Position zero was already an answer
Structure and markup refreshed for current answer engines; the original analysis is preserved.
Long before any chatbot, search had been turning into an answer: the featured snippet, People Also Ask, the knowledge panel. Position zero already meant “be the answer, not a link” — but it cited its source. As November ends, two signals of where this accelerates: a science model pulled within three days for confident falsehoods, and a research-preview chatbot released on the 30th. The direction is not new; what changes is credit and trust.
the short answer
Search has handed back answers for years — featured snippets, People Also Ask, the knowledge panel. “Position zero” already meant being the answer, not a link, but it cited its source. What is starting to change is attribution (a composed answer may carry no link) and reliability (a generated answer can be confidently wrong — a science model was pulled this month for exactly that). The work is continuous: clarity, structure and earned authority won position zero and will feed whatever composes answers next.
key takeaways
- Long before any chatbot, search was already turning into an answer: featured snippets, People Also Ask, the knowledge panel. "Position zero" meant be the answer, not a link.
- The crucial difference: a snippet is extracted from a real page and shown with a link back, so being the answer still credited — and sometimes clicked — the source.
- What is starting to change is attribution (a composed answer may carry no link) and reliability (a generated answer can be confidently wrong, unlike a quote).
- This month previewed both: a science model was pulled within three days for authoritative-sounding falsehoods, and a research-preview chatbot was released on the 30th.
- The work does not change: clarity, structure and earned authority won position zero and will feed whatever composes answers next. Treat the shift as continuous, not a rupture.
the answer has been arriving for years
On the left, the answer we have optimised for since the mid-2010s: a snippet lifted from a page, shown on top, with a link back to where it came from. On the right, what a research preview released today gestures at: an answer in the tool’s own words, no link beneath it, and — as a withdrawn science model showed this month — able to sound right while being wrong, with no link beneath it for a reader to check against.
Why frame the month this way
There is a temptation, watching a striking new demo appear, to treat it as a clean break and to reach for a fresh playbook. The more useful frame, especially while the thing is a single day old, is continuity. The work of being chosen as the answer is not new to anyone who has earned a featured snippet: you make your point clearly and early, you structure the page so the answer is easy to lift, you become a source quality guidelines would call expert and trustworthy, and you get corroborated elsewhere so the engine is confident picking you over the alternatives competing for the same spot. That is the position-zero discipline, and it is precisely the discipline a system that composes answers will lean on, link or no link. Seeing the month as an acceleration rather than a rupture keeps you doing the work that actually compounds instead of chasing whatever the newest tool seems to reward this week.
It also keeps the measurement honest. For years we could watch the snippet, the attribution, and the click, and know where we stood. If answers begin to be composed without links, some of that visibility will thin, and the temptation will be either to declare the old metrics dead or to pretend nothing changed. Neither is right. The sober reading at the close of November is that the answer has been the destination for a long time, the credit and the click that came with it may not survive the next phase, and the way you earn a place in an answer is still the slow, familiar business of being clear, structured, and trusted — measurable or not.
The bridge, in three parts
Search has been an answer for years; position zero kept a cited source that the newer answer may drop; and the open question becomes whether the answer can be trusted. Open each part for where it leads.
01 Search has been an answer for years
It is easy, at the end of a month like this, to think the answer-instead-of-a-list idea arrived with the newest tool. It did not. For most of the past decade the results page has been quietly lifting answers out of pages and placing them above the links: the featured snippet that quotes a paragraph, the People Also Ask boxes that pre-empt the next question, the knowledge panel that summarises an entity. We even named the prize — "position zero" — and built a craft around earning it. So the unit of search has been drifting from a list you choose toward an answer you are given for a long time, gradually enough that it felt like normal optimisation rather than a turning point. Holding that in mind matters, because it means the shift everyone will soon talk about is an acceleration of something already underway, not a bolt from nowhere, and the skills you built for it are not suddenly obsolete.
02 What position zero kept that the new answer may not: a source
There is one feature of the featured snippet that is easy to take for granted and central to everything that follows: it cites. A snippet is extracted from a specific page and shown with a link back, so even when search answered for you, being the answer still credited a source and often sent a click its way. Measurement reflected that — you could see the snippet, see the attribution, see the traffic. The systems hinted at as this month closes point the other way: an answer composed in the tool’s own words, with the sources folded in and frequently no link out. If that becomes the norm, the gradient we have measured for years flattens into something stranger, where you can be the basis of an answer and receive neither credit nor a visit. The answer is not new; the disappearance of the link beneath it would be.
03 And whether the answer can be trusted
The second change is reliability, and this month gave an unusually clean preview of it. A large model built to help with science was put online in mid-November and withdrawn within three days, after researchers demonstrated that it produced confident, authoritative-sounding statements that were simply false — in some cases citing papers that did not exist. A featured snippet cannot do that; it is a quote from a real page, right or wrong on the page’s own terms. A composed answer can be fluent and entirely invented, with nothing to signal the difference to a reader. That episode is worth more attention than the excitement around it, because it shows the seam that opens as answers get generated rather than quoted: sounding right and being right separate, and a company can be described authoritatively and inaccurately at once. Being found is starting to include being found correctly — a measurement and a reputation problem rolled together, and a new one for most companies.
What to do with this
Keep earning position zero, and recognise that you are already building for what comes next. Make your answer to the obvious question clear and early on the page; structure content so a machine can lift a clean, correct statement without distortion; be the kind of source that quality guidelines describe as expert, authoritative, and trustworthy; and earn corroboration elsewhere so any system is confident drawing on you. None of that is new advice, and that is the reassuring part: the discipline that wins a snippet today is the discipline that gets you named in a composed answer tomorrow, with or without the link you used to get.
Do not over-react to the demo released on the 30th, and do not dismiss it either. Treat the month for what it is: an early, vivid sign that the answer-shaped search we have optimised for is about to lose its visible sources and gain a reliability problem, neither of which changes the underlying job. The model pulled in mid-November is the warning worth keeping — sounding right and being right are coming apart — and the response to it is the oldest one there is: be accurate, be clear, be corroborated. That is the measurement-minded, unspectacular groundwork the AC Group has built for 27 years, and it is what carries you across the change beginning this week.
The measurement problem this quietly opens
From a measurement seat, the move from a sourced snippet to a sourceless answer is not just a credit problem; it is a visibility problem for the people whose job is to know what is working. The position-zero era was legible: you could see whether you held the snippet, watch the impressions it drew, and track the clicks it sent, and from that you could reason about cause and effect. An answer composed without a link removes most of those readings at once. You may be feeding a great many answers and see none of it in the places you have always looked, because the surfaces that reported your presence assumed a link existed to count. The uncomfortable truth at the end of this month is that being found is about to get harder to measure at exactly the moment it starts to matter more, and no dashboard yet exists for the part that is growing. The instruments were built for a world of links, and the world is quietly leaving links behind.
The temptation, when a familiar metric goes dark, is to treat the thing it measured as having gone away. That would be the wrong lesson. The traffic a snippet sent may shrink, but the underlying event — your company being chosen as the basis of an answer — is becoming more common, not less, even as it gets harder to see. The honest response is to hold two facts at once: the old measurements are thinning, and the activity they tracked is intensifying. For now that means relying less on a single click-through number and more on whether the answers people get about you are present and accurate, which you can still check by asking the questions your buyers ask and reading what comes back. It is a cruder instrument than a traffic report, and at the close of November it is the honest one, more trustworthy than a number that no longer counts the thing that matters.
What earned the snippet earns the answer
The most reassuring thing to say at a moment like this is also the most practical: the inputs do not change. A featured snippet was awarded to pages that stated the answer plainly and early, that structured information so a machine could extract a clean claim, and that came from a source the engine had reason to trust — the expertise and authority that quality guidelines have rewarded for years, corroborated by other credible sources. Every one of those is exactly what a system composing an answer reaches for, link or no link. It still needs a clear claim to lift, still favours sources it can trust, still leans on corroboration to decide what is safe to assert. The mechanics of delivery are changing; the qualities that make you worth delivering are not, and they are the slowest of all things for a competitor to copy.
That is why the right move at the end of November is continuity rather than reinvention. The company that has been earning position zero has been doing answer-engine work all along without the name, and the company that ignored the snippet because the click went to the searcher will find the same neglect costs it in answers too. Nothing here calls for a separate budget or a new specialism this week; it calls for recognising that the slow disciplines — clarity, structure, earned authority, corroboration — are the through-line from the search you know to the answers that are coming, the one investment that does not have to be rebuilt each time the surface changes, and for keeping at them while everyone else is dazzled by the demo.
As search becomes an answer: quick answers
Hasn’t search been giving direct answers for years already?
Yes, and that is the point worth holding on to at the end of this month. For nearly a decade the results page has been lifting answers out of pages and showing them up top — the featured snippet, the People Also Ask boxes, the knowledge panel. "Position zero" was the name we gave to being the answer rather than a link, and chasing it was already a discipline. So the idea of search handing back an answer instead of a list is not new, and anyone who optimised for the snippet has been practising for it. What is changing is not the existence of the answer but two things around it: who gets credited, and whether the answer can be trusted.
What actually changes, then, if the answer itself is old news?
Two things, and both matter for being found. First, attribution. A featured snippet is extracted from a page and shown with a link back, so being the answer still sent credit and sometimes a click to the source. The newer systems hinted at this month tend to compose an answer in their own words with no link at all, so being the answer can stop crediting you. Second, reliability. A snippet is a quote from a real page; a generated answer is assembled and can be confidently wrong. Both shifts move the ground under the familiar position-zero game without changing its underlying aim, which is still to be the trusted source an answer is built from.
Why bring up the science model that was just withdrawn?
Because it is the clearest preview of the reliability problem, and it happened this month, not in some speculative future. A large science-focused model was put online mid-November and taken down within three days, after researchers showed it produced authoritative-sounding statements that were false and even cited papers that did not exist. That is the failure mode of a composed answer in miniature: fluent, confident, and wrong, with no easy way at all for a reader to tell. It is a useful warning precisely because it is early and undramatic — a reminder that as answers get generated rather than quoted, sounding right and being right come apart, and being found accurately becomes its own concern.
So what should a company do as this shift begins?
Keep doing the position-zero work and understand why it still applies. Being chosen as the answer has always rewarded clarity, structure, and earned authority — saying things plainly, being the kind of source quality guidelines call expert and trustworthy, and being corroborated elsewhere. That is exactly what a system composing an answer leans on too, whether or not it gives you a link. So the honest move at the end of this month is not to panic about a research preview released on the 30th, but to treat the answer-shaped future as continuous with the snippet-shaped present, and to keep being the clear, credible source any answer is safest to build from.
A note on sources and timing
This is written at the end of November 2022. We have described the search landscape as it stood: answers lifted into featured snippets, People Also Ask, and knowledge panels, with links back to their sources, earned through the expertise-and-trust signals quality guidelines reward. We have noted two events from this month as early indicators, not as a settled future: a science-focused model put online mid-month and withdrawn within three days for producing confident falsehoods, and a conversational research preview released on the 30th whose adoption, if any, lies ahead of this writing. We have not predicted what either becomes. The durable point holds regardless: the answer has been the destination for years, the source beneath it is what may change, and being a clear, trusted, corroborated source is the groundwork the AC Group has built for 27 years.