Ranking #1 and still losing the click is not a bug report — it is the new baseline for some queries
A South African client came to us convinced their tracking was broken: their page ranked first for a query it had held for over a year, yet click volume for that specific term had visibly dropped. The tracking was fine. A Google AI Overview had started answering the question directly at the top of the page, and a meaningful share of searchers got their answer without scrolling down to the organic listing at all.
This is not a universal collapse of organic search, and treating it as one leads to the wrong fixes. AI Overviews appear selectively, most heavily on informational and definitional queries, and considerably less often on the local, transactional and highly specific commercial searches that drive most South African SME enquiries. The practical task is knowing which of your queries are actually affected, and responding to that specific subset rather than rewriting your entire site out of a general anxiety.
This piece stays deliberately narrow: how AI Overviews select sources, which query types they touch in practice, and how to read your own data correctly. For the broader question of Answer Engine Optimisation as a discipline, see our companion AEO vs SEO guide.
How AI Overviews tend to select sources
| Signal | Why it matters |
|---|---|
| Existing strong ranking | AI Overviews draw heavily from pages already performing well in classic organic results |
| A clear, extractable answer | A buried conclusion several paragraphs deep is harder to lift cleanly into a summary |
| Consistent facts across sources | Conflicting details about the same topic reduce confidence in any single source |
| Structured, specific content | Vague marketing language rarely gets quoted; specific, checkable statements do |
| Recency and freshness signals | Time-sensitive queries favour visibly current, well-maintained pages |
Which South African queries actually get an AI Overview
Broad informational and definitional queries — "what is," "how does," "why does" — are the most consistently affected. These are exactly the queries where a direct-answer format serves the searcher well, and where a click-through was often marginal even before AI Overviews existed.
Local and "near me" queries are affected less consistently, and Google Maps and the local pack remain a distinct, powerful surface of their own for these searches — our local SEO guide covers that ground in full. Highly specific commercial-investigation queries ("best accounting firm for a 20-person manufacturing business in Gauteng") and clearly transactional queries ("book a consultation with [X]") are affected the least, because the answer genuinely depends on a click-through to be useful.
The practical implication: audit which of your pages target broad informational queries versus specific commercial or local ones, and expect different behaviour from each category rather than applying one blanket assumption across your whole site.
Reading your Search Console data correctly
- Separate impressions from clicksFilter to the specific query, then compare the impression trend against the click trend rather than reading them as one combined number.
- Check whether position actually movedIf your average position is stable but clicks fell, that points toward answer absorption — not a ranking problem you can fix by "doing more SEO."
- Confirm the query typeInformational query showing this pattern: expected, and hard to fully reverse. Commercial or local query showing this pattern: worth deeper investigation.
- Look at the trend, not one weekAI Overview rollout and triggering behaviour still shifts noticeably; judge on a several-week trend rather than reacting to one dip.
- Track branded search as a counter-signalIf branded search or direct traffic is rising while a specific informational query’s clicks fall, awareness is likely still growing — just through a different path.
This sequence stops a business from spending money "fixing" a ranking that never actually broke, and instead directs attention to whether commercial queries — the ones that matter most — are genuinely losing ground.
What to actually change on your pages
- State a direct, specific answer to the core question within the first two or three sentences
- Follow the direct answer with supporting detail, nuance and proof — do not delete it, just reorder it
- Keep facts (pricing logic, service areas, timelines) consistent across every page that mentions them
- Add or refresh Article, FAQPage and Organization schema on pages relevant to common queries
- Update genuinely time-sensitive pages (pricing, availability, dated guides) instead of leaving them stale
- Protect the pages behind commercial and local queries — these still convert clicks and deserve continued investment
What not to do
Also resist the temptation to strip useful depth and nuance from a page purely to make it "shorter for AI." A thin page satisfies neither a sceptical human reader nor a summarisation system looking for a well-supported, checkable answer.
Query type versus recommended response
| Query type | AI Overview likelihood | What to prioritise |
|---|---|---|
| Broad informational / definitional | High | Answer-first structure; accept some click reduction is likely permanent |
| Local / "near me" | Moderate, Maps often dominates instead | Google Business Profile accuracy and location page relevance |
| Specific commercial investigation | Lower | Depth, comparison detail and proof — protect the click, it still converts |
| Transactional / branded | Lowest | Conversion path clarity matters more than answer summarisation risk |
Talking to stakeholders who saw one scary headline about AI Overviews
A common internal conversation right now: a business owner or board member reads one alarming article about AI Overviews "killing organic traffic" and asks the marketing team to explain why the SEO budget still makes sense. The honest response is to bring the actual Search Console data to that conversation, filtered by query type, rather than arguing in the abstract.
Show which queries genuinely lost clicks with position unchanged, which queries are unaffected, and which commercial and local queries are still converting exactly as before. That specific, evidenced answer tends to land far better than a general reassurance, and it keeps the conversation focused on the handful of pages that actually need attention rather than triggering a full-site panic rewrite.
The Nexus take
Do not measure AI Overview impact by how your rankings look. Measure it by how your clicks and enquiries move on the specific queries that were already carrying commercial weight.
Nexus SEO measurement principle
Why South Africa is not experiencing this uniformly
AI Overviews rolled out unevenly across markets and query languages, and South African search behaviour reflects that unevenness back. English-language informational queries in well-covered categories — general business, technology, personal finance — tend to see AI Overviews more consistently than niche local trades, regional Afrikaans or isiZulu-language queries, or highly specific B2B categories where the underlying web content is thinner to begin with.
This means two South African businesses in different categories can have genuinely different experiences and both be right about what they are seeing. A national consumer brand competing on broad informational terms should expect more AI Overview exposure than a regional industrial supplier whose buyers search in narrow, highly specific commercial terms. Calibrate your expectations, and your reaction, to your actual category—not to a general headline about "AI Overviews everywhere."
A worked example: reading one query’s numbers correctly
Take a realistic case: a Cape Town interior design studio ranks position two for "how much does interior design cost in south africa," a broad informational query. Over three months, impressions for that query hold steady around 1,200 a month, but clicks fall from roughly 90 to 40. Read on click volume alone, that looks like a ranking collapse worth escalating urgently.
Check position and the picture changes: average position barely moved, sitting between 1.8 and 2.3 across the same period. Steady impressions, stable position, falling clicks is close to a textbook case of answer absorption — an AI Overview is very likely now answering a version of that question directly, and a portion of searchers who previously clicked through no longer need to. The honest response is not a ranking fix. It is confirming that the studio’s commercial and local queries — "interior designer near me," "book an interior design consultation" — are unaffected, and continuing to invest content effort there instead.
This is the exact sequence worth running on your own top twenty queries before reacting to any single number: confirm position first, separate it from clicks second, and only treat a query as a genuine problem once both have moved in the wrong direction together.
A fifteen-minute self-audit
- Export your top twenty organic queries from Search Console for the last 90 days
- Sort by impressions, then check whether clicks fell while position held steady for each one
- Flag only the queries where position and clicks both dropped — that is your real watchlist
- Cross-check flagged queries against query type: informational queries losing clicks alone are expected; commercial or local queries on that list deserve real investigation
- Search the flagged query manually and note whether an AI Overview currently appears for it
What to do next
This week: pull your top twenty organic queries in Search Console, separate impressions from clicks, and flag any query where clicks fell while position held steady — that is your real AI Overview watchlist.
This month: restructure the two or three highest-value pages behind informational queries on that watchlist to lead with a direct, specific answer.
Pair this guide with our companion pieces on AEO vs SEO and on preparing for AI search more broadly if you want the full picture beyond this one Google feature.
