Most of the movement is noise, not signal
Structure and markup refreshed for current answer engines; the original analysis is preserved.
A core update finished rolling out last month, and now the dashboards are full of motion: rankings up, rankings down, every wobble inviting a story. But Google itself says positions are not fixed — they move constantly because the web and its users keep changing. The discipline that matters in January is not reacting to every tremor, but learning to tell the rare real signal from the constant background noise.
the short answer
Most of the daily movement in your rankings is noise, not signal. Google says positions are not fixed — they shift because the web and user expectations keep changing, and Google adjusts its systems constantly. Tell signal from noise by three tests: has the move held over a window, is it broad not isolated, and does it line up with a cause? If not, it’s noise — keep measuring, don’t react.
key takeaways
- A core update finished rolling out last month, so the dashboards are full of motion — and the temptation is to read a story into every up and down.
- Google says positions are not fixed: results are dynamic because the web and user expectations keep changing, and Google adjusts its systems constantly, often without announcing it.
- Most daily movement is noise — the field rearranging around you — not a verdict on your page and not something with a cause you could act on.
- Tell signal from noise by three tests: has the move held over a window, is it broad rather than isolated, and does it line up with a plausible cause?
- Watch trends over weeks, breadth across pages, and aggregate metrics — not a single keyword on a single day, which is the noisiest number you can stare at.
noise vs signal
The wobble on the left is most of what you see day to day, and the right response to it is none. The step on the right is rare — and it is the only kind of movement worth acting on.
The idea, in four parts
Why the dashboards are loud right now; why positions move on their own; the measurement trap of a story for every wobble; and how to read it well. Open each part.
01 Why the dashboards are loud right now
A core update finished rolling out last month, in the middle of December, and one predictable consequence is that a lot of people are staring at their analytics this January trying to read meaning into every movement. That is understandable — after a broad update, rankings really do shift, and it is natural to want to know where you stand. But it also creates the perfect conditions for a particular measurement error: seeing motion everywhere and assuming all of it is meaningful. Some of what people are looking at right now is the genuine, settled effect of the update; a great deal more of it is ordinary fluctuation that would have happened anyway, made to look significant by the heightened attention an update brings. The trap is that an update lowers your threshold for alarm. A two-position dip in late November might have gone unnoticed; the same dip in January, against the backdrop of an update everyone is discussing, gets read as evidence that you were hit. The starting point for measuring well in a month like this is to remember something Google has stated directly: positions in its results are not static or fixed in place. They were never meant to hold still. Your ranking is not a possession that sits where you left it until you or an update moves it; it is a reading taken from a system that is always in motion. So before attributing this month’s movements to the update, or to anything, the first discipline is to separate the rare signal that an update genuinely produced from the constant noise that was always going to be there — because mistaking the second for the first is how a quiet month of normal variance turns into a panic.
02 Why positions move on their own
It helps to understand why a ranking fluctuates even when you have changed nothing, because once the mechanism is clear the daily wobble stops feeling like a series of events. Your position is not an absolute score; it is a placement relative to every other page competing for the same query, in an environment that never stops changing. Google has described its results as dynamic by nature, and named the reasons: user expectations evolve, and the open web is constantly changing with new and updated content. On top of that, Google is continually adjusting its own systems — it has said it makes updates all the time, including smaller core updates it does not announce precisely because they are not widely noticeable. Put those together and a single day can move your ranking for reasons that have nothing to do with you: a competitor improved their page and edged ahead, fresh pages entered the index and reshuffled the order, seasonal demand for a query shifted what searchers want, or one of those unannounced adjustments quietly landed. None of these is a penalty, a reward, or a message; each is just the field rearranging around a page that did not itself move. This is why two sites can both “do nothing” in a given week and both see their rankings drift — in opposite directions — with neither outcome carrying any lesson. The motion is real, but it is mostly relative motion in a crowd, not a judgement on the page standing still within it. Seeing it that way is the difference between reading your data and being jerked around by it.
03 The measurement trap: a story for every wobble
The core error this month, and most months, is the human reflex to attach a cause to every movement — to look at a dip and immediately ask what did I do wrong, or what did the update do to me, as if every number had to have a story. It is a deep bias: we are pattern-finders, uncomfortable with randomness, and a chart of fluctuating positions is an irresistible invitation to narrate. The danger is not the narrating itself but what it leads to. Once you have convinced yourself that a normal wobble was caused by something — a change you made, a page you published, an update that hit you — you feel compelled to act on that story, and the action is usually a quick fix made in something close to panic: stripping out a page element you decided was the culprit, hastily rewriting content that was fine, undoing a change that had nothing to do with the dip. These reactions are not just wasted effort; they actively introduce instability, because now you really have changed something, and the next wobble — which was coming regardless — gets blamed on the fix or its reversal, and you are off chasing your own tail through a series of changes that each respond to noise. The deeper failure is one of measurement: treating every fluctuation as signal means you can never distinguish the movements that matter from the ones that do not, because by definition you are reacting to all of them equally. A measurement practice that responds to everything is responding to nothing in particular. The skill is the opposite of vigilance-as-reaction — it is the discipline to watch a great deal of movement and act on almost none of it, holding still until something crosses the threshold from noise into signal.
04 How to read it well
Reading fluctuation well comes down to a few habits that all push in the same direction: look at the resolution where signal lives, not the one where noise dominates. First, set a baseline and judge against it rather than against yesterday — where did these pages sit over a stable stretch before whatever you are investigating, so you have a level to compare to instead of a single prior day. Second, require a move to hold. A position that drops and stays down across a window of weeks is telling you something; a position that drops and drifts back is telling you nothing except that positions drift. Third, look at breadth: a single query on a single page moving is almost always noise, while a whole cluster of pages shifting together in the same direction at the same time is the fingerprint of a systemic change, because that simultaneity is exactly what random fluctuation cannot manufacture. Fourth, prefer aggregate measures — total clicks and impressions for a section over weeks, average position across a group of queries — because aggregation cancels out the random daily motion and lets a genuine trend surface. And annotate the dates of known updates on your charts, so that when something real does coincide with one, you can see the correlation instead of guessing at it. The throughline is patience. Google has said that even improvements you make can take time to show, sometimes months, before its systems confirm them — so neither alarm nor relief should be drawn from a single early reading. Watch the trend, demand breadth and duration before you believe a move, and act only on what survives those tests. Measuring at that resolution, so your effort answers signal and ignores noise, is the discipline the AC Group has worked by for 27 years.
Why this is a measurement problem, not an SEO one
Notice that nothing in this is really about search tactics; it is about how to read a noisy signal without fooling yourself, which is a problem every field that measures anything has had to solve. A position chart is a time series full of random variation around a slowly changing level, and the standard mistake everywhere such data appears is to over-interpret the variation — to see a meaningful event in what is just the spread of normal values. Scientists guard against it with baselines and significance thresholds; the discipline here is the same instinct in plainer clothes. The reason it matters so much in search specifically is that the data updates daily and sits right in front of you, so the temptation to react is constant and immediate, and the cost of reacting wrongly — panic edits that destabilise a page that was fine — is real.
That is why we treat measurement as a discipline in its own right rather than an afterthought to the work. Knowing what not to react to is most of the skill: a baseline to compare against, a threshold a move must clear before it earns a response, the patience to let a number resolve rather than chasing it. These are not search techniques; they are the habits that let you act on what is real and ignore what is not, in search as anywhere else. Building that measurement discipline, so your effort follows signal instead of noise, is part of the work the AC Group has done for 27 years.
What to do with this
This month, resist the pull to act on what the post-update dashboards are showing until it has settled and proven itself. Set a baseline from a stable stretch before the update, then ask of any drop you are worried about: has it held across a window of weeks, is it broad across many pages rather than one query on one page, and does it line up with a cause you can actually name? A move that fails those tests is noise, and the correct response to noise is to keep measuring and do nothing — not because doing nothing is easy, but because acting on noise is how you manufacture the next problem.
When a move does pass the tests — held, broad, and plausibly caused — then respond to it in proportion, and for a core update that means the patient work of improving the substance of the affected pages, not a quick technical fix aimed at the update, and not an expectation of an instant rebound, since recovery can take until the next core update to show. Above all, watch your data at the resolution where signal actually lives: trends over weeks, breadth across pages, aggregate clicks and impressions rather than a single keyword on a single day. Measuring at that resolution, so you answer signal and ignore noise, is the discipline the AC Group has worked by for 27 years.
Reading the movement, plainly: quick answers
Why do my rankings move when I haven’t changed anything?
Because your position was never a fixed property of your page; it is a relative placement in a field that keeps moving even when you stand still. Google has said this plainly: positions in its results are not static or fixed, because its results are dynamic by nature — user expectations evolve, the open web constantly changes with new and updated content, and Google itself is continually adjusting its systems, including small core updates it does not announce because they are not widely noticeable. So on any given day your ranking can shift without you touching a thing, simply because a competitor improved their page, new pages entered the index, demand for a query changed, or one of those quiet algorithm adjustments landed. None of that is a verdict on you; it is the normal weather of a living system. The mistake is to treat your ranking as a still object that should only move when you move it, and therefore to go hunting for a cause every time it does. Most of the day-to-day motion has no cause you could act on, because it is not about you at all — it is the field rearranging around you. Once you accept that positions breathe in and out naturally, the daily wobble stops looking like a series of events demanding explanation and starts looking like what it is: background noise, against which only occasionally a real signal appears.
How do I tell a real drop from normal fluctuation?
You compare against a baseline over a window of time rather than reacting to a single day, and you look at breadth. Normal fluctuation looks like a position oscillating within a band — up a little, down a little, around a steady centre — over days and weeks, with no sustained direction; it is noise around a level. A real drop looks different: a clear, sustained step down that holds across a meaningful window and does not drift back, ideally lining up with something you can name, like the dates of a known update or a change you made. Breadth is the other tell. If one query on one page slipped, that is almost certainly noise; if a whole cluster of pages moved together in the same direction at the same time, that is far more likely to be signal, because broad simultaneous movement is what a systemic change produces and random fluctuation does not. So the practical test has three parts: has the move held for long enough to be more than a wobble, is it broad rather than isolated, and does it coincide with a plausible cause? If you cannot answer yes to at least the first two, you are most likely looking at noise, and the correct response to noise is to do nothing — to keep measuring and let the picture resolve rather than reacting to a number that will probably drift back on its own.
A core update just finished — should I act on what I see now?
Not on the first thing you see, because the period right after a core update is the noisiest time to read your data, not the clearest. Rankings can keep settling for a while as the update finishes propagating, so an early reading often shows movement that has not stabilised, and acting on it means responding to a number still in motion. Give it time to settle, then assess against a baseline: where were these pages before the update announced, where are they now that it has clearly finished, and is the change broad and sustained or scattered and drifting? If you confirm a genuine, held drop across a meaningful set of pages, then the response is the patient, holistic work a core update calls for — improving the substance and quality of the affected content — not a quick technical fix aimed at the update itself, because core updates are not about page-level tricks you can reverse. And crucially, recovery from a core update typically does not arrive the moment you make changes; Google has indicated improvements can take time to be reflected, often only confirmed at the next core update as its systems re-assess. So the worst move after an update is to panic-edit based on a noisy early reading and then panic again when the edits do not produce an instant rebound. Measure first, confirm the signal, improve the substance, and give it the months it may genuinely take. Reading your data with that patience — separating the settling noise from the real signal — is the measurement discipline the AC Group has practised for 27 years.
What should I actually watch instead of daily rankings?
Watch trends over weeks and months, watch breadth across pages and queries, and watch the metrics that aggregate away the daily noise. A single keyword’s position on a single day is the noisiest number you can stare at, and staring at it is how people talk themselves into pointless changes. More useful is the shape over time: is a page or a cluster trending up, flat, or down across a window long enough to mean something? More useful still is to segment — to ask whether a move is concentrated in one corner of the site or spread across it, because that distinction is what separates an isolated wobble from a systemic shift. And aggregate measures — total clicks and impressions for a section over weeks, average position for a group of queries rather than one — smooth out the random day-to-day motion and let a real trend show through. The point is not to ignore your data; it is to look at it at the resolution where signal lives, which is rarely the daily, single-keyword resolution where noise dominates. Set a baseline, annotate the dates of known updates so you can correlate real events, and judge movement against that frame rather than against yesterday. Measuring at the right resolution, so you respond to signal and ignore noise, is the discipline that keeps you from chasing your own tail — and it is the work the AC Group has done for 27 years.
A note on sources and timing
This is written in January 2021, a few weeks after the December 2020 core update finished rolling out. The framing rests on Google’s own description of how its results behave — that positions are not static or fixed, that results are dynamic because user expectations evolve and the open web constantly changes, and that Google continually makes updates including small, unannounced ones — along with its guidance that improvements can take time, sometimes months, to be reflected as its systems re-assess. The advice to judge moves by duration, breadth, and a plausible cause, and to prefer baselines and aggregates over single-day, single-keyword readings, is the ordinary craft of reading a noisy time series, applied to search. The durable point outlasts any one update: most of the movement is noise, and the skill is telling the rare signal from it — the measurement discipline the AC Group has practised for 27 years.