If you disappeared for a week, could the people using your database keep the business running without you? Or would your phone start ringing the moment you got comfortable on the beach? That is one of the best tests you can give any business application: can ordinary day-to-day work continue when the developer is unavailable?
A database is not finished just because it works on your computer while you are sitting there to explain every button, repair every bad import, and answer every "what do I click next?" question. Good software should make people's jobs easier. It should not turn the developer into a permanent member of the staff, on call forever.
Back when I was doing consulting work, I would often walk into businesses with information scattered all over the place. There might be Excel spreadsheets, Word documents, paper forms, and sometimes an old DOS program that only one employee understood. Yes, DOS. I am old.
Building an Access database meant learning how that business actually operated, bringing those disconnected pieces together, and giving the employees a system they could use. But the job was not done when the database technically worked. It was done when the people using it could work confidently without me standing behind them.
That leads to a simple design principle: build software that explains itself. Buttons should say what they do. A caption like "Run Monthly Sales Report" is much better than "Button 17." Use Control Tip Text when it helps. Put short instructions directly on forms where users need them. If a process is complicated, include a Help button that opens information relevant to that specific screen.
Do not expect someone to remember a six-step procedure they only perform once a month. If the database needs a monthly import, report generation, archive process, or other occasional task, guide the user through it. Better yet, automate as much of it as you reasonably can.
Error messages are another big one. A message such as "Error 3022" may mean something to an experienced Access developer, but it means absolutely nothing useful to the person trying to enter a customer order. Tell the user what happened in plain English, what it probably means, and what they should do next. For example, explain that they entered a duplicate value where a unique value is required, and tell them which record or field needs attention.
One of the best ways to find weak spots in an application is to sit down with someone while they use it. Then keep your mouth shut. Let them perform their regular tasks without coaching them through every screen. Every time you feel the urge to say, "Before you do that, you need to..." you have probably found something that needs to be improved.
Maybe a button is confusing. Maybe instructions are missing. Maybe the form should automatically fill in a value, verify a condition, or prevent the user from taking an invalid step in the first place. If you have explained the same thing to three different people, the odds are good that the problem is not those three people. The interface is trying to tell you something.
Another common mistake is building a separate report for every question a customer asks. At first, that seems reasonable. They want a report for this month, then another one for last month, then one filtered by department, employee, customer, or date range. Before long, you have a database full of nearly identical reports and a client who still needs to call you whenever they think of a new variation.
Whenever it makes sense, build tools instead of just one-off reports. Give users a form where they can choose their own date range, customer, department, status, or whatever criteria are appropriate. A flexible reporting form can answer dozens of routine questions without requiring dozens of report objects and a support call every few months.
Sure, if you bill by the hour, repeat calls for tiny changes may sound like job security. I always preferred building a better system up front. The customer gets software that can grow with the business, and you get fewer interruptions when you are trying to enjoy your vacation. Everybody wins.
Even a well-designed database needs some documentation, but that does not mean you need to write a 300-page manual that nobody will read. Start with the practical information people need: which file they should open, where the data is stored, what must happen before monthly reports are generated, and what an import process does.
Pay particular attention to processes that can cause trouble when repeated. If running an import twice could duplicate records, users need to know that. Better yet, design the process so it can detect whether it already ran and protect users from making that mistake. A warning is helpful. Prevention is better.
For your own development notes, keep track of important objects, business rules, unusual design decisions, and anything a future developer would need to know. A simple spreadsheet is often enough. The documentation for users, however, is usually most useful when it is built directly into the application, right where they need it.
Also, every captain needs a number one. If you are the only person in the organization who understands the database, that is a risk. Find someone who is reasonably comfortable with computers and teach them the basics. They do not need to become an Access developer or learn VBA. They should simply know how to handle routine issues, confirm that regular processes have completed, locate basic instructions, and recognize when a problem really does need technical help.
This becomes especially important if you built the system and you maintain it too. What happens if you get sick? What if you have an emergency? What if you decide to extend your vacation because a couple of Ferengi talked you into investing in beachfront property? The business still has to function.
In my own business, I have systems that need to keep operating whether I am at my desk or away from the office. That means thinking about automation, error handling, logging, backups, and notifications. When something goes wrong, the system should either recover gracefully or provide enough useful information that I can understand what happened without remotely babysitting it from a hotel room.
You cannot anticipate every possible problem. Hardware fails. Networks go down. Users will always discover creative new ways to break things. But you can eliminate a surprising number of support calls by asking a few questions while you are building the application.
Can users recover from common mistakes? Can they tell what a button does? Can they get the information they need without asking you to modify a query? Does someone get notified when an automated task fails? Is there a reliable backup plan? These are design questions, not afterthoughts to consider while you are packing your suitcase.
Although I am talking primarily about Microsoft Access databases here, the same idea applies to Excel automation, SQL Server applications, websites, and just about any custom business software. The goal is not merely to build something clever. The goal is to build something understandable, dependable, and useful to the people who rely on it.
Try the seven-day test before your next vacation. Pretend you will be completely unreachable for a full week. No phone calls, no email, no remote desktop sessions from the hotel. Can the office keep doing its job? Will routine operations continue? Does someone know what to do when something unexpected happens?
If the answer is no, that is not necessarily bad news. It is a very useful list of improvements waiting to be made. A database that works while you are there supervising it is one thing. A database that keeps the business moving while you are relaxing on a beach is something else entirely.
Watch the embedded video for more thoughts and real-world examples. Then ask yourself a simple question: if you disappeared tomorrow, what would your users get stuck on first?
Live long and prosper,
RR
No comments:
Post a Comment