If you A/B test that against any improved design, for example one which replaces the subtle "get started with Tarsnap" link with an attractive 200x60 button, the graphical alternative will annihilate the performance of your minimalist alternative.

If you don't have the time, skill, or desire to do this, there is an Internet full of starving web designers out there. I got my current website design from a nice young lady in India for $150. With the degree of improvement you'll see, that would probably pay itself back within a few days to a week.

Touche! To be fair, it's not just a fashion statement, though -- it's also a matter of branding.

... improved design, for example one which replaces the subtle "get started with Tarsnap" link with an attractive 200x60 button...

Good idea. One question comes to mind, though: Is increasing the attractiveness of signup links as effective for commercial services as it is for free account-creation? I don't dispute that it would help -- I just wonder if it would give me a large number of new accounts but only a far smaller number of new paying users. (Only one way to find out for certain, of course...)

If you don't have the time, skill, or desire to do this, there is an Internet full of starving web designers out there.

I have a few friends who are interested in graphics and web design, so I'll throw this in their direction and see what they can come up with -- at least now I have a basic structure for the content to get started with.

Thanks for the advice!

Ask a simple question, get a simple answer: Yes.

Still simply does not explain to a typical user why they're any better off with you than with Mozy. In fact, even for technical users, Mozy's security page might sound better than yours.

Lose "does not support Windows". Nobody in your target market cares, but it sends a "this is half-baked" message --- which is obviously not justified.

I'd rather be honest than rich. :-)

Still simply does not explain to a typical user why they're any better off with you than with Mozy.

True -- but I don't want typical users. Let's face it: Tarsnap is a computer geek's backup system. The design principles are security, flexibility, efficiency, and utility -- not ease of use or eye candy.

A year or more from now, once tarsnap has a GUI and supports windows, and once I'm shipping precompiled binaries? That's when I'll want to compete with Mozy. Until then, there's no point trying to attract users who won't be happy with tarsnap.

even for technical users, Mozy's security page might sound better than yours

This is probably a dumb question, but... Mozy has a security page? I can't find one.

Lose "does not support Windows"

I'll think about it. I put that in mostly because I get too many emails already from people asking if tarsnap runs on Windows (it's nice to know people are interested, but replying to such emails is a time sink) -- but I see your point that it doesn't necessarily give a good first impression.

Otherwise, I've read your code and watched you work on this for awhile now and you're just doing yourself a disservice.

If it's just a hobby, then no, you don't need fancy graphics or much in the way of an interface or anything else. It's your project, and you're sharing it with people. That's great, and you get to let your personal style preferences rule the project.

But if it's a business, then you have to run it like one. Even if your target market isn't a bunch of point-and-click Windows users, the site needs to be professionally designed, which includes adding some graphics, paying attention to typography, getting rid of the horrible black borders around everything, etc. You're a damn good programmer, from what little I know of you, but that doesn't mean you have any skill at graphic or interface design. If you want this thing to pay the bills, or even do better than that, then you have to sell it, and at the moment ... I don't think the new design is cutting it, even though it's a hundred times better than the current "design".

Yet another way to look at it is that your time would be much better spent working on tarsnap's infrastructure and code, which you're very good at, rather than developing a new layout for the website ... which you're mostly only OK at.

There is a distinction between honest and forthright. He's saying that you're screaming it. Not screaming it is NOT dishonest. Having an attractive web site that could conceivably belong to a company with a big team is not dishonest.

I see what you're going for and it might work well for a very specific audience. But I think it will be hard to reach people beyond those that "get" where this bare bones design stands for.

One thing that could be better in my opinion is answering some very obvious questions directly on the homepage like

- How does it exactly differ from other backup services out there?

- Who would use this and who wouldn't?

- A nice example of cost breakdown for a common setup

- Is there any one successfully using this?

- It feels like a one-guy shop. Should I trust that for my backups?

Then, some random opinionated observations about your design choices:

- It will most likely only attract a very specific audience, is that the goal?

- All the information is very subdivided into lots of pages, lists and blocks. Some people may like it, from what I found out simpler one-page overviews work better. An example could be our versionsapp.com site, which obviously has a very different style going on.

- It looks like an open source project page. I wouldn't expect to have to pay for something at a page like this.

- The whole no-logo thing is a pretty bold statement, but could work against you.

- The small serif font in the menu could be hard to read on some screens. I would make it bigger and/or non serif.

- The design is very timeless, and therefore it's hard to see if this is something recent or not.

Koen

Seriously: I was attempting to evangelize tarsnap to my company the other day (despite the fact that I have yet to actually use the service...) and was hindered by the fact that there was just nothing to link to. This is exactly what I needed last week.

Let your first priority be: Launch this right now. A good second priority would probably be to listen to patio11 and tptacek, but I encourage you not to let this distract you from the first priority.

(I already have, to people like H.D. Moore on Twitter. The HD's of the world is a really small target market).

Ok, I don't know why you, of all people, don't consider security to be a valid reason for using tarsnap; but if you don't like that, how about

* Tarsnap's snapshotting model of backups is very efficient (equal to or better than incremental backups) while providing the flexibility of having archives behave as if they're completely independent of each other.

* Tarsnap is priced based on actual usage, so if you don't have much data to store, it can be incredibly cheap.

* Tarsnap is written by someone with the expertise required, and who is willing to take the time necessary, to get things right.

That last, of course, is relevant to security; but it's also relevant to simple robustness, too.

I don't have much hope that anyone will automatically see anything. This is why I go out of my way to educate -- explaining things like why simple code is less likely to have security flaws, pointing out my expertise and experience, and demonstrating that I'm not afraid to have people look at the tarsnap source code.

There's a reason you designed tarsnap the way you did. It's not a little bit different from Mozy. It's a totally different model. Explain how in simple terms. And remember even if your beachhead market is sysadmins, most sysadmins don't code. Don't talk down to them, but don't make your value proposition be about code.

There are multiple levels on which this site is problematic.

1. Marketing

Who are your customers? Who is landing on this webpage? What information do they have? What information do they need to make a decision about whether or not to use/purchase/whatever your product?

What is your product, anyhow? Why are your prices in crazy units? Picodollars? Just talk like a normal person: "$0.30 per gigabyte per month."

It's totally self-defeating from a marketing perspective. I have no idea what you're selling or who you're supposed to be selling it to.

The goal of a landing page is to either convince me I need something or, if I know I need something, convince me you can meet that need better than anyone else.

2. Design

Nothing on this page catches my eye. I don't know where to look or click. There's too much text; don't make me think.

If you don't know how to design (I don't), pay someone to do it. You need to figure out what the first, second, third, etc. most important things are for your customers and then design the site in a way that directs them down that hierarchy.

Also: BIG. ASS. BUTTONS. See http://getdropbox.com for a nice example.

3. Brand

It looks like the homepage for yet-another-open-source project (I know it isn't). Very academic.

If I were really concerned about the security of my online backups I don't think I'd trust this product.

If "backups for paranoids" is what customers really care about (did you go talk to customers about that, or did you just pick the tagline because you liked it?), I'd like to see evidence that people just as paranoid me are using it, too.

Tarsnap is an awesome product. I think this site doesn't do it justice.

And any service that lists its prices in picodollars has made me a fan for life.

change the current css to something like this:

tarsnap.css line 107:

div.box h3 { border-color: #DDDDDD;

tarsnap.css line 91:

div.box { ... border: 1px solid #CCCCCC;

Maybe also change the 'Start using tarsnap now!' box header from pastel red (for stop) to pastel green (for go):

change tarsnap.css line 121:

div.box h3.cta { ... background: #A5F5A5 none repeat scroll 0 0;

I would also think about giving that boxes content div a background-color: #D5F4D5 or something, to make the call to action a tiny little bit stronger.

This is really inspiring. Congratulations. Goes to show that if you make something genuinely useful, even non-paying computer users take out their wallets.

Now for a few comments on the look of the site (which I think is pretty good overall, considering your current target audience).

I suggest turning off the full justification in the text boxes. Ragged right lends itself to much cleaner inter-word spacing.

Try to make the text boxes have the same margins on all four sides. With large top-bottom and small left-right margins, they look unbalanced.

I personally would prefer a menu strip along the top of the page to the menu bar you have. Several reasons for this. One, there's a lot of wasted space below the menu box. Two, I am not a fan of the indented submenus. If you make a menu strip along the top with a secondary strip for the submenus, I think it would make the entire site look more balanced. You should even consider tweaking your content to eliminate the submenu altogether, although that may be easier said than done. If you can pull it off, though, perhaps by putting some pages together with others, you will make it easier to grasp the contents of the site at a glance.

Finally, what is the proper capitalization? Is it "tarsnap" or "Tarsnap"? Make up your mind and be consistent. If you settle on "tarsnap", do not hesitate to use the lowercase form as the first word of a sentence. It's a small thing, but I think it clarifies the branding.

Maybe I've spent too long reading text typeset by TeX, but I find non-justified text to be very distracting.

Try to make the text boxes have the same margins on all four sides. With large top-bottom and small left-right margins, they look unbalanced.

Oops. Those were supposed to be the same.

...menus...

Yeah, I spent a long time trying to decide between side and top menus -- in the end I decided on the side menu in part because I have not only submenus but also subsubmenus. Things like man pages deserve to be on the website... but they don't really deserve a prominent space in the menu structure.

Finally, what is the proper capitalization?

That's a very good question. Can I dodge it by answering "sentence case", or is that likely to be too confusing to people?

I'd like to stick to referring to the client code as "tarsnap" (lower case) since the software is installed with a lower-case name and UNIX filesystems are often case-sensitive; is it possible to talk about the "Tarsnap backup service" (capitalized) but the "tarsnap software" (non-capitalized) without horribly confusing the matter?

The only thing place I think might need improvement is that first 5 minutes while someone internalizes what exactly tarsnap is. I can't judge it myself since I already got over that hump, but something as simple as a transcript of a backup session with a hint toward the strong security features would go nicely on the front page.

Yeah, that's always the hardest part -- my hope was that the words "Online backups for the truly paranoid", the pricing in picodollars, and the clean web design would at least start people off with the right idea...

... something as simple as a transcript of a backup session with a hint toward the strong security features...

That would be fantastic... but I'm not sure how to do that. The security of tarsnap is built deep into its core -- there really isn't anything people can see while they use tarsnap which shows them how secure tarsnap is. (I suppose I COULD add some random "AAA blocks encrypted with SHA-256, BBB bytes authenticated with HMAC-SHA256, CCC RSA-2048 private-key operations performed" messages -- but that would run directly counter to the philosophy of cleanliness and minimalism.)

The security part doesn't need to be anything more than a nod toward that core. Maybe something along the lines of:

  Bomb-proof backup in two commands:
  <generate a key as simply as possible>
  <backup some files>

* Make the front page simpler in content, richer in design/graphics:

In the front page, I'd put WHAT exactly is tarsnap (few words, like "Tarsnap is an online backup service for UNIX operating systems with emphasis on security" and avoid the "probably will compile" stuff, and WHY should someone use it instead of other products/solutions, in a few words. Drop the "start now" box and make a nice button instead that says "Start now with $5" or something simple like that.

Make a simple logo in GIMP, there are hundreds of tutorials for this.

Avoid listing the pricing in the front page, but make sure it is accessible from a nice viewable button. Use normal units, not "picodollars".

* Use a sans font. Serifs are nice, but seem "oldish" and are not the best fonts to describe a new product.

* I'd use blue and shades of grey as the dominant colors. Grey speaks "solid". I'd use other colors such as yellow,red etc. to achieve contrast to the buttons that might interest the visitor (pricing,documentation,legal etc)

That said, the new site is missing the mark. There is plenty of good advice here on how to improve the content. You best bet is to copy some ideas from your competitor's content. Even if you do that, it will still look like you did the web page and look ugly. There my be no way around you shelling out a few hundred bucks on a web designer. Every time I have tried to make my web sites look better, it was painful and I simply wished I had paid a designer to do it for me.

- put this up right away and iterate from there.

- no-one reads text. The best line you've got is "online backups for the truly paranoid". Put a graphic button somewhere so people can reflexively click on it and get taken to a payment page.

- please don't tell us what it DOESN'T do on the front page (i.e. Windows support). My website doesn't make ice cream, but it is irrelevant. Also, you don't want negative sentiment anywhere near your payment process.

- Improve the start area. - 1. Download and Install (5 mins....) - 2. Deposit $5 or more - It's that easy!!!!

The message sent by the design has changed from "Absolutely no effort has gone into designing this; I have other priorities" (see also, e.g., Craigslist, Drudge Report, useit.com) to "I thought this could do with a bit of thought about design, so I sat down for half an hour and hacked together a style sheet". This may or may not be an improvement.

Purely superficial visual-stylistic remarks (important note: I am not a graphic designer or anything; and I've looked at your pages using exactly one web browser on exactly one machine):

I think there are too many rules (mostly box borders), which are too prominent and too close to one another and to the text. And, generally, not enough whitespace.

I tried moving the boxes about 1em further apart, putting the text within the boxes a uniform 1em from the edges, and changing the borders from #000 to #999 (and #ccc for the rule between the box header and the main content) and I think it was a definite improvement. They still look, well, awfully boxy, but changing that would mean a pile more work.

The hover highlighting for your navigation sidebar would be greatly improved, I think, by changing the background (e.g., to the same as for a.current -- that "doesn't work" for the link representing the current page, but that may be a feature rather than a bug) instead of adding a border.

The majority of commercial websites seem to use sans-serif faces for most of their text. I like serifs :-), but there is something to the argument that they don't work well on the crappy low-resolution monitors we all have to put up with and you might find that, say, adding font-family: sans-serif and (something like) line-height: 140% to div.box gives a more "professional" look. You may or may not care about this. For some reason, I find that reducing the contrast just a bit (color: #333; you wouldn't want this for the headings) also has this effect. This may all just be because any change away from default settings produces the impression that more effort has gone into the design, or something.

Marginally less superficial comments:

On the About page, you should probably make "portsnap" and "FreeBSD Update" into links.

In addition to not being a graphic designer, I am also not a lawyer; but I wonder whether your legal page needs to say explicitly that the "Why?"s aren't part of the legal agreement.

Too many parentheses on the "Utility" page.

On the why-picodollars page, you probably mean 2^-40, not 2^(-30).

On the why-10^9 page, you claim that nothing other than RAM is addressed in hardware. This is, of course, not true. You might consider replacing "RAM" with "memory", or something.

On the scrypt page, PBKDF2 and bcrypt should be links.

It's a pity that the Why? links bounce the user to a new page. (Especially on the accounts page, which has forms on it.) You might consider some AJAXy nonsense to avoid this, but it's probably about ten times more trouble than it's worth.

Ouch! Thanks for the honesty, though. :-)

I tried moving the boxes about 1em further apart

I'll meet you half-way -- a 2em gap looks too big to my eye, but I like 1.5em more than the original 1em.

putting the text within the boxes a uniform 1em from the edges

Oops! I didn't realize that this wasn't uniform.

changing the borders from #000 to #999 (and #ccc for the rule between the box header and the main content)

That looks a bit too weak to my eye, but I agree on the principle -- I've dropped them to #555 and #aaa.

The hover highlighting [...]

I spent an obscene amount of time trying to decide what I wanted to do there -- changing the background was the first idea I had, but I couldn't get it to look a way I liked. I think I'll have to come back to this once I've gotten away from looking at that menu for a while.

...sans-serif faces...

Sigh. I hate the idea of specifying fonts in web pages... why can't web browsers just pick a reasonable default? But you're right, it does look much better with sans-serif -- consider it changed.

... reducing the contrast just a bit...

Lowered contrast helps readability, but I hate the look of grey text (maybe this is just a personal thing -- but I'd like to be able to look at my website without wincing), so I've done this using an off-white background instead.

On the About page, you should probably make "portsnap" and "FreeBSD Update" into links.

I considered that, but there's nothing sensible to make them link to -- I have pages about portsnap and FreeBSD Update on my personal site, but they are several years out of date.

I wonder whether your legal page needs to say explicitly that the "Why?"s aren't part of the legal agreement.

My law student friends tell me that this isn't a problem; and it doesn't matter anyway since I really don't care if the explanations are somehow interpreted as part of the terms.

Too many parentheses on the "Utility" page.

Good point, I didn't realize how many parenthetical remarks I was making when I wrote that text.

On the why-picodollars page, you probably mean 2^-40, not 2^(-30).

Oops!

On the why-10^9 page, you claim that nothing other than RAM is addressed in hardware. This is, of course, not true. You might consider replacing "RAM" with "memory", or something.

Is it not true? I can't think of any non-random access memory which is addressed in the same way as RAM is.

On the scrypt page, PBKDF2 and bcrypt should be links.

Good idea.

It's a pity that the Why? links bounce the user to a new page.

Those links should be opening up in a new window -- aren't they? That's what happens in my web browser, but it's entirely possible that I did something non-portable.

Thanks for all the excellent suggestions -- this is exactly the sort of comment I was hoping to get here. I'm not necessarily going to follow all of your suggestions, but even those I'm not following have provided very useful food for thought.

should be opening up in a new window: ah, I see. I've got so firmly into the habit of middle-clicking such links that I didn't even try left-clicking :-). (Which, yes, means that if you make them do fancy AJAX stuff then I will lose out...)

1em versus 1.5 vs 2em? That's premature optimization.

And statements like "[l]owered contrast helps readability, but I hate the look of grey text" tell me you're designing for yourself, not for your customer.

That's fine, if that's what you're doing, but I've been assuming the opposite, viz., that your goal is to increase the number of people who pay for tarsnap.

Not quite. In the absence of hard evidence either way (yes, I know, A/B testing -- but there isn't enough traffic on the tarsnap website right now to get any statistical significance for minor changes) website design is purely a matter of taste... and if it's going to be a matter of taste, why not go with mine?

And what you said isn't true, although most engineers I know believe it. It's not A/B testing or bust, and design isn't series of arbitrary aesthetic decisions.

You can also gather (non-numerical) data by going out and talking to potential customers, figuring out exactly what they need, and then making design decisions with that as an input.

That's not entirely true. Design and typography pros have over the centuries developed a series of rules and guidelines for what constitutes good and usable design. These have generally proven to be fairly universal and are more than simply a matter of taste.

I like it. It instills confidence. It's spartan, which to me communicates that you value performance over style (think road bikes vs. fixies). Very fitting for your target market.

I wish google understood your units so you could provide a link to the calculation. Perhaps you could provide your own calculator?

Only problem I have is with the floating boxes. I'd rather have one box than two columns. Makes for easier read.

On most pages I use just a single box -- I ended up going with multiple boxes in two columns on the front page because I figured that (a) there wasn't a huge amount of text to read, and (b) separating 'what is tarsnap', 'start using tarsnap', 'tarsnap pricing', and 'tarsnap news' into individual boxes some much-needed organization.

Is your comment about prefering a single box a general comment (in which case I think you'll find that most of the site does what you want) or do you also think that the front page would benefit from that?

I certainly like it better than the glossy competitors with their sneaky marketing-speak and buzzword bingo.

Thanks! The google-like "clean" style is very much what I was aiming for.

You have a point, though, in that he should explicitly put 'sans' in his CSS for the 99% of the population who can't be bothered to change their browser default font.

Ok, that last one is facetious -- but in all seriousness, I'd love to hear something more specific than "unprofessional".

- "can probably be compiled on many other UNIX-like operating systems" comes across well from one nerd to another, but probably not to managers. This gets into the question of who your target audience is.

- In the intro, talk about what tarsnap does instead of just saying it's a secure online backup service and talking about the "client code". What makes you different from the hundreds of other backup services?

- The red yellow and blue color scheme isn't doing it for me. Looks better if you make the red yellow, but I know you want to draw attention to the call to action.

- The gap between the left column and the news is ugly. Consider not having news on this page at all.

First of all, I am an ideal target user of the service. I know my way around applied cryptography, I can appreciate properly designed system, and I am really protective of the privacy of my data when it leaves my immediate vicinity. Properly encrypted backups is my kind of thing.

Your existing website is not minimalistic. The lack of the design is not a minimalism, it's a manifestation of the visual presentation of the product not being a priority.

The site looks like a personal webpage from the early 90s, implying lots of things but none for those that I would want to see in the backup service provider. I have visited tarsnap website on multiple occasions before; primarily through the references on the cryptography maillist. And I have always assumed it was either a hobby or an academic project.

Since you have had the site redesigned, I am assuming that you realized the impression the existing site gave was wrong.

The new design is certainly a step in a new direction, but again there is a number of problems with it. First, know your audience - who are your target users ? Let's assume I was right that it's the people like myself - crypto-savvy techs in a need of encrypted backup. What would I like to know first if I stumble upon your site ?

  1. How it works
  2. How much it's going to run me

Little digging on the site (that I did only because you asked to review the site, and would've not probably done otherwise) revealed that the data is stored on Amazon's servers. Good. Next question - does my data go directly to the Amazon or do you proxy it ? If it's former, then why do I need tarsnap as a service rather than a simple client product that handles data gathering, compression, encryption and uploading to Amazon ? If it's either, then, again, the issue at hand is tarsnap being an additional point of failure on route to my data (should, god forbid, anything happen to you or the company).

Assuming the data goes straight to Amazon, I will be happy to use your service just not to deal with the Amazon directly. What I need is a re-assurance that the data is recoverable, should tarsnap cease to exist.

To sum this part up - all of the above needs to be on the front page, or at the very least - one click away from it under an obviously labeled link.

Second point - how much is it - is equally important. The "picodollar" term as ingenious and geeky as it is needs to go. It is something that detracts the attention from other information on the page. I am reading the text, bam, picodollars, I keep reading, but I am actually trying to remember what power the pico is. So anything that I read after the picodollars doesn't really digest.

Additionally, the cost calculator is a must. Say I have 60GB of data to backup. Let me punch it in on the page and show me how much it would cost to run the initial upload and how much it's going to be monthly. Also an estimate of an initial upload time would be very handy to have. If you are on Shaw like me, the 50 KBps upstream bandwidth is not exactly online backup friendly.

Lastly, design-wise. It is not bad, but it is pretty far from being great. You clearly have a preference for a clean looking site, and I think this is an absolutely correct (visual/brand) positioning of the product. However neither the existing nor the new site are it. The site can be clean and appealing to the tech crowd and not look like it was designed for the Lynx browser. Have a look www.minimalsites.com, lots and lots of great examples of truly minimalistic, clean and light looking sites.

And that you maintain a FreeBSD (only?) port.

On the About page I talk about being a FreeBSD committer, security team member, and Security Officer, and about some of the code I've written in FreeBSD -- is this insufficient?

And that you maintain a FreeBSD (only?) port.

There's also a slackbuild (Slackware), an AUR (Arch Linux), and the tarsnap source code contains debian packaging bits -- so FreeBSD isn't unique in having packaging metadata available for tarsnap.

I plan on providing pre-built packages in the future too -- those will go onto the download page once they exist.

Nope. It sounds like exactly that for which I was looking. I couldn't (or rather, didn't) find it because I thought the indented bits under the Tarsnap label were telling me which things were included on the page I was on. I know, for example, that I saw News there.. and also News included just above the fold of the page I was on.

In fact, I suppose realized later that they were sub-menu items. When I went to the Design page and clicked on the Security link, I recognized that the menu on the left changed. I'm pretty sure it would not have occurred to me to click on them directly, however.

I spent maybe 90 seconds trying to find info about your work and the company, unsuccessfully. I don't know if this is, in any way, useful feedback. I mostly seem to be pointing out my inadequacies as a user, not providing any solutions to my so-called problems.

FreeBSD isn't unique in having packaging metadata

I looked on the Download page and expected to be told (especially around the time you mention that BSD systems usually have the OpenSSL header files already installed) that I could also find it in sysutils/tarsnap.

I hope my comment proves more helpful than it appears from my end.

Aha, that explains it. I did the menus this way because I (a) wanted to avoid javascript, and (b) thought it looked good -- but if it's confusing people I'd better see if I can find a different way of presenting the menu structure.

I looked on the Download page and expected to be told that I could also find it in sysutils/tarsnap.

Good point. Yeah, I should look at revising this page to point people at OS-specific instructions first, with the full install-from-autoconfed-source instructions only as a fallback for systems where I don't have special instructions.

Thanks for the feedback!

much better front page. not a tar snap user, but have followed the tarsnap story, and I think this is a good move.