Free Mockup
SEO

The first hundred words of a blog post now do most of the work, for a human skimming and a model summarising.

Generative engine optimisation is mostly a structure problem. Here is the answer-first template, section by section, and how to retrofit it onto articles you already have.

Two guides answered the same question. One got quoted.

A Durban bookkeeping practice and a Joburg competitor both published a guide to provisional tax deadlines within a month of each other last season. The competitor's article ran longer, covered more edge cases and was, by most measures, more thorough. When a client later asked an AI assistant about provisional tax deadlines, it summarised the Durban firm's page and named it as the source. The longer, more thorough article never came up.

The difference was not accuracy or effort. It was structure. The Durban article opened with a plain, two-sentence answer to the exact question a person would type, then organised the rest of the page into sections that each stood on their own. The competitor's article built its argument the way a print magazine feature would, context, then history, then nuance, then finally the answer buried in paragraph nine. A human patient enough to read the whole thing would have gotten more value from it. A model skimming for an extractable answer never got that far.

This is the practical shape of generative engine optimisation, stripped of the acronym: write so the point can be found and lifted cleanly, near the top and again at every section break. It is not a separate content discipline bolted onto SEO. It is the same clarity SEO always rewarded, just enforced more literally now that something other than a human is doing the skimming.

What "answer-first" changes about a normal blog structure

ElementNarrative defaultAnswer-first structure
OpeningScene-setting, then a slow build to the pointDirect answer in the first two to three sentences
Headings"Some thoughts on X" or vague section titlesPhrased as the real question a buyer would type
Section independenceLater sections assume you read the earlier onesEach section restates enough context to stand alone
Proof placementSaved for a "results" section near the endSits directly beside the specific claim it supports
FAQ blockGeneric filler questions bolted on for SEOReal buyer questions, answered plainly, no padding

The direct-answer paragraph is not a summary, it is the actual answer

A common mistake when teams first attempt "answer-first" writing is opening with a summary of what the article will cover, instead of the answer itself. "This guide explains provisional tax deadlines for South African sole proprietors" is a table of contents dressed as an opener. It tells a model nothing it can quote with confidence.

The working version answers the literal question in the first sentence or two, using plain language and a specific figure or fact where one exists: "Provisional tax for South African sole proprietors falls due in two periods each year (the last business day of August and the last business day of February) with a third voluntary top-up payment available in September." That sentence works as a standalone citation. It also happens to be exactly what a human scanning on a lunch break wanted in the first place.

Everything below that opening sentence should support, qualify or extend it (exceptions, edge cases, what happens if you miss a date) without contradicting or diluting the plain answer sitting at the top. If your honest answer genuinely is "it depends," say so immediately and then explain what it depends on, rather than making a reader (or a model) wade through the dependency logic before finding out that a simple answer does not exist.

Write sections that survive being read alone

AI answer systems frequently quote or summarise a single section of a page, not the whole thing. A section that only makes sense after reading three paragraphs above it is a section that gets misquoted, or skipped entirely in favour of a competitor whose section stands on its own.

  • Each H2 restates just enough context that a reader arriving cold understands the claim
  • Avoid pronouns referring back to something explained two sections earlier ("this approach," "the method above")
  • One core claim per section, not three loosely related ideas sharing a heading
  • Specific numbers, names or examples sit inside the section they support, not in a separate "proof" section
  • A section title phrased as a real question a buyer would type, not a clever label

This does mean slightly more repetition than a traditionally flowing article would use, restating the topic briefly at the start of a section instead of relying purely on "as mentioned above." Readers skimming on mobile rarely mind the repetition. They are usually skipping between sections anyway, which is exactly the behaviour this structure is designed to support.

Headings are questions now, not labels

A heading like "Our Approach" or "The Bigger Picture" tells a search engine and a reader almost nothing about what is actually inside the section. A heading phrased as the real question a buyer types ("How long does provisional tax registration take?") does two jobs at once: it orients a human scanning the page, and it gives an AI system an almost pre-formed question-answer pair to extract.

This does not mean turning every article into a wall of identical "What is X?" headings with no personality. It means testing each heading against a simple standard: would a real customer plausibly type this exact question into a search bar or ask it to an assistant? If the honest answer is no, the heading is probably serving the writer's outline more than the reader's actual question.

Pull your best headings straight from real customer emails, sales call transcripts and support tickets. The phrasing customers actually use is almost always more specific (and more citable) than the phrasing a content calendar template suggests.

A retrofit sequence for articles you already have

  1. Audit your top pagesStart with pages already ranking or driving enquiries, retrofitting proven content usually beats writing something new from scratch.
  2. Rewrite the opening two sentencesReplace scene-setting with the direct, specific answer to the page's primary question.
  3. Re-order sections by real question, not narrative flowMove the most commonly asked question to the top, even if it feels less like a natural essay structure.
  4. Make every H2 a real questionRewrite vague section labels using language pulled from actual customer conversations.
  5. Add or fix schemaArticle and FAQPage schema where the FAQs are genuinely drawn from real buyer questions, not invented to hit a quota.

Retrofitting a proven page for answer-first structure is usually a higher-return afternoon than writing a brand-new article nobody has validated yet.

Nexus content principle

The FAQ trap: padding a page instead of answering questions

FAQ schema became popular fast, and plenty of teams responded by bolting five generic questions onto the bottom of every page purely to qualify for the format. "What is [service]?" and "Why choose us?" answered in two vague sentences each does not fool an AI system, and it does not fool a sceptical reader either, both recognise filler when they see it.

A genuinely useful FAQ section answers the specific follow-up questions a real buyer asks after reading the main content, pricing logic, edge cases, what happens if something goes wrong, how long something takes. Pull these directly from your sales team's inbox, your support tickets and the actual questions people ask on a discovery call. If your sales team cannot recognise a question in your FAQ block, it probably should not be there.

Keep answers specific and short enough to stand alone as a citation, but resist compressing a genuinely nuanced answer into one dishonest sentence just to keep the format tidy. "It depends on X and Y" is a perfectly good FAQ answer when that is the truth.

What this structure should never sacrifice

Watch for a second failure mode: chopping a page into so many short, choppy sections that it loses any argument or narrative thread a genuinely complex topic actually needs. Some subjects deserve a longer, connected explanation. Use answer-first structure for the parts of a page that are genuinely question-and-answer in nature, and let more complex reasoning breathe where it has to.

Measuring whether the structure change actually worked

SignalWhat to watchCaveat
Featured snippet / AI Overview appearancesWhether your page starts appearing as a cited or summarised sourceInconsistent visibility across tools; treat as directional
Branded search volumeGrowth in searches for your business name after a period of visibilityTakes weeks to months to show a trend
Time on page for retrofitted articlesWhether readers who arrive are actually finding the answer fasterA drop is not automatically bad if the page now answers faster
Qualified enquiries from the pageThe commercial outcome that actually mattersThe only number worth defending in a budget conversation

A retrofit example, side by side

Before: "In this article, we will explore the different aspects of provisional tax registration and what South African sole proprietors need to consider when starting this process, which can often feel overwhelming for new business owners." That sentence sets a friendly tone and states nothing a system, or a rushed reader, could act on.

After: "Provisional tax registration for a South African sole proprietor typically takes one to two weeks once SARS eFiling access is confirmed, and can be started the same day you register for a tax number." Same topic, same honesty, radically more useful in the first eight seconds a reader or a model spends with the page.

The retrofit did not remove anything — the original article's context about who needs to register, common delays and what documents to prepare all stayed on the page, just below this new opening line instead of buried three paragraphs beneath it.

What to do next

This week: pick your three highest-traffic or highest-intent articles and rewrite only the opening two sentences of each into a direct, specific answer. That single change is often the fastest-return edit available to a content team right now.

This month: retrofit those same three articles fully, reordered sections, question-phrased headings, and an honest FAQ block built from real buyer questions rather than invented filler.

Pair this with our broader guide on preparing your business for AI search for the entity-consistency and technical side of the same shift, and read our SEO guide for South African business owners if your foundations need attention first.

FAQs

Questions this article answers.

Mostly, yes. It sharpens existing good practice (clarity, direct answers, useful structure) because AI systems now extract and summarise content more literally than a human skimming a search results page ever did.
Not if applied selectively. Use it for genuinely question-and-answer content. Longer narrative pieces (case studies, opinion pieces) can keep a more traditional structure where the format actually suits the content.
Only where you have real, specific buyer questions to answer. Schema on a page with invented, generic questions adds markup without adding genuine value for a reader or a citation.
Rewritten, in most cases. A page already earning some visibility usually retrofits faster and cheaper than a brand-new article has to earn visibility from zero.
Sales call transcripts, support tickets, WhatsApp conversations and search console query data are more reliable sources than a generic keyword tool alone.
Yes, and arguably matters more there, local categories are often thin, so a genuinely well-structured, specific local page has an easier time standing out and getting cited.

Make your best pages easier to cite

We audit your highest-value pages and retrofit them for answer-first structure, without stripping out the voice and proof that make them yours.

Want a site that earns the enquiry?

Request a free homepage mockup, or WhatsApp Nexus with your brief, we respond within a business day.