Tuesday, May 25, 2010

Why Our Website Re-Design Failed

This is a cautionary tale about how, sometimes, logic does not prevail when it comes to selling goods online...

A few weeks ago I visited the Chocri, the brilliant "design your own chocolate" website. I was instantly hooked, and quickly designed some chocolate to buy. The only thing holding me back was their shipping policy - they don't ship to Australia!

However, the whole experience really resonated with me. It was fun, easy and I had my credit card in my hand in less than 5 minutes.

After all of this, it made me think about our own website for Shoes of Prey and how we could improve it. Admittedly, we're not selling $6 bars of chocolate - but the principle, I thought, should be the same.

I set upon redesigning the site using a style heavily inspired by Chocri's experience. This is what I ended up with:

The design I came up with was definitely different - in fact it completely moved our positioning on the spectrum between being a "designer shoe" brand to something closer to a transactional site (like Zappos) or a shoe making service (not unlike a dress maker, just for shoes):

By all accounts, this new design seemed much easier to navigate, more "fun", and provided a better explanation of the service. The overwhelming majority of the feedback was positive. Our friends and acquaintances seemed to love it. Here's a sample of what they said:

"Wow - that's heaps better. Doesn't really give the "bespoke" feel that you were going for before but does look more "legitimate"."
"This version definitely appeals to me more."
"f*^% yeah, heaps better!"
"100% better than the current design."
"In short - that new design rocks my socks off."
"Love the new site design. So much cleaner and cooler and more professional. The look and feel is great."

Our Japanese partners were the only ones who had reservations. In the nicest way possible, they said the site looked "cheaper."

But buoyed by the positive reviews, and with high expectations, we launched the site to our $USD traffic (roughly 50% of our visitors).

The result? .... Let's just say it didn't set the world on fire. If anything, our sales went backwards. Our bounce rate increased and time on the site dropped as well.

We wanted a revolution, and it just didn't deliver. More importantly it cheapened our brand; and in the long run, that wasn't acceptable. Consequently yesterday we switched it off.

Why didn't it work? It's probably because buying premium shoes is an emotional, self-defining experience. Consumers need to connect to the shoes above and beyond merely having an easy to use website. While our existing design has many flaws, it conveys certain emotional cues about how you will feel once you receive our shoes. For that reason it continues to sell shoes.

There's probably a lot more we can do from this point. In fairness, we might have tried to move too far along the spectrum towards being transactional. In the coming weeks we'll probably play around with some alternative designs that try to find some sort of compromise.

So, in summary, it's interesting to know that building a "better" website won't necessarily improve your sales. A lot depends on how you position yourself in the market, and how your customers want to perceive you. For a brand new retail concept like ours, it's really hard to know instinctively how to get this right. Experimenting, though, will certainly help.


  1. I have a really hard time believing this. There just has to be another explanation.

    Well, sure, _maybe_ you should stop accepting advice about your high-end-womens-shoe-store from a slobby, 30-year-old male computer programmer, BUT I just can't wrap my head around why the worse design (its not a subjective when its MY value judgement) is winning here.

    A few thoughts: how much of your sales are from the "ready made" section? That seemed harder to get to in the awesome-er design - could that have been a factor?

    Did you do any latency measurement or other tests to see if something technical could have been at fault? How were you conducting the A/B testing (I think you told me but I forgot). Cookie based? Obviously bouncing between designs is an awful experience...

    Did you split up geographies at all? The only thing - inspired by the japanese critique - that I can think is that maybe ideas of "cheapness" vary greatly between locations. The darker design (the one up now) looks WAY cheaper to me than the new one you tried. Sort of faux-commerce, like Grandma's first website. Haha sorry, I know thats harsh, but I really, really don't like it. The "new" one looked to me like a fresh, navigable storefront - but much more like the retail sites I'm used to visiting and purchasing from... so perhaps the sites your primary demographic visits are just that much different than those I visit? Possible, but still not a satisfying explanation.

    Something has to be up. Wish I knew what it was.

    In any event - have you guys considered "naming" the pairs of your "Ready Made" shoes? I like the way other sites do that - it makes them feel hipper and more in tune with my lifestyle. I'm thinking of something like bonobos.com, which gives their pants style an unconventional name. I just imagine it being more of a brand connection for somebody to say "oh, these? they're from ShoesOfPrey - they're the 2010 Lavender Lipsmackers" or something, rather than somebody being like "they're from ShoesOfPrey, click on 'ready made' and they're in the 3rd box down on the left"

    Just a thought - but be forewarned that accepting advice from me generally results in plummeting sales!

  2. In the future, I recommend selective A/B testing for a percentage of your traffic. ie: You don't have to guess! Just test it and you will know....I understand Google tests ANY change, right!?

  3. LOL at "2010 Lavender Lipsmackers"! Your talent is currently being wasted Cameron. You need to get into shoe designing ASAP! :-)

    Good thoughts too on other possible explanations. The new site should have loaded faster for people, and as far as I know, there weren't other variables at play..... It's possible though that there were, but we just never found out people were having difficulties. It was seen by a good sample of people though - at least unique 10,000 visitors.

    (We selected participants for the test based on their location/currency, which was then stored in a cookie. The people that saw the new design were basically all for the US. Our theory was that they were most likely to be receptive to an e-Commercey website.)

    Ready made is only a new section too -- we added it a day before we took down the experiment, so I don't think that would have caused any variation.

    I guess at a high level it is a trade off between being "cool" and being frictionless in terms of closing the sale. The new design required us to give up some fashion credibility in favor of making it easy to sell. Our basic theory was that the new website would outsell the current website by a lot. When that didn't happen we decided it wasn't worth taking a reputation hit.

    I think the way forward though is to continue to work on the design. I for one liked using the new design a lot more (maybe it's my male brain at work)... so I actually would switch deliberately to it. I'd like to try and find a design which achieves this AND increases sales by a noticeable amount too.... I'm sure it can be done.

    I'm currently in Tokyo at the moment and I've been really inspired by the amount of creativity here... particularly in the retail space. I'm looking forward to channeling this inspiration hopefully into a new experiment.... so watch this space! :-)

    Jeff - Good thoughts. It's difficult though to A/B test an entirely different website using Website Optimizer (although I'd say it is possible). Because we automatically select the currency based on the user's geography, it was easy to set it up like this.... although keen to do some more Website Optimizer experiments too in the near future as well!

  4. I like the new design (I've been around for all your designs and am not a fan of the current one) but do agree it is slightly "cheap" I think it's on the right track though and I much prefer the layout and navigation to the current site.

  5. I loved the redesign. I thought it was much easier to navigate and more appealing, so I'm really surprised it had a negative effect. I"m also surprised that feedback said it looked cheap - I think the one you went back to looks cheaper to be honest!

  6. How long did you run the experiment for? Are you sure you're not accidentally optimizing for a local maxima here while users get used to the new look? Did you run it for over 30 days? Did you look at search funnel data to see if your users were converting on different ad paths? Dude - I have to believe that there is something else going on here, this just doesn't make sense.

  7. Christine - Thanks for your thoughts. Yeh, I loved it too! I wish it had worked, as I much preferred it for navigating the site. Hopefully we can nail it soon... although, that said, the current website is performing at or slightly above the targets we've set for ourselves.

    Cameron - Am I right in thinking you hated watching magic tricks as a child without knowing the secret? ;-) We ran the experiment for 2 weeks. You're right - there are possibly some other factors at play too. Like you said, it might have taken some people a while to get to warm up to the site. However, our analytics data (for what it's worth) suggests that about 85% of people buy within 2 weeks of visiting the site for the first time (or last clearing their cookies!). That makes me think that 2 weeks is probably an okay period for a trial. In terms of search funnels - at the moment we're not really doing that much advertising. However well over half of our revenue comes from direct visits, or organic keywords. So, given all of this, I was expecting the needle to move a lot more than it did.... and, I guess that's the point. In order to re-position our brand to be more service orientated, we would have needed to see a bit uptick in sales. When that didn't happen, I think we just thought "what's the point of changing something that is working the same, if not better than the alternative...?"

    However, all is not all lost. I'm committed to exploring new designs that are more "ecommercey". However I just think I need to dial it back so that we keep more of our brand personality...

  8. Hah! Worse, I was the kid that was trying to ruin the show for everybody else by saying "I know how he does that!". Just one of the reasons for my 19-year birthday party invitation drought, I'm sure.

    Anyway, I don't blame you for switching back or anything, but I'd encourage you to remain incrementally experimental, which I'm sure you will. Really interesting to think about though... thanks for being so open about all this stuff, it makes it fun to follow along. I feel like I'm starting an innovative, fun, ingenious business vicariously through you guys!

  9. Good Job Michael!

    Im planning on opening an online shop for my father's business myself. Can I have your eamil, so I can ask you a few questions?

  10. wow, I am very surprised. But again, I'm not the target demographic. I guess this is more a lesson in female psychology than it is a website design experiment.

  11. Hi Foxy, that is surprising feedback as the site refurb looks so good. May I suggest your old site provided more intrigue therefore required more exploration where the new one pretty much says what it is and hence you make a decision a lot quicker to stay or go. Maybe a question on how you can take people on more of a journey to become engaged with your brand. Alot of ecommerce site are the same which fails to differentiate from a brand perspective so I think people respond well to something a bit different. Recent research has shown the power of video to drive online sales with some good examples. I can flick this over to you if you like. Cheers Jasper

  12. Hi Jasper,

    I think you're likely right that the intrigue, exploration and the site being different to most e-commerce sites resulted in people becoming more engaged with the brand. Certainly when we launched this version of the site our target market loved it, while our online savvy friends didn't, which probably fits with your explanation.

    Would love to see your info on the power of online video. We've been starting to do a little more with our Shoes of Prey YouTube channel so would be really keen to see what's worked for other retailers in the video space.

    Thanks Jasper.

  13. Currently being deep in a redesign, this is timely reading. How has the transition been to the old website? Are you able to get feedback from users? When iinet transitioned to a new website not too long ago they gave you the option of running the old one if you preferred. I imagine they would have been able to gather a lot of the data by running both sites concurrently.

  14. Thanks Ryan for the comment. You make a good point about getting comments. I guess we were running off the theory that if people had strong feelings about the website they'd email us and, ultimately, they'd buy our shoes if they really liked the website.

    Running 2 sites site by side is probably okay for a consumer internet product, like Facebook or iinet, but I'd be worried it might confuse our users, because both of the sites took different approaches to the way we positioned ourselves in the market...

    Good thing to think about though for the future.

    I should say too, it's been about 1.5 weeks since I wrote this post. We've had slightly better sales in $USD since we reverted. Even so, I'm beginning to think there was something in what Cameron said. Perhaps there was another explanation.

    Someone recently told me that a lot of the stores in the US are all on sale (and they are here in Australia too). Massive sales, in fact, because the stores can't afford to buy their holiday stock...

    So it's very possible that we got unlucky with the timing.