KES 29,999 paid once. Week 10 is when the one-to-ones and portfolio review start, so the balance falls due before that phase begins. Standard price only, not early bird.
Apply for payment planAgencies and in-house teams. Quoted per organisation and invoiced properly so your finance team can process it. Send how many people and what they currently do, and a quote comes back.
Get a quote on WhatsAppCapped at 20 seats. Admission to the cohort is by application. When the room is full, enrolment closes.
Starts 15 SEPTEMBER 2026. 20 seats.
Apply for a seat