Why a Standing Desk Shakes

The founding charter

The first purchase in any sit-stand build isn't the chair or the keyboard — it's the desk itself, and the order matters. A desk is the only piece of gear that can put your typing surface at elbow height for two different bodies, sitting and standing, and everything else in a setup — monitor arm, cables, even a walking pad underfoot — hangs off it. Buy the desk last and you're stuck working around decisions already locked in; buy it first and everything after becomes a choice.

Deliver a sit-stand desk beat

At sitting height nearly every powered desk feels solid — that's just physics, since a low, collapsed frame is squat and stable. The real test comes at the top of the travel, where a desk turns tall and thin and starts trading stability for height. Four things decide how much it wobbles once it gets there: whether a steel crossbar ties the two leg columns together low across the frame, how many telescoping stages each column has (more stages buy height range but stack up play in the sliding joints), how deep the steel feet reach front-to-back, and how heavy and well-anchored the desktop itself is.

Side-to-side sway and front-to-back rock are different failures with different causes — a desk can pass one test and fail the other, so any check has to push the top sideways and also type a hard line while watching the screen, always at full standing height with real gear loaded on, never at sitting height.

Named examples anchor the spec talk: the Uplift V3 covers the widest height range and carries independent stability certification; the FlexiSpot E7 is the crossbar-braced, three-stage benchmark most competitors get measured against for value; the ApexDesk Elite Pro shows a case where the frame is fine but a wide, light desktop is doing the flexing; and a single-motor, two-stage budget desk illustrates the simplest failure of all — a top-of-range height that never reaches a tall user's standing elbow height in the first place.

The rest of the sheet is downstream of that: dual motors beat a single shared motor, lift capacity should be read after subtracting the desktop's own weight, memory presets and adjustment speed decide whether the desk actually gets used standing, cable slack has to survive fifteen to twenty-five inches of travel, and any warranty should be read by its shortest-covered part — usually the motor and controller, not the steel frame.

By now you should have a number written down somewhere you can find it again — the height from your floor to the underside of your elbow, taken sitting, with your shoulders loose. Maybe you have the standing one too. Keep it handy, because this is the episode where that number starts spending money.

The first purchase in this build is the surface. Not the chair, not the keyboard, not the second monitor. The desk. There are two reasons for that, and they are worth being exact about, because the order you buy in decides how much of your budget does any work at all.

The first reason is that the desk is the only piece of gear that can put your typing surface at your elbow height for two different bodies — the sitting version of you and the standing version of you. Those two heights are nowhere near each other. Everything else you might buy is downstream of that. A beautiful keyboard on a surface that is three inches too high is still a surface that is three inches too high, and the keyboard cannot fix it, because the keyboard is sitting on the problem.

The second reason is structural rather than ergonomic. Everything that comes later in this build hangs off the desk. The monitor arm clamps to the edge of it. The cables run under it. The walking pad goes beneath it, which raises the floor you stand on and therefore raises the height the surface has to reach. If you buy the surface last, you buy it with constraints you have already locked in. If you buy it first, everything after it is a choice rather than a workaround.

So: the desk. And here is where I want to narrow this down hard, because standing desk listings are a wall of numbers and most of those numbers will not decide anything for you.

At sitting height, almost every powered sit-stand desk on the market is fine. That is not a compliment and it is not a dodge. It is a description of the physics. A desk at sitting height is a short, squat object with its weight low down and its telescoping legs mostly collapsed into themselves. Squat objects are stable. You can put a cheap desk and an expensive desk side by side at twenty-nine inches, lean on both, and struggle to tell them apart. People do exactly this in showrooms and walk out having learned nothing.

The difference between a good sit-stand desk and a bad one lives entirely at the top of the travel. That is the axis of this whole episode. When a desk climbs from sitting height up to somewhere in the high forties, it stops being squat. It becomes tall and thin, with its mass held up in the air on two extended columns. And the further it extends, the less stable it gets — not a little less, and not in a straight line either. A desk is a machine that trades stability for height, and it trades away more of it the higher you ask it to go.

That single sentence is the frame for every specification on the page. Once you understand that a rising desk is spending stability, you can read the spec sheet as a ledger. A crossbar buys stability back. More column stages spend it. A deep steel foot buys some. A big thin desktop spends some. Every line item is on one side or the other of that trade, and the wobble you feel at standing height is the balance at the bottom of the ledger.

Two pieces of housekeeping before the mechanics, and then I will not interrupt again. The review tables on the site earn a commission on the links they carry, and a commission has never bought a pick — the picks come from what the house measured, and the ranking would be identical if every link on the site paid nothing. And this is a gear show, not a clinic. I will tell you what a change does to the mechanics of a desk and what owners report about living with one. I am not going to tell you what it will do to your body, because I cannot know that, and nothing here prevents, treats or cures an injury. If something has been hurting for weeks, that is a conversation with a professional, not a shopping decision.

Now: why does a raised desk shake, and what can you do about it before you pay?

The four things that decide how much a raised desk moves

Start with the lever, because everything else is a variation on it.

Hold a broom by the very bottom of the handle and give the bristles a tiny nudge sideways. The bristles travel a long way. Now grip the handle halfway up and nudge again — much less movement, same nudge. That is your desk. The legs are clamped to the floor at the bottom, and the desktop is the far end of the handle. A small amount of looseness anywhere down in the frame becomes a visible sway up at the surface, and the amount it becomes depends on how far up the surface is. This is why the same desk that felt like a rock at sitting height can feel like a card table at standing height, with nothing having changed about it except its geometry.

Four things decide how much sway you actually get.

The first is whether there is a crossbar. A crossbar is a steel bar running low across the frame, tying the two leg columns together. Without one, the two legs are effectively two separate posts, each one standing there resisting sideways force on its own, joined only up at the top by the frame under the desktop. Push the desk sideways and the legs can lean independently, which means the frame can lozenge — go from a rectangle to a parallelogram — and the top swings. Bolt a bar across low down and that stops being possible, because now any lean in one leg has to drag the other leg with it. It is the same reason a garden gate gets a diagonal brace nailed across it: on its own, a four-sided frame of timber is a hinge waiting to happen, and one member across it turns four floppy joints into a rigid shape. Of everything on a desk spec sheet, the presence or absence of that bar is the single biggest lever on side-to-side sway, and it is also the one most likely to be quietly missing from the minimalist-looking frames, because a crossbar is exactly the part a designer removes to make a desk look clean and leave room for your knees.

The second is column stages. A desk leg is a telescope. A two-stage column is two tubes, one sliding inside the other — one sliding joint. A three-stage column is three tubes, so two sliding joints per leg. Inside each joint there are plastic or composite glide shims, little bearing surfaces that let the steel slide smoothly instead of grinding. Those shims cannot be a perfect fit. If they were, the leg would bind and the motor would stall. So each joint is built with a deliberate hair of clearance, and that clearance is play.

Here is the part nobody explains. Play does not average out across joints. It adds. Two sloppy joints stacked on top of each other give you the first one's wiggle plus the second one's wiggle, and then the lever multiplies the total. Worse, the amount of overlap between the tubes is what stiffens a joint, and overlap is at its absolute minimum exactly when the desk is at full height — at the moment you most need the leg to be rigid, it has the least tube inside tube holding it straight.

So three stages is worse than two? Not that simple, and this is the trade the listings never lay out. Three stages is what buys you the wide height range. Three-stage columns are how a desk gets down near or below twenty-four inches for a short person and up toward fifty for a tall one, and they collapse shorter, so the desk is lower when it is down. You are not choosing between stability and nothing; you are choosing between stability and reach. A well-machined three-stage column with tight shims will beat a sloppy two-stage every day of the week. A cheap three-stage, with two loose joints per leg, will be worse than either. Stages tell you where the risk lives. They do not tell you the outcome.

The third is the feet. Look at the steel footings the columns stand on and ask how far they reach forward and backward. That reach is your base of support against tipping. When you lean on the front edge of a desk, you are trying to rotate the whole thing forward around its front feet, and the only thing arguing back is how much foot there is behind the pivot. A deep foot has plenty of argument. A short, stubby foot has almost none, and the desk answers a forearm on the front edge with a nod. Related: whether the column sits centred over the foot or offset toward the back, which changes how much of that foot is in front of the load and how much is behind it.

The fourth is the desktop itself, and this one gets ignored constantly because it looks like a matter of taste. Mass matters. A heavy top — thick butcher block, solid hardwood — is slow. It has inertia, so a sharp keystroke does not have enough energy in it to get the top moving, and any movement that does start dies quickly. A light, wide, thin laminate top is the opposite: it is a drum skin. It flutters under fast typing, and a wide one flexes in the middle where nothing is holding it.

And how the top is attached matters as much as what it is. A frame that bolts up into threaded metal inserts, or spreads its load along wide steel rails running most of the length of the top, is one object with the frame. A top held on by a handful of wood screws driven into thin particle core, on two small brackets, is a separate object resting on the frame and hoping. That joint can rotate a little, and a little rotation up at the surface is a lot of visible movement at the far edge.

Now the thing I most want you to take away from the mechanics: side-to-side sway and front-to-back rock are two different failures with two different causes, and a desk can pass one and fail the other.

Side-to-side sway is the frame lozenging — shear, driven by weight shifts and by lopsided typing, and it is mostly a crossbar-and-column-play problem. Front-to-back rock is the desk pitching toward and away from you, driven by your wrists loading the front edge, by heavy keystrokes, by resting forearms on the front third of the top, and by a monitor riding on an articulating arm that has its own swing. That one is mostly a foot-depth and desktop-attachment problem. A desk with a solid crossbar and shallow feet will hold rock-steady against a sideways shove and still bob at you all day. So when you test, you have to test both, and when you feel wobble, the useful question is not how much but which way.

Which brings us to the test, and you can run this tonight on the desk you already own, or in a showroom on one you are thinking about.

Raise it to your full standing height — your actual number, not a comfortable middle setting — with everything on it that normally lives on it. Monitors, laptop, the lot. An empty desk tests nothing, because the mass on top is part of the machine.

Then, before you touch the frame, get down and check the leveling glides at the bottom of each foot. If one of them is not making solid contact with the floor, you have found your wobble and it is free to fix. There is no point diagnosing a frame that is standing on three feet.

Then take hold of the top of the outer leg sleeve and shake it. You are feeling for play in the sliding joints — a small dead zone of movement before the leg starts to resist. That is the shims talking.

Then put the heel of one hand against the side edge of the desktop and push sideways — a firm, gradual push, ten to fifteen pounds, not a slap. Watch two things. How far it goes, and, more importantly, what happens when you let go. A stiff frame stops when your hand stops. A loose one keeps going, swinging back past centre once or twice before it settles. That continued oscillation after the push is the tell, and it is the thing you will notice at four in the afternoon when your monitor is drifting in and out of focus and you cannot work out why.

Then type. Not gently — type a hard line the way you actually type when you are annoyed, hands on the front third of the surface, and watch the screen rather than your hands. Tap your heels firmly on the floor while you do it. If the screen swims, or the water in a glass on the desk ripples, you have front-to-back rock and you now know which of the four causes to go looking for.

And shaking a desk at sitting height tells you nothing at all. If you take one habit from this episode, take that one.

Once you know what you are feeling for, the house desk table stops being a wall of numbers. There is a column on it called wobble, and it carries the house's own stability score rather than anything a manufacturer said about itself — stability is one of those figures nobody self-reports honestly. Read the score the way the table reads it: higher is better, and on a quality column like that one a bare number is a floor. Ask for a wobble score of eight and you get everything at eight and above. On a cost column like price it works the other way round, as a ceiling: ask for six hundred and you get everything at six hundred and under. You can also give it a proper range, a minimum and a maximum with a dash between them, and leave either end off to make it open-ended.

The page is at ocdevel dot com slash ergo slash desks, typed straight in — there is no slash ergo landing page above it, so do not go looking for one, and the exact link is in the show notes. To filter, put a question mark after the address and then filter underscore wobble equals eight. For the affordable end, filter underscore price. Stack them with an ampersand between and both apply at once. And the two columns worth adding to any wobble filter are the two structural ones from the mechanics — filter underscore crossbar, and filter underscore column underscore stages. Those do not restate the wobble score. They explain it. A desk with a high stability score and a crossbar is stable for a reason you now understand; a desk with a high score and no crossbar is a frame doing something clever elsewhere, and worth reading the writeup on.

One warning about that syntax, because it fails in the nastiest possible way. A column name is spelled in lower case with underscores, exactly as the table spells it. Get the case wrong or the spelling wrong and the filter is dropped silently. No error, no empty result, no warning of any kind — you get the full unfiltered page and it looks exactly like a page that filtered fine and matched everything. So if a filter comes back suspiciously roomy, suspect your spelling before you suspect the catalog.

The desk table is one of the two families on the site with a full written review behind every single product, which means I can actually name names here. Everything that follows is where things stood at the end of August, twenty twenty-six; check the table for the live answer, because prices in this category move constantly.

The stable-and-expensive case is the Uplift V3. C-frame, three-stage columns, a crossbar, dual German motors, and certified to the current version of the industry stability standard for desks and tables — the one that puts a frame through static load, racking and drop testing at an independent lab rather than in a marketing department. It travels from about twenty-two and a half inches up to just under forty-nine, which is the widest useful span here and covers both of the groups that usually get designed out. It also moves at about two inches a second, which sounds trivial and is not, for a reason I will come back to.

The stable-for-the-money case is the FlexiSpot E7. T-frame with a crossbar, three-stage columns, dual motors, three hundred and fifty-five pounds of lift capacity, the same stability certification, and a range from around twenty-four inches to a shade over forty-nine. It is the benchmark most of the field gets measured against, and the house calls it the best overall value in the family.

For the third case — where the frame is not the problem — look at the ApexDesk Elite Pro. Dual motors, a T-frame, a crossbar, all the structural pieces present and correct. But it is a sixty by twenty-seven inch laminated engineered-wood top, and that is a lot of wide, light surface. If you meet a desk like that and it bobs when you type, the frame is not what you are feeling. You are feeling the top: the mass of it, the flex across that span, and how well it is fastened down along its length. This is worth knowing because it changes what you do about it. You cannot fix a floppy frame, but a top is a part, and how it is bolted to the rails is something you can go and look at underneath your own desk this evening.

The counter-example at the cheap end is instructive too. The AmazonBasics electric desk is a single motor and two-stage columns, and its top end stops just under forty-six inches. That number is the one to hold against your own standing figure before you consider anything else about it, because a desk that cannot reach your standing elbow height is not a stable desk or an unstable desk. It is the wrong desk, and no other specification on the row can rescue that.

Which is why height range at both ends is the first column to check, ahead even of wobble. Not the range as a headline — the two ends of it, separately, against the two numbers you wrote down. The bottom end is what the short get designed out of; the top end is what the tall get designed out of. And if a walking pad is going under there eventually, that pad raises your standing floor by several inches, and every one of them comes off the top of the desk's range.

The rest of the spec sheet, briefly, because it is all downstream of what we have already covered. Dual motors — one per leg — are the standard on a serious desk; they lift more, they lift more evenly, and a single motor driving both columns through a shaft is where you find slower travel and more noise. Lift capacity is quoted including the desktop's own weight, so subtract the top before you decide whether three hundred pounds is generous: two large monitors, arms, a machine, a laptop, a lamp and the usual sediment add up faster than people expect, and the number is worth having as margin rather than as a target.

Memory presets and adjustment speed sound like conveniences and are actually the difference between a sit-stand desk and an expensive fixed desk. A desk you have to hold a button on for twenty seconds is a desk you will stop raising by the second week. Four presets is the useful count — sitting, standing, and two more for whatever else you do. Speed matters for the same reason: at two inches a second the change is a shrug, and at half that it is a decision.

Cable management has to survive fifteen to twenty-five inches of travel, which is the part people discover late. Every cable on that desk has to have that much slack, routed so it takes up and pays out cleanly instead of going taut at full height and pulling a plug half out of the wall. And assembly is worth reading honestly: a desk frame is a heavy, awkward object, most of these take somewhere between twenty and forty minutes, and the step where the top goes on wants a second person, whatever the box says.

Last, the warranty, which is where the pattern from earlier in this show turns up again. Read it by its weakest component, never by its headline. A fifteen-year warranty on a desk almost always means fifteen years on the steel frame, with the motors, controller and keypad covered for a much shorter term — and the electronics are the part that actually fails. So the working number is the short one. Same discipline with a "starting at" price: that figure is generally the smallest frame in the range, and often the frame alone, with no desktop on it at all. The desk you actually want, at the size and material you actually want, is a different number.

So, tonight. Raise your desk to your real standing height with your gear on it, push the side edge with the heel of your hand, then type a hard line and watch the screen. Write down which way it moved — sideways, or toward you and away. Then open the desk table and filter it by wobble before you look at a single other column.