The scourge of Document Driven Design

Documents, or rather words are the scourge of product design. Because words can never convey the true meaning or emotion of what is really required. All to often, software development projects are driven by the documentation – agile projects can be equally guilty of this- driven by words on paper (or card) that convey what the requirement is. Issue #1. Developers don’t read!

Except for a few exceptions, software is all about a user interacting with a system in order to accomplish a goal. Trying to describe that premise in terms of features and functionality is fundamentally flawed as it will be nested in the language of technical implementation, not in the user interaction. “Find” becomes “search”, “buy” becomes “shopping cart”, “check” becomes “validation” and so on. Issue #2. The desired outcomes become lost in a smog of technical jargon.

Solution: Picture Driven Design. Up front. Yeah! I’m all for up front design! A picture tells a thousand words. It has the power to remove ambiguities, clarify the vision, showing what the journey to realise outcomes look like. Start with a day in the life of… scribble out the flow, the user journey. Nothing complicated, boxes and arrows. Then scribble out what the user interface might look like. Done. That’s your up front design. That’s your documentation. That’s your scope. What you do next is up to you, write loads of documents that describe it and produce the software in a waterfall way if you want. I’d prefer you were more lean and adopted agile practices, but whatever you do, start with the picture. I’m convinced it will save much pain later on.

Bag of risk

I’ve only thought of blogging about Lois Vuitton once before and that was on how they positively encourage queueing outside their stores during busy periods. It’s a pretty strong brand that can tell its customer to hang about before being allowed to come in and shop.

This time I’m not blogging about them in a positive light, and nor are many others. Jeremiah Owyang describes the situation they are in well. Their brand has been hijacked by Nadia Plesner, an artist trying to raise awareness about Darfur and how the media considers Paris Hilton with her “designer bags and ugly dogs” to be more worthy of attention than genocide in Darfur. She uses an image of a LV bag in her T-shirts. LV take offence and sue, she refuses to budge and suddenly the image, the issue and LV all hit the spot-light. And in this David and Goliath contest, who is going to come out worst? There can only be one looser.

So why didn’t LV just ignore it, or even as Jeremiah suggests, harness the issue, turn it into a conversation that would paint them in a good light? I’ll argue that it is because they don’t understand risk.

There was always a risk to the brand be de-valued by being associated by asociation with Dafur. And this is what the marketing and legal team jumped on with such zeal. Did no-one think about the risk to the brand of turning this into the issue it has become on the web? Laying out the options and doing a risk analysis would have been a worthy exercise.

Option 1. Assess the global impact of nadia plesner, assume it is minimal and do nothing. Risk to brand: minimal.

Option 2. Follow standard route of brand defamation and sue. Ignore association with ‘good cause’, ignore blogosphere. Risk to brand: potentially significant.

Sadly, it seems that LV ignored the whole concept of risk and went with the default option – sue. They are not alone in failing to assess the risks properly before pursuing a course of action. In IT this approach is endemic. Where is the greater risk? Placing all your eggs in one basket, investing heavily in a desired outcome that will be many months before it sees the light of day. Or take a more gradual approach, investing ‘just enough’ to get ‘just enough’, ‘just in time’. The latter approach is lean and agile. A good agile project is a lesson in risk management, building resillience into the process and testing options as you go. It is organic and evolutionary, (rather like nature), as opposed to the plan and control approach of waterfall which is brittle and will struggle to react to or accommodate risk appropriately. I should write more but there is a day’s work ahead.

Agile Hong Kong

Exciting times for us in Hong Kong. The inaugral meeting of Agile Hong Kong is next Tuesday, 5th February. Martin Fowler will be attending, hosting an informal Q & A session. It’s sponsored by ThoughtWorks who will be providing the drinks. Check out for more details.

Was it just a simple database query?

So the sensitive personal details of 25m people has been lost and there is a huge political furore over it. Whose fault is it? As far as I can see, (and this is my personal opinion,) blame must lie with IT, specifically the IT contractor and either the contract they work with or the perception of that contract.

The National Audit Office asked HM Customs and Excise for child benefit in “desensitized form”. Sensitive details were specifically asked to be removed, ostensibly to make the file size smaller. This would require a bespoke query to be run. It was deemed too costly so it was assumed that a full extract of the data would do. The fact that this was then burned to a CD, posted unregistered mail and lost is not the point (that is stupidity). What is the point is the IT contract prohibits the business (in this case the governmental offices) to do their job properly.

What sort of contract demands extra payment for a simple database query for “NI numbers, child benefit numbers and children’s names in order to select a risk-based sample of cases to audit as part of anti-fraud work“?

Surely this is an extra request that an experienced database analyst could easily run in the course of a day? If not you must ask why not – is it because the database is badly designed with nested tables and stored procedures and stuff that would make a decent DBA go eugchhh (I’ve seen that happen). If this is the case, the IT contractor has done a bad job; if an electrician worked in your house and left a mess of an electrical installation, would you keep employing them, even if they were cheap?

Maybe however it would not have incurred a cost and this was just the perception; “we must not… run additional scans/filters that may incur a cost to the department”. If this was case it suggests a breakdown in the relationship between the business and IT, with tendency towards the confrontational and transactional rather than co-operation and partnership.

Organisations that outsource their IT often fail to realise what the true costs are. Anything outside the terms of the contract is a change request. It is not unusual for the request itself to incur a cost (someone has to write the documentation, specifiy the design, estimate the effort) before a line of code is written. (At one organisation I worked with that had outsourced their IT function, I was told that to add some basic client-side field validation to a single field on an application form on their website was likely to cost in the region of £60k). The business starts to believe that everything costs and IT becomes a hindrance and a vicious cycle commences.

How could things have worked differently? Let’s say the HMRC IT department was run on more lean/ agile lines. With agile it embraces change. The request comes in (let’s assume such requests are not regular occurances) and in the morning stand-up the BA describes the request and asks the developers for its feasibility in a word. Someone says “yes, I ran a simiar querry last month, it’ll take me ten minutes”. (In reality double or treble that estimate), but it will not have an imact on the developers ability to get thier prioritised work completed. Alternatively the developers say “given the database structure we have inherited that’s a lot of effort” or the project manager says “another request?! pritoritise it like the others!” and it is prioritised in the weekly iteration planning meeting (pushing something else out) and then it gets done.

My hope is that when the inevitable investigation takes place, they don’t just look at the policies and procedures, but also at the underlying structure of the way that IT is managed.

Consistency when things are poor

Call it a pattern, a heuristic or a rule of thumb. A fundamental one of those to ensuring usability is consistency. This may be external consistency – for something behaves in the software in a similar way elsewhere. A good example might be the ‘x’ button in the top right hand side of an open window. It is universally a call to action to close the window. If the designer created a button labelled “C”, and placed it on the left hand side this would result in confusion. It is not consistent with the users’ expectations from using other applications. The second type of consistency is internal – do things behave in a consistent manner throughout the application? This may be both in behaviours (e.g. buttons with the same titles perform the same action), and in look and feel – a website has a visual identity and coherence, assuring a continuity of experience.

There may be examples of where internal consistency is not possible. For example a brochureware site that uses a third party for fulfilment or payment. Paypal is a good example of this – the user is taken out of the shopping experience and into a paypal experience. This can be successful if there is clear signposting and use of the paypal imagery on the shopping site to assure the user.

So what happens when you have a large, legacy website that you acknowledge to be pretty poor in the usability front and want to introduce new functionality, or want to rebuild it. If you play the consistency card too strongly you may continue to be consistent with the old design and behaviours. This begs the question, is it better to introduce something that is internally inconsistent, but fundamentally better? This becomes even more an issue when you look to rebuilding your site in an incremental fashion.

As an information architect I can help you design your site architecture – the look and feel, navigation structure, user journeys etc., but this will probably be in its entirety. To build this new site will take time, and assumes a “big bang” whereby a completely new site will be (re)launched. Yet there are probably business imperatives to fix specific areas. If we build in an incremental fashion, and take the agile approach of focussing upon delivering business value, we are not going to have a fully redesigned site to go live with. We are probably going to have (for a short while at least), some parts of the site that are new and some that are old.

Going back to my original question, we can either build this to be consistent with the old site, or do something tangentially better. If we do the later it will probably be significantly inconsistent from other parts of the site, or the original parts of the site. It is in this scenario that I am inclined to throw away the consistency pattern. You may have internal inconsistency if you have a clear roadmap to throwing the old and the new functionality / design is proven to be usable, accessible and intuitive. With this the case, the interaction behaviour and visual identity of the new functionality must become the benchmark to which future functionality is consistent with. And you must clearly signpost to the user what is going on; customers will generally be forgiving if they understand that the changes are in their interests.

Software Dams and recipient participation

There once was a time that international development was all about big capital projects, building dams and the likes. Times change, now the focus is on eliminating poverty; DFID “focus [their] aid on the poorest countries and those most in need”. There is a realisation that those big projects did very little to address poverty, indeed they kept countries poor forcing them into debt (read Confessions of an Economic Hit Man for a cynical view of this). And besides, dam projects are rarely successful and before you know it they silt up.

A focus on reducing poverty requires a new approach. It requires an understanding of the root problems, it means spending time with the poor to understand their circumstances to be able to create appropriate and sustainable solutions rather than prescribed programmes that develop and maintain a dependency culture.

There are parallels here with the IT industry. Much of the IT game remains focussed upon those big projects. Software dams that can be launched with great fanfare but do little sustainable good to those most in need. The customer.

Before I wound up in IT I worked in international development. My PhD. “Ergonomics tool and methods for use in Industrially Developing Countries” was based on working with farmers in Sub Saharan Africa, looking at how technology is transferred and how it can be made more appropriate, sustainable and usable. Many of the tools and techniques I used in the bush I apply with the corporations I work with today. These came under the umbrella of “Participatory Technology Development” and “Participatory Rural Appraisal”.

Rather than the delivering the white elephants of expensive machinery that you see littered around Africa, Participatory Technology Development is an approach for developing simple low cost innovative solutions that have the ownership of the community who work with researchers to build them. The PTD framework starts with gaining a shared understanding problems and opportunities. This is followed by defining priority problems then experimentation. Experimentation is collaborative with options derived from indigenous knowledge and support from the researchers experience and expertise. The farmers own the experiments and the results. This leads to the next step of the framework; sharing the results with farmer led extension. (Traditionally dissemination of agricultural advice is done by agricultural extension officers – government employees who despite their best intentions preach too the farmers, sharing centrally defined agricultural advice rather than the more appropriate, locally developed technologies that the farming community have developed themselves). The final step to the process is the researchers withdrawing, leaving the community with the capacity to continue the process of change.

(Sounding like agile?)

If PTD is a framework, then PRA is a basket of tools and techniques that can be used to support it. These can be broken down into nine categories:

  • Secondary data reviews – reviewing existing sources of information
  • Workshops – getting key stakeholders round the table (or more appropriately under the banyan tree)
  • Semi-structured interviewing – talking to people with a loose conversation direction
  • Ranking and classification techniques – identifying “things” and ordering them according to different criteria. (Often this will involve moving pebbles around boxes drawn in the sand).
  • Diagramming, illustrations and graphics – pictures to convey ideas and concepts, through “boxes and arrows”, Venn diagrams and charting to cartoons and imagery
  • Mapping – drawings or models that represent the local environment
  • Structured observation – watching people doing
  • Timelines – What happens when, for example seasonal calendars, a line in the sand and people put pebbles down against the time to show when crops are sown, harvested, how the price fluctuates, labour migration etc.
  • Community meetings – meeting the whole community rather than just the immediate stakeholders who participate in stakeholders. Showcases?

Are you building a Software Dam? Or are you focussing your aid on those most in need? PTD and PRA are approaches that have developed to help introduce appropriate, sustainable improvements to the life and wellbeing of subsistence farmers. Much of their content can be transferred to IT projects, helping introduce appropriate, sustainable improvements to the life and wellbeing of customers / users.

Ditch the feature shopping list. Think customer journeys.

Let’s imagine a mobile phone provider that that has an on-line presence as well as a high street retail network. Their current website was built several years back on legacy technology; it is slow and has a conversion rate lower than industry norms. Along with having poor usability, the current shopping cart functionality does not support the concurrent usage figures that are hitting the site. The business takes the output from their web metrics and throw these at IT and demand improvements. And a new IT project is born. Rebuild the existing site on a new platform. They get a design agency to handle the look and feel. The functional requirements are built upon the existing functionality and chunked into functional silos:

  • Content managed brochureware site
  • Phone selection
  • Tariff selection
  • Shopping cart
  • Order management

A typical project process. That is flawed. It is starting with a functional premise. An alternative is to think in terms of the customer and customer journeys.

We can start by asking who the customers are. Almost certainly the marketing department has a customer segmentation model – this is a good place to start. That may give us basic attitudinal and behavioural details on customers, but we need richer data. How do customers shop for phones? So we go and spend time in the stores and observe how people buy them. It soon becomes clear that the choose phone – choose tariff buy model is a journey that is only taken by a certain number of customers. Other customers come into the store with intentions – they don’t know what phone or tariff they want, but they know what they want a mobile phone for. Other customers come into the store asking lots of questions, they are doing research; they leave the store with the sales guy’s number, and costs for a couple of phones and tariffs. We look at the competition and see how they sell phones. We look at Amazon and eBay and expedia and see how other retailers sell products and experiences on the web. We synthesise our research and suddenly we have a whole bunch of new requirements. Gasp! Scope creep. Indeed if we list them out into functional silos again:

  • Content managed brochureware site
  • Lifestyler phone selection wizard
  • Save a quote
  • In-store quote save for on-line fulfilment
  • On-line purchase for store fulfilment
  • Phone selection
  • Tariff selection
  • Shopping cart
  • Order management
  • Etc

The business is excited and IT is despondent. When the business announce the date all this must go live by, the letters of resignation land on the IT directors desk and they start looking for contractors. It does have to be this way. You can have your cake and eat it.
Rather than thinking about vertical functional silos, we should think about horizontal slices through the functionality. Slices that support customer journeys. We can start with simple journeys to start and build complexity as we prove our process and new platform, mitigating technical risk as we progress.

don't functionally silo, slice by customer journey

The first release probably needs to demonstrate the new platform works: we can deliver on time; that a new shopping cart interfaces with our legacy warehouse order management system; etc. What’s the bare minimum we can do that does this and delivers business value. How about a microsite that sells a single product. Customers are directed to the microsite via a URL published on a flyer handed out in the stores.

As a customer who has picked up a special offer N80 deal flyer, I want to buy a Nokia N80 on a pay as you go contract

Our acceptance criteria:

Given I enter my personal details and credit card details When my credit card details are validated Then send order to warehouse to dispatch phone and activate contract.

We can decompose this requirement into discrete requirements, “stories” of sufficiently small size and estimate them. We’ll soon get a project velocity and because it is only a small release expectations, risks and dependencies will be easy to manage. We’ve not had to wait for months to get something out, everybody is happy.

We identify that a profitable segment of the market are the aspirationally clueless. People who want a mobile phone, realise they are a minority who don’t have one but are too afraid to admit they have no idea what they want. So we build a new customer journey.

As an aspirationally clueless I want say what I do and how I live my life and have a suitable phone selected for me.

OK, not the finest story, but you get the point. This story might take elements of the phone selection and tariff selection functional silos, but just enough to realise the required functionality. Because we are building around customer journeys, just enough to realise customer intentions that support those journeys we build only what is required. We are not building by feature list. We don’t over promise and under-deliver. We are building trust with all stakeholders. Surely this is the better way to build software, thinking about customer intentions and incrementally delivering to support more complex customer goals. Sadly, people all too often get fixated by feature lists. Because that is what they are used to. Because that is how products are sold. But isn’t there a marketing 2.0 where we sell experiences and go beyond the product. Isn’t that what Virgin does? But I’m probably off on a tangent.

4 of 4