What makes Cart45 different?

An installation process that takes a few minutes, good quality SEO and ease of use for both customers and website owners. We’re hoping Cart45 will do for ecommerce what WordPress did for blogging.

We believe that this marketplace is saturated with difficult to use shopping carts (yes, I am looking at you Actinic) and we’re hoping Cart45 will change all of that.

Cart45′s development

The six months of Cart45 development have been enormously fruitful for me in terms of my professional development. My object-oriented PHP has improved tremendously and my programming thought process has progressed too. I’ve been very lucky to have been working on this project and I’m really hoping it will be a success.

Architecture

Built using MVC (well close to), Cart45 should be easy for novices to setup and super-easy for professionals to customise for clients. Who knows, this time next year there may be a whole cottage industry based on Cart45 installs and theme developments – just as there is with WordPress.

Read about Cart45′s Folder and file structure

System requirements

Cart45 needs PHP5, MySQL5 and Apache. I know a lot of PHP frameworks and shopping carts cater for PHP4/MySQL4 as well as 5 but we want Cart45 to help push things forward. If there is sufficient demand for PHP4/MySQL4 then we may explore supporting them.

Competition time

To celebrate the launch of Cart45, I’m giving away 3 copies to readers of this blog. To enter simply leave a comment or a question about Cart45 on this post (see form below) between now and 31st July 2008 and I’ll pick 3 winners at random.

Any questions?

Want to know something about Cart45 – ask me a question. I’ll do my best to answer it.

Some background

Whilst working for JJB Sports Plc (2005-2006), I became increasingly aware of most web agencies’ inability to master search engine optimisation (SEO). In particular, JJB Sports’ main e-commerce site (which cost a lot of money) was woefully unoptimised.

JJB’s website started out with several obvious SEO errors.

1. All pages had the same page title which was JJB Store
2. Poor category naming with extensive usage of industry jargon
3. Dynamically generated URLs e.g. www.jjbsports.com/search.php?supercategory=FR&wcategory=FREP
4. The e-commerce site was launched at the domain www.jjbsports.com but all the PageRank was on www.jjb.co.uk which remained as an information site.
5. There was practically no text on the site besides product images and lacklustre product descriptions.
6. The description meta tag was missing

Problem 1: All pages had the same page title which was JJB Store

Two weeks after starting at JJB in my first web development job proper, I went to see a demo of the new ecommerce website. I was flabbergasted – it was awful. Then I learnt the price tag of the site (which I won’t reveal) and I nearly died!

One of the first things I noticed was that every single page had a title of JJB Store. Firstly, the brand name was JJB Sports or JJB so inventing a new persona was (in my opinion) a bad idea, the fact that it was the only title on every page set alarm bells ringing that the company who had designed the site was quite inexperienced.

Note: When I say page title I’m referring to the window title that you see at the very top of your browser, which is stored in the <title></title> tag of the page source.

Solution: New titles

I changed the titles to better reflect the content of each page and also maintain the brand of JJB so the title of JJB Store on every page became:

Product – Product Category – JJB Sports

So for pages with nike trainers on them, the title became:

Men’s Nike Trainers – Nike Trainers – Nike – JJB Sports

Problem 2: Poor category naming

The category names on the website was also poorly thought-out. Too often, the categories were named after the internal business names so the football section was ‘Replica‘ and the trainers section was ‘Footwear‘.

This oversight was no doubt causing problems for SEO but also for the people using the site who had no idea that the football shirts which they wanted to buy were stored under a category called ‘Replica’.

Solution: New category names

Thankfully, a colleague was also on the ball here too, and over time everyone involved in editing the site improved category names as they created them. Top selling items were grouped into their own categories e.g. Replica became ‘Football Shirts‘ and ‘Football Accessories‘ as football shirts were the best selling items on the site it made perfect sense to optimise for them.

It’s important that products are categorised properly for the user and for the seller. Supermarkets like Asda spend a lot of time working out what to products go in which aisle at what height and next to other items so that they sell the most of their high profit items.

Problem 3: Dynamically generated URLs

URLs need to be readable by users and also by search engines. In Google, when you search for something e.g. ‘stuff‘ every instance of that keyword is bold within the search results. That means if your website’s page title mentions stuff, it’ll be highlighted, if your description mentions it, it will be bold and if your URL features it, it will be highlighted there too. Maybe users don’t notice this but, I’m convinced they do even if it’s subconsciously.

Solution: user friendly URLs with mod_rewrite

I set-up mod_rewrite rules for the site so the URLs properly reflected their content e.g. jjbsports.com/football/. I was still limited to that fact that the content management system only allowed 3-4 letter category IDs. I could have accounted for this with a huge mod_rewrite file but that would future proof the site.

Instead I renamed big categories as jjbsports.com/fitnessequipment and then stuck to removing the horrific query strings from the URLs on subpages so:

jjbsports.com/searchcategory.php?supercategory=FR&branch=FWF

became

jjbsports.com/fr/fwf/

still not too pretty but a little better.

Problem 4: Multiple domains, one goal

When JJB launched their new e-commerce site in August 2005, they launched it at www.jjbports.com which had previously been a domain alias for www.jjb.co.uk (an informational/data-capture site). JJB wanted to keep the www.jjb.co.uk site as data capture and a portal to other JJB websites but, it was damaging the PageRank of the new site.

The result was that www.jjb.co.uk came top of the search results for JJB related keywords with www.jjbsports.com at number 2. www.jjb.co.uk got masses of traffic that simply clicked on the first link they saw to the e-commerce store. This wasted customers time, confused them and didn’t help JJB much either.

Solution: one domain to rule them all

The solution was to set up a mod_rewrite directive to direct the actual www.jjb.co.uk homepage to www.jjbsports.com with a 301 redirect thus making the www.jjb.co.uk obsolete but, still allowing subsequent pages on that domain to remain in operation. Within days, the fortunes of www.jjbsports.com websites in Google’s SERPs improved dramatically.

Problems 5 & 6: Lack of text and meta tags

The site made no use of the meta description tag which is very powerful. Google often the meta description as a website’s description in their results.

Solutions: Improving the use of text/meta description

Therefore the solution was to firstly, improve the text of all the pages then put some of that new text into the description tag.

The website had practically no text on it, category pages had images as links to subcategories and subcategories had images as links to actual products. I changed that so every page had a mini-description e.g.

Nike trainers for all the family. We’ve got Nike Air Max, Nike Tennis Classic and many more of your favourite Nike trainers.

This bumped up the keyword count of all pages.

An extra solution: An opportunity to cross-promote

More text on the page meant more chance to add links and therefore cross-sell. All product descriptions were possible were given links at the end e.g. looking for nike trainers? This was done on virtually every page. I did this for cross selling as much as SEO. When a user landed on the men’s nike trainers section they were given the chance to quickly find women’s nike trainers section.

Summary

At every stage, the changes made to the site were measured with Google Analytics and they were consistently proved to be making more money as people found products more easily.

The mistakes JJB made on their website, are very common and are due to a lack of experience on the part of the website owner and certain web agencies. A website is a complex bit of kit and good search engine optimisation (SEO) should come as standard with any website.

This article first appeared in February 2007.

1. Be as descriptive as possible

Ensure you give as much information as possible to potential buyers. Some eBayers will ask you questions if they don’t see the answers they crave but most won’t and that majority won’t bid on your item either. I often copy the sales copy from another seller’s item if I intend to sell a similar item – only if it has achieved the price that I want.

2. Keep the starting bid low & have no reserve

You may be nervous about this rule but it’s imperative that you get people bidding on your item as soon as possible. The higher the reserve/starting bid the less likely people are to bid. As soon as 1 person bids a £1 other people become convinced that this is a good item because someone else is interested enough to buy it.

Think of this analogy; You’re starving and there are two restaurants next to each other neither of which you’ve ate in before. Restaurant A is (comfortably) full of people whereas Restaurant B is empty. Which one are you going to choose to eat in?

3. Use as many (good) pictures as possible

People need to see what it is they’re buying to feel confident about buying it. They don’t want to make a mistake and be ripped off so if they can’t see good pictures they won’t bid. Take as many pictures from different angles as possible so yoru potential buyers are not left wondering if there’s something hidden they can’t see.

Also make sure you use pictures of the actual item you’ve take yourself. I’ve made the mistake before of using pictures from the manufacturer’s website and assumed that people would realise that my item was used and so would have some wear and tear – the didn’t and my mistake resulted in an unhappy buyer and a partial refund.

4. Make delivery as cheap as possible (if not free)

People hate to see those high delivery prices. Putting a high delivery cost on your item can push up your margin and it is tempting to imagine people are so stupid that they can’t calculate that with your heavily marked-up delivery price and the item’s listing value that they are being ripped off – but they can and do.

A high delivery price may seem like a good idea because it’s more money in your pocket but it’s a false economy because less people bid on your item and consequently you’ll get a lower sale price.

5. Make your auction end on Sunday night 8-9pm

Late Sunday night works for me – it may not be the optimal time but it is a time when people aren’t doing much; they’re practically sat at home waiting for Monday morning. Too many times, I’ve got bargains because people have set their auctions to end during business hours – when most people are banned from using eBay by their IT departments. Set your auction to end at a time when most people are at home and looking to buy, that way you’re more likely to get a bidding war at the end of your auction.

Summary

These tips really work for me. Do you use similar rules are do you have your own eBay selling tips?

This year prior to the big Glasto ticket sale you had to pre-register which for the first time in history the tickets companies (in this case seetickets) knew exactly how many people would try to use their website/phone lines at once. Yet when we logged-on at 8:50am (10 minutes priror to sales) the site was already down. How? How is this possible? Give the profit these guys must be raking in not to mention that this is their speciality – selling tickets but they can’t even do it right.

Thankfully, the heavens smiled on some people sat next to us who actually managed to get to the website (as opposed to the server busy message that we kept getting) and after they got their tickets they let us use their machine. Great. Tickets bought! Confirmation screen gives confidence that an email will arrive with proper confirmation. However, it’s now the day after and still no confirmation. We can only pray that their email system has broken down and no-one else has received their confirmation. How sad it is that after criticising the terror of the seetickets machine we are now hoping for more seetickets mediocrity which’ll mean we actually have tickets.

Certain figures get bandied about for checkout drop-off rates from 60-90%. Every ecommerce site’s products and therefore, customers are different and so are their checkout processes, so it’s difficult to state what is the average complete rate for an online checkout. The best solution is to monitor what drop-off rate and then try to improve it.

Rising completion rates

At JJB, our completion rate rose from 30% to 32% at the tail end of August and since September we’ve been looking at a completion rate of 33%. In July the completion rate was a mere 23%.

The reasons for this our two-fold, the first is we had a real boon with one product creating a real buzz and creating huge sales which, in turn, pushed more people through the sales process. The 2nd reason was that I removed all the majority of the site navigation from the checkout process.

Decisions, Decisions

If you give people the option to make a decision, chances are they will. Great! We’re all for user choice… or are we? The answer to that question is no, not really.

(Most) People still don’t know what they’re doing online so presenting as few choices as possible is the best option if you want higher success levels. If there is an option to get to another part of your site and then there will be people who take it by mistake.

CPC

This technique has also been recommended for CPC landing pages, by a Google rep, to me before, however I can’t believe it’s the same (who am I to argue with a professional!). In the case of someone coming from a Google search (natural or CPC) I think they need the navigation to establish where they are in the site. If they land on your green candles landing page but they wanted something slightly different you have to be able to give them the opportunity to find that.

Summary

There are times when navigation guidelines (Where am I? Where have I been? Where can I go?) need to be altered for your and your visitors’ requirements.

Assume the worse of your e-consumers and never underestimate their inability to read and think whilst online. Do your best to guide them around your site so they go where you want them to. Much like the supermarkets guide you through fruit and veg then onto meat then tinned goods and so on web designers must ensure that customers are given the biggest possible push in the right direction so sales are made and information is consumed

The Zoom link is a very common feature on ecommerce sites therefore, you’d expect it to be highly used by the general public. I monitor everything on site with Google Analytics and I can tell you that less than 0.5% of people on the JJB site either click the zoom image text link below the image or the click the actual image.

This is heartbreaking on many levels:

1. We don’t have room to offer a bigger image than 205*205 pixels for our main product image due to the design so our enlarged/zoomed image usually is much bigger and contains different angled shots of the product in question.
2. We use the JavaScript lightboxing technique which I believe is better than using pop-up windows for usability/accessibility and looks damn fine also.

The experiment

I decided to experiment with the wording of the text links: some categories’ products had one wording whilst others had another, all the while I checked the number of clicks each wording was getting.

Image link vs text link

The first result shows that 64% of people who ‘zoom image’ click on the text link (regardless of it’s wording) and 36% click on the actual image. It’d be interesting to see if the same number of people would click on the image if the text link wasn’t visible.

Zoom Image text link wording

Summary

This is not fully conclusive as some categories get far more traffic than others so another technique I’ll be testing will be to swap around the wording with the same products on different days of the week. If the results hold out next week then it will give conclusive proof which wording is best.

Other experiments

Even though I’m striving for the optimum link text wording, I’m very aware that it won’t make much difference and maybe how the link looks or where the link appears may have to be experimented with in the future too.

Last week, on the JJB website I decided to make some small changes and of course every change was monitored by Google Analytics.

The Changes

Cross Linking

Firstly, I added some cross links to product categories for SEO purposes and also for usability. The idea being, that the site would have better links sitewide to terms like football boots and nike trainers but also that if a female customer found themselves on the men’s trainers page via SERP they could easily get to the women’s trainers section.

Therefore this, change has more than one benefit… hopefully: SEO & Usability.

Hero Items

The hero items are our supposedly best selling items. On our site it can be difficult to promote individual items in sub-category listings so our top instore selling items can get lost on page 5 of products listings and therefore never get seen nor bought. As most people don’t scroll we’ve found that putting the top sellers at the top of the page after just one customer click gets great conversion.

Obviously, we can’t prove that customers wouldn’t have found the product anyway but, we can see that nearly 20% of people who click on the football shirts main category clicked on one of our hero items (which is more than clicked on any subcategory) and a lot of purchases were made.

We hope this technique makes browsing and buying from the site much easier which, in turn should lead to repeat purchases and happier customers.

Improved Search

The search improvement can be seen in the example (to your right) which shows a site search for school brings back the most popular school shoes we have for sale. Not shown in the picture we also promote the best selling backpacks and school coats below the shoes.

The diagram above clearly shows, that although the actual number of searches decreased (mainly due to it being a Bank Holiday weekend in England and consequently all round less traffic) that the value of those searches went up by nearly 20% which is astounding. This is great for the business because it helps customers choose easier.

Also, we’re aware that our search engine is quite poor but the solutions to fix it completely are quite expensive and beyond my capabilities. So this fix costs nothing except the time to write the code (1 day) and the time for somebody every week to make sure the recommended products are in stock and correctly priced (2 hours a week).

Admittedly, isn’t as awful as some (we’ve all seen worse) it isn’t as forward thinking as the website would suggest. A quick View Source reveals the culprit… Microsoft Visual Studio.NET 7.0

What JohnLewis.com does right

1. Fluidity. This site looks as though it was designed for 800*600 and 1024*768 and doesn’t look too shabby on even big screens either. This is hard to pull off but then again they’ve used tables so they have cheated.
2. No fly-out menus. When I first saw this site. that’s what I was expectin and I didn’t get it thankfully.
3. A quality footer. Packed with information that was clearly part of the designer’s thoughts early-on. It’s very welcome and lets people find useful information with ease (well those who scroll anyway).
4. Their related products are easy to add to basket at the same time as adding the item you want. I can only imagine upsale/items per order is quite high on this site.
5. The customer service page is full of information and instead of hiding behind a link called FAQ (which by the way no non-web designers understand the meaning of) they show the customer the most important questions in plain English.
6. The order telephone number is clearly displayed as soon as you get to the checkout.
7. You are only asked to register once you have bought. No usernames & passwords if you don’t want them. Hurrah!

What JohnLewis.com does wrong

1. I’d find it very interesting to see the stats or see some people using this site as I don’t think it’s overly obvious which parts of the page are clickable. Each department has a landing page with images linking to the main subcategories but to me they just look like decoration and not links.
2. It could be argued that the design is overly simple. I might have added a splash more colour but it is very fittng with John Lewis branding so it isn’t a big issue.
3. The contrast on buttons makes it difficult for me to tell what they say let alone someone with poor vision.
4. The checkout process is too reliant on JavaScript. I turned JavaScript off and I wasn’t able to click any buttons and progress with an order. That’s just poor.Once again .NET is the criminal mastermind behind this.

Summary

I would dearly love to know the conversion rates, checkout process drop-off rates and everything about this site. I feel from a customer point of view it’s close to perfection but could do with a little TLC to fully maximise profits. I am going to be using John Lewis as a huge inspiration for the next stage of JJB’s development.

This is a mistake that we’ve made at work, but have recently rectified.
This is also a mistake that quite alot of UK (newly-online) retailers make – especially our competitors.

This mistake in the wild

Sportsworld homepage

Take for instance Sports World, a customer finding themselves on www.sports-world.com is greeted with two options:

1. enter sports world retail
2. enter sports world online

As a potential customer my first question is, What’s the difference? If you’re making your customers ask this question your making them think and people don’t like to think on the web and won’t click either option or worse will click the wrong one and get frustrated by the result and not return.

Why this mistake is consistently made

You may find it strange that I would critique a competitors’ website for free but I can’t foresee them changing this situation so I feel safe. The reason for retailers using techniques such as this is pressure. Pressure from their high street stores who think the internet will steal their business and their jobs – and of course it will to a certain extent but, that’s going to happen anyway so why fight it?

If XYZ Shoes refuse to create an ecommerce site for fear of upsetting their store managers then ABC Shoes will simply control all the online shoe market with their website and they’ll probably bid on the keyword XYZ Shoes on Adwords.

How to fix it

At JJB I’ve recently removed this awkward feature – although ours was slightly different – for various reasons our store locator and a competitions data-collecting feature needed to be on a separate server and therefore a different designed website that had different goals but having this separate website must’ve seemed ludicrous to an outside observer.

Removing yourself from the pressures of people above and around you and looking at your website from the perspective of a customer can be hugely rewarding. Doing this allows you to see the site not from the perspective of a politik but from someone actually spending money and it makes you see where and when you’re made to think or led to confusion by your own website.

Disclaimer: Putting yourself in someone else’s shoes is not a complete alternative to usability testing with real people.