Managing custom web development is managing a software project. More and more, average people in average businesses are managing software projects in the form of customized blogs, websites, and e-commerce shopping carts. Even more so, if they are trying to deploy a custom intranet application, for internal use by their company’s employees, or an extranet application, for use by their customers and corporate partners, they may find themselves faced with developing a fully custom software system.
For example, one project I recently worked on is a custom extranet application that a certain company’s clients use to submit jobs to the company. As each job request goes through the system, a number of people in the company need to add information and sign off on it. This automated system was thick with business rules for this particular company and as a result was almost completely custom, because no off-the-shelf software got even close to doing the job the way they needed it done. It was expensive for them to develop, but it allowed them to automate a process that previously had been manually intensive. (Before this, they would email Excel spreadsheets around. Now, the computer handles all the grunt work, and frequently, all a human has to do is click “Approve.”)
This is an extreme example, but more moderate examples also exist. For example, a Gilmore Girls fansite I put together years ago had custom database features in order to store memorable quotations from the show with fan commentary, linked to episode and topical guides. That’s something that has not to my knowledge been done before or since. It required custom programming and configuration to process user submissions, store the data, and display the data in the right format.
Unfortunately, if you’re facing this type of project, many of the consultants and service providers you talk to will not be aware of the software development issues involved and how to manage the software-development risk. They may try to shoehorn your problem into something off-the-shelf software can accomplish. Or they may just “hack together” custom programming, without managing the complexity of the underlying software design. Or they may want an up-front specification of everything your project needs to do and may want to spend months working out these details, under the guise of good planning. But how can you plan a project you have zero experience with? You can’t. And neither can they.
Now, this is not a problem if all you need is a stock installation of off-the-shelf software. For example, if all you need is a WordPress blog with off-the-shelf plugins and an off-the-shelf theme, the traditional approach is the right one. Figure out what you need, install it, make sure it works, and release it to the world. Even if all you need is a custom theme for a stock website, the traditional graphic design approach–3 mock-ups, choose one, then implement it in code–may be the best approach. Even if you have a semi-custom installation of a more complex web application, like a content management system or an e-commerce shopping cart, the traditional approach may be acceptable.
But the more customization your project requires, the more software engineering concerns will impact it, and the more ignoring these concerns will put your project at risk. This is especially important in a financial squeeze, where you need to maximize what you get for your software-development dollar and minimize the risk that the project will go off track or bust.
If you find yourself faced with this prospect, here are 13 “lucky” software development tips to planning and executing your next web development project, briefly:
- Resolve to take tiny steps. Don’t develop and deploy the system in one big leap. Rather, do it in small chunks so that you can monitor its progress and adjust its direction each step of the way.
- Find an expert you can trust, then trust him. Don’t just look for someone to bang out code for you, but for someone to advise you on technology and on the software development process.
- Don’t commit to which features you want until you need to develop them. Wait until the last responsible moment before committing. That gives you more time to gain the experience at each step to choose the most important features to add next.
- Make sure the contract allows for changes, and make changes before committing, but not after. Work with a development team who will allow you to make changes to the plan, up to the point at which they begin actually implementing your requests.
- Provide software requirements in the form of objective test procedures. Go step by step through a typical usage scenario so that the developers can understand what you envision, and so that your in-house testers can know that they implemented what you asked for.
- Set objective acceptance criteria for each feature. Be as flexible as you can without giving up your core requirements. Think through ahead of time what you really need and what you can get away with. Then talk to the developers about what they can do to give you the former.
- Ask the developers how long each feature will take to implement, and whatever they tell you, trust but verify. Many developers are over-optimistic in their estimates. So depend on their expertise in estimating time-to-completion, but verify it with your own measurements.
- Make an “expert user” available to advise the developers. Good developers will ask you questions about the system that you probably never thought of. Assign an “expert user” to answer these questions, someone in your company who can represent the users of the system to the developers.
- Get regular status updates. Have someone in-house look at each release. Meet with the development team leader every week to address issues. Stay in constant communication.
- Expect the possible, but not the impossible. If your developer says such-and-such a task will take a month, don’t argue him down to a week. Rather, talk about what parts of it you can leave out to reduce the time it will take.
- Be prepared to choose which features are more important, and which are not. If the developer asks you to choose which of two features is more important, don’t insist that both are critical and must be done next week. Make the hard choices. Pick one to do this week, and leave the other until next.
- Expect to fail in the short term, even if you’re optimistic in the long term. Microsoft didn’t get Windows right until version 3. You also may need to wait until the third monthly release before you see anything usable in your application.
- Don’t be afraid to pull the plug. At some point, you’ll find that all the most important features are already working and that the ones that are left aren’t actually worth what you’re paying for them. At that point, stop buying new features, and move onto the next project.
Not every development shop will be able to work within these bounds. These tips do come from many years of experience, failed projects, and successful ones. But there’s also a certain amount of faith involved, because despite the years of experience, we still have scanty scientific data to prove which software-development practices are most effective. That’s why some developers may not be used to working with the kind of process I’ve outlined here and may even refuse to. So you may need to shop around a little to find a developer who can. But at least now you know what questions to ask.