M6.L1 · Search & SEO
How search works in 2026
What you'll be able to doMap the 2026 discovery landscape — classic rankings, AI-generated answers and LLM assistants — and identify which surfaces your business can realistically win.
Learn
Two sentences that will each waste your year
"SEO is dead." "SEO hasn't changed." Both are wrong, and each one costs a year if you believe it.
What's true: fewer clicks leave the results page than a few years ago, because search engines now often generate an answer at the top and many people read it and stop. What's also true: search remains the largest intent-driven channel that exists — the only place strangers announce, in their own words, that they want what you sell. And a second prize has appeared next to the old one. It used to be rank and get the click. Now it's rank and get the click, or be the source the generated answer cites. Sometimes both.
One warning before the map. This module teaches mechanisms, not screens. The interfaces — what the answer box is called, where the citations sit, which assistant is fashionable — will have changed by the time you read this. The mechanisms underneath change slowly, and they are what you'll optimise. Whenever a tactic in this module depends on a mechanism, it will survive a redesign. Whenever advice you read elsewhere depends on a screenshot, it won't.
The pipeline underneath everything
Classic search still runs on three steps: crawl (a bot fetches your page), index (the engine stores and interprets it), rank (results are ordered by relevance and trustworthiness for each query).
Nothing else in this module works if this fails. A page a crawler can't reach can't rank, can't appear in an AI answer, and can't be retrieved by an assistant — the newest surfaces eat from the oldest pipeline. Lesson 6.5 checks yours.
The answer layer
On many queries, the engine now generates an answer above the results — Google calls its version AI Overviews; the name matters less than the mechanism: the engine synthesises an answer from pages it has retrieved, and credits some of them as sources.
Three consequences:
- On queries the answer fully absorbs — definitions, general how-often, general how-to — clicks fall. Ranking #1 there wins less than it used to.
- Being a cited source puts your name in front of the searcher even without the click. That's not nothing: it's the cheapest brand exposure a small business can earn.
- What gets cited is retrievable, parseable, specific content — which is exactly what the rest of this module builds.
Assistants as a discovery surface
People increasingly ask ChatGPT, Claude, Perplexity or Gemini directly: "best client-onboarding software for small accountancy firms", "good physiotherapist in Graz for a running injury". The assistant retrieves from live search indexes and its own crawls, synthesises, and names or cites sources.
Hold on to the mechanism: assistants mostly recommend what they can retrieve and what other pages say about you — review platforms, comparison articles, forums, directories. Your own site is necessary but not sufficient. Lesson 6.6 turns this into a strategy; today you only need to observe it.
Which intent still clicks
Queries come in rough kinds — you'll formalise this in Lesson 6.2:
- Informational ("how often should you see a physiotherapist") — most absorbed by generated answers. Prize: citation, not click.
- Commercial ("stoneware vs porcelain plates", "physiotherapie graz kosten") — comparing before choosing. Still clicks, because choosing needs detail.
- Transactional and local ("physiotherapie graz", "handmade dinner set buy") — booking, buying, contacting. Still clicks, because you cannot book an appointment inside an answer box.
Notice whose problem the zero-click shift really is: publishers who monetised informational traffic. A business that sells things needs commercial and transactional intent — and that intent still arrives at websites. Panic is for a business model you don't have.
Strings lost. Things won.
Old search matched strings: pages that repeated "physiotherapy Graz" often enough. Modern search models entities and topics — things and their relationships. It knows "rotator cuff", "dry needling" and "sports injury rehab" belong to one topic; it knows your business is an entity with an address, reviews and people.
So let's bury the corpse formally, once, here: keyword density is dead. Repeating a phrase adds no meaning to a page that engines read semantically, and heavy repetition reads as spam. You optimise in 2026 by genuinely covering a topic and by being a clearly identified, corroborated entity. Every "repeat your keyword N times" guide you meet from now on is telling you the year it was written.
Herzog runs the five searches
Watch the reconnaissance method on Herzog Physio — a business whose entire digital funnel is ~800 mostly-branded visits a month. Tomas runs five searches and writes down what each results page actually contains:
- "wie oft zur physiotherapie" (informational) — an AI-generated answer citing two national health portals. No clinic appears anywhere. Verdict: ranking here is pointless; being a cited source is the only prize, and a small clinic won't outrank a health ministry. Skip.
- "was kostet physiotherapie" (informational–commercial) — a generated answer with vague national ranges, then People Also Ask. Nobody publishes real local prices. That gap becomes a Lesson 6.3 brief.
- "physiotherapie graz" (local–transactional) — a map pack of three clinic profiles with review stars, then ads, then organic. And here it is: Herzog's own unclaimed profile — 23 reviews averaging 4.6 — is already ranking in that pack, without Tomas ever having touched it. Strangers are reading unanswered reviews and finding no booking link, on the most valuable surface in his market. Lesson 6.4 lands this properly.
- "physiotherapie graz kosten" (commercial) — local results and plain organic, no generated answer. Click-rich, winnable, directly connected to money.
- Perplexity: "good physiotherapist in Graz for a running injury" — names three clinics, with citations drawn from review pages and a local directory, not from any clinic's own website. The third-party layer, visible in the wild.
One page of notes, and Tomas knows something most competitors don't: his branded traffic means search currently only confirms people who already have his name. Every surface above is unbranded territory — and one of them he's already ranking on by accident.
Why this channel still compounds
A ranking or a citation, once earned, keeps paying without a per-click fee. That makes search the highest-ROI channel available to a business with €150 a month — precisely because it's slow. Competitors with ad budgets won't wait three months for a result. You can. That asymmetry is the whole strategy, and the next five lessons execute it.
Do
Exercise 6.1.1 — Your search landscape map
Map your search landscape. Run the five searches below for your business's world and record what each results page actually contains. This is reconnaissance — you're learning what game is being played on each query before you play it.
Write these down — in your plan document, or on the worksheet at the end of this lesson. Five rows: two informational queries, two commercial/transactional, and the same topic asked to an AI assistant as row five.
| What to record, per search | Guidance |
|---|---|
| The query | The exact search used, in the language your customers search in |
| The surface | Google · Bing · ChatGPT · Claude · Perplexity · Gemini. Row 5 must be an assistant |
| What the results page contains | Any of: AI Overview / generated answer · ads · local pack · shopping · videos · People Also Ask · plain organic only · assistant answer with citations · assistant answer without citations |
| Who wins | Up to 25 words: which sites are ranked or cited — names, not "various websites" |
| Would a searcher still click? | Yes — the answer needs a visit · partially · no — the answer is fully absorbed |
Then one takeaway for the whole exercise, 40–100 words: where are the click-rich or citation-rich openings for my business? Herzog's five completed rows above are your worked reference — and if all five of yours come out "fully absorbed", your queries are probably all informational.
Where this goes: section 7.1 — Search landscape — of your Marketing Plan.
Check
Rubric
Mark your own work against these criteria.
| Criterion | 8–10 | 5–7 | 1–4 |
|---|---|---|---|
| Query spread | Genuinely 2 informational, 2 commercial/transactional, 1 assistant — in customer language | Spread present but queries are marketer-phrased | Five variations of the same query, or brand searches only |
| Observation accuracy | Features recorded match what the page really contained | Some features missed or guessed | Rows describe what the student expected, not what appeared |
| Winner specificity | Named sites and surfaces per row | "Big sites" and "competitors" without names | Blank or generic |
| Takeaway quality | Names a specific click-rich or citation-rich opening for this business | Restates the rows without a conclusion | "SEO is important" |
Pass: 5+ on every criterion. Distinction: 8+ on all four.
Quiz — four questions. Pick an answer to see whether you were right.
1. Which query type still reliably sends clicks to websites in 2026?
- a) "what is physiotherapy"
- b) "how often should you see a physiotherapist"
- c) "physiotherapie graz kosten" ✔
- d) "history of ceramics"
Why: commercial and transactional intent needs a visit — you cannot compare local prices or book an appointment inside a generated answer. The informational queries are the ones the answer layer absorbs.
2. "Repeat your keyword 15 times so Google knows what the page is about." Why is this obsolete advice?
- a) The limit is now 10 times
- b) Search engines model topics and entities semantically; repetition adds no meaning and heavy repetition reads as spam ✔
- c) Keywords only matter in meta tags now
- d) Because AI Overviews don't read body text
Why: engines stopped matching strings years ago. Coverage of a topic — the concepts a genuine expert would touch — is what relevance means now, which is why Lesson 6.3 teaches entity coverage instead.
3. Tomas's most valuable discovery in the five-search reconnaissance was…
- a) That an AI answer cites health portals for "wie oft zur physiotherapie"
- b) That his unclaimed profile — 23 reviews, 4.6 — is already ranking in the local pack, with unanswered reviews and no booking link ✔
- c) That Perplexity exists
- d) That his competitors run ads
Why: the local pack is the most valuable surface in his market, and he's on it by accident. Claiming and completing that profile costs an evening and €0 — it's the single cheapest win in his entire plan.
4. An AI Overview fully answers "how often should you see a physiotherapist". What is the realistic prize for a clinic on that query?
- a) Outranking the answer with a longer article
- b) Nothing — abandon search entirely
- c) Being one of the cited sources — name exposure without the click ✔
- d) Buying an ad above the answer
Why: on absorbed informational queries the click is gone, but the citation isn't. A cited name in front of a future patient is cheap brand exposure — and the qualities that earn citations are built in Lessons 6.3 and 6.4.
Advance
You now hold an accurate map: which of your queries still click, which only cite, and where an assistant gets its recommendations. Most people optimising search in 2026 have never once looked.
Next: M6.L2 — Keyword & topic research on a €0 budget. The map showed you the surfaces; now you choose the exact queries to fight for — thirty of them, in five clusters, built from your customers' own words without spending a cent.
Mark your own work
| Good | Not yet | |
|---|---|---|
| Real searches run | You looked at five actual results pages today | You filled rows from memory |
| Customer language | Queries phrased as your customers speak | Industry jargon your Module 2 interviews never used |
| Winners named | Specific sites and surfaces per row | "Competitors" and "big sites" |
| An opening found | Your takeaway names a click-rich or citation-rich gap | A summary with no decision in it |
Worksheet
THE SCHOOL OF NET MARKETING
Lesson 6.1 — How search works in 2026
RUN FIVE SEARCHES — write down what the page ACTUALLY contains.
Features key: AI = generated answer ADS = ads LOC = local pack
PAA = People Also Ask ORG = plain organic
CIT = assistant answer with citations
1. INFORMATIONAL query: ________________________________
features: ____________ who wins: ____________________
would a searcher still click? ☐ yes ☐ partly ☐ no
2. INFORMATIONAL query: ________________________________
features: ____________ who wins: ____________________
still click? ☐ yes ☐ partly ☐ no
3. COMMERCIAL query: ________________________________
features: ____________ who wins: ____________________
still click? ☐ yes ☐ partly ☐ no
4. COMMERCIAL query: ________________________________
features: ____________ who wins: ____________________
still click? ☐ yes ☐ partly ☐ no
5. ASK AN ASSISTANT (ChatGPT / Claude / Perplexity / Gemini)
question: ____________________________________________
who was named: ________________________________________
which sources were cited: _____________________________
MY TAKEAWAY (40–100 words)
The click-rich or citation-rich openings for my business:
____________________________________________________
____________________________________________________
SELF-CHECK
☐ I ran real searches today, in customer language
☐ I named the actual winners, not "competitors"
☐ I know which of my queries click and which only cite
☐ My takeaway names a specific opening
Next: Lesson 6.2 — Keyword & topic research on a €0 budget.
theschoolofnetmarketing.com/learn/keyword-research-zero-budget