If you landed here directly, check out the first two posts of this series of - The complete industrial designer:
Stage 1: Research & observation
Stage 2: Sketching
I have about four to five bottles from different brands, they have about the same volume of 720ml, and all of them roughly have the same weight. One of them looks expensive from across the room and the other three just looked like bottles. The expensive-looking one wasn't the expensive one actually. It was a differently shaped Mijia/Xiaomi Tritan series bottle.
What catches the eye
Three things, more or less in this order.
How slim it looks, which is just height against width. I wrote a blog last year on car design which has similar underlying concept. In a bottle, the ratio of height to width matters. 3:1 would look like a cylindrical container but make it slightly taller, it'll feel like an object. If you make it taller, it'll probably look like a tube than a bottle.
This slim profile also goes hand-in-hand with where the weight of the bottle sits, which you should work out by finding the widest point and checking how far up it is.
Same diameter for most of the height and then a fast shoulder looks industrial. A slow taper starting low looks soft. A slight belly down in the lower third looks like a vessel, and that's why old brass water pots still look better than most of what's on a typical bottles shelf.
And how much of the whole thing the cap takes. A cap at a fifth of the height looks like a lid sitting on a tube. A cap at a seventh or less looks like part of one object.
That third one matters more now than it did, because a click-open cap has a button and a hinge and a spring inside it and every one of those needs height.
Volume of water picks the diameter and the diameter picks everything
An 80mm body puts about 199mm of water in the tube, and with the neck and headspace and cap it comes out somewhere near 270 tall, which is about three and a half to one. This is a good ratio to work with as it clears most fridge doors.
Take it to 74mm to make it slimmer and the same volume of bottle goes past 300mm. Now it's over four to one, it goes over on a car seat, and it stops standing in a fridge door, which is often only about 200mm of clear height. Slimmer looks better in a drawing but becomes an irritant everywhere else.
Go the other way to 86 or 88 and the thing gets fatter and the hand problem gets worse.
So somewhere around 78 to 82 is the window, and everything after this should be worked out inside it.
The grip and proportion
At 80mm across, nobody grips this bottle. A hand closes comfortably round something roughly 30 to 50mm and at 80 the fingers are spread and the palm is flat against the side, which is visibly okay but becomes non-ideal when walking down a corridor.
Probably look at the waist of the bottle; pulling the middle of the body in to maybe 65 or 68mm so the hand can grip it neatly. Now, that will compromise volume, which has to be adjusted in height, and it puts a curve into a clear body where the wall thickness has to stay even or the optics go weird.
Or a loop, which is what Nalgene did, so the hand carries it by a finger rather than by the body. So volume is not compromised but the cap will have a prominent feature.
Draw both and probably come up with a shallow waist for holding and a much better loop for carrying, because they're solving different moments of bottle handling.
The water line is also part of the silhouette
An opaque bottle has one outline all day, right? A clear bottle has three, because the water line is a hard horizontal line that cuts the form, and it keeps moving. When the bottle is full, it reads as one solid volume. Half bottle will read as two stacked blocks with a bright edge between them. In an empty bottle you mostly see the cap and the base and whatever coloured parts you'd be giving it.
Which means the coloured parts will carry the identity for a good part of the day, and the empty state is the one you should design first, because that's the version with the least going on and the one that has to still look like something interesting.
So draw all three levels before committing to anything.
Fill it in black to test the outline
Everything you'd like about the surfaces disappears in the black fill and what's left is the outline, and the outline is what somebody sees from two metres away, which is where the decision to pick it up happens. If the black shape isn't interesting then no amount of detail later is going to save it.
This matters more than usual here because of the differentiation problem. Muji and Xiaomi look alike from six feet. Both of them are a clear cylinder with a white cap. Their identities are cheap to copy, so a year later somebody else would have it too.
The Stanley looks different from a distance because the banding cuts it into three zones, which is stronger, and it's still copyable.
The thing nobody can copy in a season is the outline itself. I mean, they can but the identity would be tightly locked with the brand like the Stanley quencher. So if the black fill is distinguishable at matched height then we are onto something, and the distinguishable property can't be just in colour.
Then do a squint eye test to see which one stands out.
Where the bands fall
A clear body with coloured parts has an identity made of horizontal and vertical divisions. The Xiaomi Tritan series water bottles are great examples of these with the accented green as the infuser inside the frosted white body.
The orange sleeve at the base of Stanley flask is doing three jobs at once, which is why I like it. It's the identity band, it protects the base which gets dirtiest the most, and it takes the condensation ring off given it'll be full of cold or hot water too.
So the height of that sleeve is a real decision. Low and thin looks technical and leaves the water visible. Up to a third and it starts looking as a two-part object and you lose some of the visible-water job that transparency was doing in the first place.
I'd start around a fifth of the body height and test it against the empty state, because empty bottle is where the sleeve is doing the most work and too much of the colour will look like a mistake.
How the bottle stands
Three things decide whether it looks planted or looks like somebody's about to knock it over.
Base width against the widest point, so if it bulges above a narrow base it reads top-heavy even when it isn't. The shadow gap, where about a millimetre of recessed rim sits the thing on a ring and throws a thin dark line that makes it look as placed rather than resting. And the base radius, where something tight like a thin groove would look precise and something soft like a wide recessed band will feel a bit cheap.
Reminds me of a principle by Dieter Rams that good design is unobtrusive. His products were tools and he wanted them out of the way, the way a good butler is.
A flame orange base is the opposite of that and Rams would not have made this bottle. But see, his objects mostly lived in one room and belonged to one person. This one goes in a bag and onto a shared desk and into a house where we want multiple people to own similar bottles. Being identifiable across a room is the problem statement for us so the colour is doing work rather than decorating.
We should not go overboard as well. If the base is loud then the cap and the button and the sleeve edge and the loop all have to be subtle, because we have to spend the boldness once. A bottle with an orange base and a shaped cap and a contrast button and a logo is going to look like flee-market knock-off product.
Taking things out
Print the current design and go through it feature by feature asking what happens if this isn't here.
Even if the answer is nothing, still take them out and look again, and maybe half the time the object gets better and you won't be able to articulate why, and the reason is that the eye now has one thing to look at instead of four.
Some practice along the way
Take ten objects you love, with a vernier and a notebook. Write down height, width, every radius, every gap, the height of any secondary part against the whole. Write down the ratios and do about ten and you'll start carrying proportion around in your head, which is what people mean when they say somebody has an eye.
Draw things from memory and then measure the real one, then see the difference because that difference of the error will tell you which dimension you weren't seeing.
And the shop exercise too. Go to a super mart and look at aisles of bottles and compare the cheapest and dearest side by side. Ignore the brand and find ten proportional differences.
Then a library of a couple of hundred images with a reason written on every one. Don't write nice. Write a number or a mechanism, so the cap is about an eighth of the height, or the sleeve stops at a fifth, or a millimetre of base groove throws a shadow line. An inspiration library without annotations is a mood board, and a mood board is pretty useless for decision making.
Where should the project be by now
Diameter is chosen within the 78 to 82 (ballpark) window with the height required for a litre of water, and both written down with the reasons next to them.
The grip is closed, as a waist or a loop or both, with a hand outline drawn over the design at full size. Cap height details are closed too. The decisions should lead to a design where nothing can be removed anymore.
Then black-fill it at full and half and empty and stand all three next to the black fills of Muji and Xiaomi and the Stanley at matched height. If ours isn't standing apart in silhouette then the job is still incomplete.