sylleptic - mark hopwood’s blog

(This is a piece I wrote for the retail solutions conference guide.)

If you’re a retailer or a consultant involved in choosing an e-commerce solution, the approach you take, and in particular where to start the process, is critical. You might start with a decision on the underlying application server (Microsoft or Java, for example). Some look for something that fits well with a system they already have (like their ERP or warehouse management system), or with a “must-have” application or consumer offering, like facet-based navigation. Others start from the ownership or implementation model: are you looking for a hosted solution or something you can host and support yourself? Perhaps you would like your own unique instance of the application you can adapt totally to your needs? Do you need to be live in 12 weeks, or will you be doing a detailed analysis and customisation project? Subjective feel about fit with the vendor or vendors often carries a lot of weight, and budget is always a deciding factor.

In practice, many people come out of this process with a solution that is quite complex: the level of flexibility and integration offered by many products today, and the long lists of existing integrations seen in so many vendor presentations make choosing a hybrid solution a normal outcome. This raises some difficulties that must be considered when choosing the partner or partners who will deliver the project, but it also raises some real issues about the nature of the solution for the people who have to operate it once it is live.

The problems around project delivery relate to responsibility and complexity. Let’s say you’ve chosen a search product, a base commerce application, an image and video delivery tool, and a payment service provider, and you need integration with your existing warehouse and ERP. Who’s going to take responsibility for making all of those pieces fit together? Do you have the right people in-house to do the complex technical management that’s required, or will you use a consultant? Will you appoint one of your vendors to manage the overall implementation? Who will lead and manage end-to-end testing of the fully integrated solution? If you don’t consider this basic governance and accountability, problems are certain to arise. Aside from these questions, you need to consider who will manage the technical integration of all the pieces. All sorts of different interfaces are required between systems: these need to be specified, developed (or adapted from existing ones) and tested. Someone needs to make sure all the interfaces click together. This isn’t trivial, and it seems it’s sometimes assumed to happen by collaborative means: our experience is that it’s a lot simpler and a lot less risky for all partners if one party to the project is explicitly identified as being the lead for management, integration and verification of the solution.

If you’ve completed a successful multi-vendor e-commerce project, there remain questions about the integration of management tools, and their impact on the efficiency of your back office. I was recently at a presentation for a site content personalisation tool, and it became clear that whatever the benefits the tool had in terms of user experience, conversion and basket value, it was going to make the retailer’s management task a lot more complex. When a retailer chooses different vendors for different parts of the solution, it adds many steps to the management processes. Workflow for these tasks sometimes seems to be under-played in solution definition, and end-to-end documentation and testing seem to get lower priority than they ought. Let’s take  the  example of adding a new product and putting it on promotion – this really should be quite simple and yetthe merchandiser might have to use the base commerce platform, the search management console, and an image management console to get everything up and running.

This is a problem that some of the big vendors with a comprehensive offering would suggest they have the answer for: if you buy their product, it all fits together, and all the tools are integrated, so management is easy. It’s worth remembering that several of these products have been built by acquisition, and weren’t designed in an integrated way, so the reality may not be what is promised in the literature. We would recommend customers to choose a base product, like Microsoft Commerce Server, that works well for most of their requirements, and only add to it in really high value areas: most people should challenge customisation and additional complexity wherever it’s introduced into a solution.

In the longer-term, we would expect much easier integration between all the diverse products that are on the market. The evolution of web services as a standardised and flexible way for applications to talk to each other already makes it easier for applications to integrate in many ways. For example, putting data into the big search products today is a matter of generating some XML from your product catalogue and firing it into the index server. We’ve built our own management tools on top of web services so that they can be integrated and re-used in all sorts of ways, but it would be really exciting if we could integrate the management tasks of multiple vendor products in this way: as vendors expose more of their management capabilities to systems integrators as web services, we should be able to turn my new product example into a single page transaction in the future.

E-commerce continues to be an exciting and innovative space to be involved in technically, so we can expect to see a host of ideas coming forward to manage complexity, both in project delivery and management of the solution once it’s live, in the coming years.

Earlier today, I was trying to book a train ticket when one of those “Verified by Visa” prompts appeared on the screen. If you buy much online, you will probably have seen one of these by now. They’re supposed to prevent fraud, because they use a password in addition to the 3-digit security code on the back of your credit card, and in theory only you and Mastercard or Visa knows what that password is. But there’s a problem:

• I keep forgetting my password
• It’s actually pretty easy (too easy to be secure) to reset it online
• But when you do forget it and fail to reset it, it locks your card for online use, and you have to phone the bank

What this leads many people to do is to stop using websites that have the “Verified by Visa” logo showing on them, maybe because they don’t have the patience to remember and re-key an extra password, maybe because they’ve forgotten it, or perhaps they’re locked out and can’t be bothered to call their bank when it’s easier to buy the thing you were going to buy somewhere else, which isn’t signed up to this scheme.

Which is probably why most retailers who implement what the industry calls 3D Secure are losing customers when they do it.

When it happens online, apparently. This article reminded me of 2 meetings I’ve had recently. A fashion brand which is seeing enormous attempted fraud on its website showed me a spreadsheet, each row of which represented a fraud or attempt at fraud. There were thousands of rows for one month, and most were from addresses in the UK. Will the police do anything to address these cases? I’m sure every online retailer experiences the same things, and they’re essentially on their own as far as the government and police are concerned, which is why companies like Third Man and Cybersource are so valuable to them.

The second conversation was with the website manager at a retailer about the factory outlet in the building next to his office. Someone ran out with a pair of trainers a few days before our meeting, and there were security guards, dogs, he was stopped and apprehended at a barbed-wire fence, and arrested when the police arrived. The trainers were worth £14.99…

So physical crime is definitely getting priority at the moment, yet the value of online crime seems to be so much greater. Are the police (and the government) lagging behind everyone else? Criminals are channel-shifting, just like the rest of us.

The blog’s been pretty quiet the last few days. Here’s why. I’m sure nobody wants to read a blog about how busy I’ve been in my new job…

More soon.

Amazon used to be one of the text book examples of usability in e-commerce. Famously, “Don’t Make Me Think”, an excellent introduction to user centred design by Steve Krug, includes a chapter called “If you love Amazon so much, why don’t you marry it?”

Recently, I was given an Amazon gift voucher, and I’ve been using the site a fair bit as a result. I’m afraid the experience hasn’t been very good.

Firstly, the Amazon stock and the market place are mingled in the most confusing ways. I can only use my voucher on Amazon stock, and the fact that I’m going to have to pay cash for something isn’t made clear until very late in the checkout process. In fact, it isn’t really made clear in the late stages: I had to look very closely to see what happened, and in one case had to cancel a market place purchase after I realised they hadn’t used the voucher.

Secondly, tracking missed packages has become incredibly hard. You’d think clicking the “where’s my stuff” link in your account would help, but you have to go so much deeper than that, and the navigation is impossibly unclear. I felt like I was testing the site for them, and in fact the tracking didn’t work when I did find it.

Generally the whole site feels a lot more complex and clunky than it used to.

What did work well was the tool that sets up calls to their call centre, facilitated by estara (owned by ATG, who compete with Amazon in some regards).

When you compare amazon.co.uk with Amazon’s excellent Endless shoe site (still actually a beta) you can see what they are capable of, so why is the main UK site so poor?

From the Snow Valley blog it seems that Amazon are going to offer same day delivery on Christmas Eve, if you order certain products, and your delivery is to London or Birmingham. Good news for those of us who leave Christmas shopping to the last minute. It got me thinking about how early Christmas seemed to start this year, though. At the risk of sounding like an old git, it used to be that if you saw any Christmas advertising before November 5th, it was too early. It seemed to start in October this year.

There are lots of reasons for this, but the one I think is the most interesting is the shift in Christmas shopping from offline to online. If we need to wait for deliveries, and the Royal Mail made sure that was the case this Autumn, then we need to remember to order earlier. So it seems likely advertisers brought everything forward a bit to allow for that.

Yesterday, I read at Gizmodo that Reevoo (a leading provider of online review services to retailers) is warning shoppers to be careful when relying on reviews they read on the Internet. This might seem like an odd thing for a provider of reviews to be doing, but Reevoo’s business model involves pooling reviews across all its customers, and building its own brand as a trusted supplier of independent information, perhaps a bit like Which? Magazine. So Reevoo’s message can be translated as “our reviews are independent and reliable, beware other peoples’”.

I wrote last week about the benefits of consumer reviews on e-commerce websites: they help with search engine optimisation, and there is evidence they improve average basket value, conversion rates and return rates for many retailers. There is a real concern about the extent to which some retailers might edit unfavourable reviews, while others at the low end might just make up good ones. If 6 out of 10 consumers are strongly influenced by consumer reviews, there’s obviously an incentive for some to improve, or even manufacture, better feedback than they would otherwise receive.

This issue is on a continuum of questions regarding online trust. At one end we see trusted, peer reviewed academic writing, at the other unauthenticated random messages on Twitter, Facebook and Bebo, which might be inaccurate, slanderous or even criminal in some cases - the name of someone who might have been involved in a murder briefly appeared in comments on a Youtube video last week, according to this article.

Clearly you shouldn’t believe everything you read, and this is why trust models like ebay’s have become so successful. Tim Berners-Lee addressed the question at his lecture at the BCS in March. There might be better technical solutions on trust soon, but for now it’s not surprising that people are questioning consumer reviews. If Wikipedia can be amended by the CIA, to suit their world view, we should be careful before trusting anything we read.

One of the powers of a brand is to create that kind of trust for the information it provides. Would John Lewis fix its reviews? The reason most people would say not is that their brand is so strong. And that’s based on decades of hard work.

I was at e-commerce expo in London on Tuesday, speaking to technology vendors, listening to presentations, all the usual exhibition things. The e-commerce space is very complex: if you examine a typical modern online store, it has many components, probably including the following:

• a core e-commerce platform (perhaps ATG, IBM, MS Commerce Server, Demandware) which can have many features including promotions, the basket and checkout functionality
• a merchandising tool, which the retailer uses to organise how products are presented to consumers, and increasingly merging with search (all those named above have one, but Mercado and Endeca are well-liked today)
• analytics and reporting tools like Coremetrics and Omniture (and again lots of others bundle this in)
• product imaging solutions like Scene 7
• product review features,  perhaps from Bazaarvoice, Reevoo or Feefo

… and these are just some of the options. Throw in order management, fraud detection and payment handling… the vendor landscape is very complex, and there is a healthy economy in figuring out the best combination of products for each retailer, implementing it, and integrating it with their back-end systems.

A presenter from one company was talking about behavioural targeting on retail websites, which his product did, and which involves detailed, specific configuration of web pages, content on those pages and even the different routes shoppers take, depending on what we know about them, and what they do. (This is a terrible definition of BT, but it’s the one he used). He said that using his company’s product meant shoppers got the right offer every time, leading to massively improved conversion, sales and profits.

Another presenter talked about how his company’s consumer review features were the most important contribution to his clients’ success: consumers loved to read what others had said, and higher sales, lower returns, better search engine optimisation all followed.

It’s clear that the claims of the vendors are largely contradictory. Buying all their products can’t be the answer. They can’t all be right! So where is a retailer to start? I think the answer is in thinking about the consumer: what customer journeys are best for the company’s product or service, for their consumers and ultimately for their retail performance? And then to design the experience around those insights, choosing the right combination of vendors to support the consumers’ journeys.

Another critical factor to consider is integration: buying a stack of products from different people is only going to make putting all the pieces together harder. There might be specific examples where buying products from multiple vendors adds value, but this needs to be balanced with the ease of implementing a suite that has been built to work together.

Finally, as one speaker was very keen to point out, renting a hosted solution, where you pay a monthly, quarterly or annual rent for a service rather than operating it yourself, makes many companies a lot more agile than they would be otherwise. In my management consulting days I would always have to remind clients to consider the “buy, build or outsource” question, and that question is even more important today than it was then, because what we can build for ourselves has become so much more complex.

Charles Tyrwhitt have just launched their new online store, which Agency.com helped to develop the user experience for. It was a collaboration with Javelin Group, and a host of e-commerce technologies are used, including:

• Mercado, for the structured product search and merchandising
• Scene 7 for the product image presentation
• Feefo for the user reviews of products

As an occasional wearer of suits and such like myself, I’m glad this site’s gone live, as it’s a great improvement on its predecessor and is applying a lot of current e-commerce best practice.

I’ve known the people at Emailvision for 6 years, and have worked with them on several occasions, as a client and as a consultant. The company has grown amazingly, especially over the last couple of years. This morning I was at the Institute of Directors for their breakfast seminar about version 6 of the Campaign Commander email marketing platform.

As they explained it, version 6 is about 3 things: easy integration with other web-based applications, better reporting and control over user rights and workflow.

Integration is all about APIs (the interfaces developers use to integrate applications with each other) and web services , a specific kind of API used for web applications. As they explained it, you no longer need to use the Emailvision front-end at all if you don’t want to: you can control and access anything Emailvision can do just by using their APIs. It’s unlikely that anyone is going to write their own user interface to Emailvision’s email platform: there’s a good one already. It does mean that most integrations you might like to do will be possible.

Suppose, for example, you’ve got an e-commerce website that’s been doing its own email marketing, but you’ve grown out of the built-in capabilities. Now you can choose between enhancing your own email platform or building an interface and using Emailvision’s. Given the amount of R&D they do each year, the idea of moving up to their platform could be a lot more attractive than attempting to enhance your solution. If you’re using Salesforce.com or Highrise for your customer relationship management, or Coremetrics for your analytics, you could easily benefit from this integration.

There’s something about this that reminds me of yesterday’s piece about Dopplr: building something very functional and making it easy to plug into (and well-documented) makes it much more likely that people will pick up your application and start doing interesting things with it.

Chris Combemale, Emailvision’s COO, and I were talking about APIs for the Emailvision platform in 2002, so it’s great to see them doing such a comprehensive job now, and placing it at the centre of their product strategy. And very web 2.0!

The reporting enhancements looked very nice, and I would expect so too! This is a very competitive area and keeping the reporting ahead of its rivals is a necessity.

Finally, on workflow and user rights, they’ve implemented controls and structures that remind me very much of those you see today in commercial content management systems. User access can be defined at the role level, and then users can be assigned to different lists (effectively databases) with very specific control over access rights. Different workflows for campaign development and execution can be set up, and these can again vary from list to list. It’s very powerful, allowing customers to set things up so that the CEO can run a dashboard report but can’t inadvertently push the campaign live. Agencies can be allowed to upload creative (and perhaps to test it) but not to create campaigns, if that’s how you want it.

My only reservation on this point is to ask whether they might open up the workflow and user architecture, so that big enterprises can use their existing LDAP or Active Directory for user access, for example. Might a corporate website built on Sharepoint 2007 link to Emailvision for its email campaign management, and use Windows Workflow Foundation to integrate web and email development workflows?

Finally, it’s interesting to see how many pre-configured packages of services Emailvision are offering. Having acquired Barnes & Richardson, the Brussels-based viral experts, last year, they’ve designed a full-service viral solution that looks very competitive. They also have post-sale engagement and up-sell solutions that would benefit a lot of retailers.