8 Best Practices To Reduce Technical Debt

When past actions in software development return to haunt you…

Is your business being bogged down by technical debt? Let’s look at measures that you can take to reduce it and scale your operations without the weight pulling you back. 

 

Work with a flexible architecture.

Right from the word go, you want to use architecture whose design is malleable, especially with the rapid rate of software evolution witnessed today. Going with an architecture that keeps calling for too much refactoring, or whose design won’t accommodate future changes will leave you with costly technical debt. Use scalable architecture that allows you to modify or add new features in future releases. While on this, complex features required in the final product should be discussed at the planning stage, that way simplified solutions that will be easier to implement can be identified, as this will lead to less technical debt in the long run. 

 

The Deal with Refactoring 

This is basically cleaning up the code structure without changing its behaviour. With the updates, patches, and new functionalities that are added to the systems and applications, each change comes with the threat of more technical debt. Additionally, organisations are increasingly moving their IT infrastructure from on-premises facilities to colocation data centres and deploying them on the cloud. In such scenarios, some workarounds are often needed to enable the systems to function in the new environments, which they hadn’t been initially developed to accommodate. Here, you will need to take some time to refactor the existing system regularly, streamlining the code and optimizing its performance – and this will be key to pay down the tech debt. When working with a flexible architecture from the start, the amount of work that goes into this will be reduced, meaning there’ll be less tech debt involved. 

 

Run discovery tests

Discovery testing essentially takes place even before a line of code is written for the system or application. This takes place at the product definition stage, where human insight software is used to understand the needs of the customer and is particularly helpful in setting priorities for the development work that will be carried out. It gives your business the opportunity to minimize the technical debt by allowing customers to give you a roadmap of the most pertinent features desired from the product. 

 

Routine code review

Getting a fresh look at the product or application from different sets of eyes in the development team will improve the quality of the code, thus reducing technical debt. There’s a catch though – this should be planned in a convenient way that doesn’t end up becoming a burden for the developers. Here are suggestions:

Break down pull requests

Instead of having complex pull requests where numerous changes in the code are introduced at a go, have this broken down into smaller manageable pull requests, each with a brief title and description about it. This will be easier for the code reviewer to analyse. 

● Define preferred coding practices

Documenting the preferred coding style will result in cleaner code, meaning the developers will focus their effort on reviewing the code itself, not losing time on code format debates.

 

Test automation

Relying only on scheduled manual testing opens you up to the risk of technical debt accruing rapidly, and not having sufficient resources to deal with the accumulated problems when they are identified. Automated testing on the other hand enables issues to be uncovered quicker, and with more precision. For instance, you can have automated unit tests that look at the functioning of the individual components of a system, or regression testing where the focus is on whether the code changes that have been implemented have affected related components of the system. However, establishing and maintaining automated testing will require quite some effort – making it more feasible for the long-term projects.

 

Keep a repository that tracks changes made

Do you have a record of changes made in the software? Keeping one in a repository that is accessible by the development team will make it easy to pin-point problems at their source. For instance, when software is being migrated to a new environment, or legacy software is in the process of being modernised, you will want to have an accurate record of changes that are being introduced, that way if there is an undesired impact on the system this it will be easier to zero-down on the cause.

 

Bring non-technical stakeholders on board

Does this conversation sound familiar?

Development Team: “We need to refactor the messy code quickly”

Product Team: “We have no idea what you are saying”

On one hand, you have the management or product team defining the product requirements, creating a project roadmap, and setting its milestones. On the other hand, there’s the software development/engineering that’s primarily focused on the product functionality, technical operations and clearing the backlog in code fixes. Poor communication between the two teams is actually a leading cause of technical debt.

For you to take concrete steps in managing your technical debt, the decision-makers in the organisation should understand its significance, and the necessity of reducing it. Explain to them how the debt occurred and why steps need to be taken to pay it down – but you can’t just bombard them with tech phrases and expect them to follow your thought process. 

So how do you go about it? Reframe the issues involved with the technical debt and explain the business value or impact of the code changes. Basically, the development team should approach it from a business point of view, and educate the management or production team about the cost of the technical debt. This can include aspects such as expenses in changing the code, salaries for the software engineers especially when the development team will need to be increased due to the workload piling up, as well as the revenue that is lost when the technical debt is allowed to spiral. 

The goal here is to show the management or production team how issues like failing to properly define the product requirements will slow down future software development, or how rushing the code will affect the next releases. That way, there will be better collaboration between the teams involved in the project. 

 

Allocate time and resources specifically for reducing technical debt

With management understanding that working with low-quality code is just like incurring financial debt and it will slow down product development, insist on setting time to deal with the debt. 

For instance, when it comes to the timing of application releases, meetings can be conducted to review short- and longer-term priorities. These meetings – where the development team and product team or management are brought together, the developers point out the software issues that should be resolved as a priority as they may create more technical debt. Management then ensures that budgets and plans are put in place to explicitly deal with those ongoing maintenance costs.

 

Retire old platforms

While most of the resources are going into developing new applications and improving the systems being used, the organisation should also focus on retiring the old applications, libraries, platforms, and the code modules. It’s recommended that you factor this into the application release plans, complete with the dates, processes and costs for the systems involved. 

 

Total overhaul

When the cost and effort of dealing with the technical debt far outweighs the benefits, then you may have to replace the entire system. At this tipping point, you’re not getting value from the technical debt, and it has become a painful issue that’s causing your organisation lots of difficulties. For instance, you may be dealing with legacy software where fixing it to support future developments has simply become too complicated. The patches available may only resolve specific issues with the system, and still leave you with lots of technical debt. Here, the best way out is to replace the system in its entirety. 

 

Final thoughts

Every software company has some level of tech debt. Just like financial debt, it is useful when properly managed, and a problem when ignored or allowed to spiral out of control. It’s a tradeoff between design/development actions and business goals. By taking measures to pay down your organization’s debt and address its interest as it accrues, you will avoid situations where short term solutions undermine your long-term goals. This is also key to enable your business to transition to using complex IT solutions easier, and even make the migration between data centres much smoother. These 8 measures will enable you to manage your technical debt better to prevent it from being the bottleneck that stifles your growth.

Check our similar posts

Becoming Nimble the Agile Project Management Way

In dictionary terms, ?agile? means ?able to move quickly and easily?. In project management terms, the definition is ?project management characterized by division of tasks into short work phases called ?sprints?, with frequent reassessments and adaptation of plans?. This technique is popular in software development but is also useful when rolling out other projects.

Managing the Seven Agile Development Phases

  • Stage 1: Vision. Define the software product in terms of how it will support the company vision and strategy, and what value it will provide the user. Customer satisfaction is of paramount value including accommodating user requirement changes.
  • Stage 2: Product Roadmap. Appoint a product owner responsible for liaising with the customer, business stakeholders and the development team. Task the owner with writing a high-level product description, creating a loose time frame and estimating effort for each phase.
  • Stage 3: Release Plan. Agile always looks ahead towards the benefits that will flow. Once agreed, the Product Road-map becomes the target deadline for delivery. With Vision, Road Map and Release Plan in place the next stage is to divide the project into manageable chunks, which may be parallel or serial.
  • Stage 4: Sprint Plans. Manage each of these phases as individual ?sprints?, with emphasis on speed and meeting targets. Before the development team starts working, make sure it agrees a common goal, identifies requirements and lists the tasks it will perform.
  • Stage 5: Daily Meetings. Meet with the development team each morning for a 15-minute review. Discuss what happened yesterday, identify and celebrate progress, and find a way to resolve or work around roadblocks. The goal is to get to alpha phase quickly. Nice-to-haves can be part of subsequent upgrades.
  • Stage 6: Sprint Review. When the phase of the project is complete, facilitate a sprint review with the team to confirm this. Invite the customer, business stakeholders and development team to a presentation where you demonstrate the project/ project phase that is implemented.
  • Stage 7: Sprint Retrospective. Call the team together again (the next day if possible) for a project review to discuss lessons learned. Focus on achievements and how to do even better next time. Document and implement process changes.

The Seven Agile Development Phases ? Conclusions and Thoughts

The Agile method is an excellent way of motivating project teams, achieving goals and building result-based communities. It is however, not a static system. The product owner must conduct regular, separate reviews with the customer too.

Migrating from CRM to Big Data

Big data moved to centre stage from being just another fad, and is being punted as the latest cure-all for information woes. It may well be, although like all transitions there are pitfalls. Denizon decided to highlight the major ones in the hope of fostering better understanding of what is involved.

Accurate data and interpretation of it have become increasingly critical. Ideas Laboratory reports that 84% of managers regard understanding their clients and predicting market trends essential, with accelerating demand for data savvy people the inevitable result. However Inc 5000 thinks many of them may have little idea of where to start. We should apply the lessons learned from when we implemented CRM because the dynamics are similar.

Be More Results Oriented

Denizon believes the key is focusing on the results we expect from Big Data first. Only then is it appropriate to apply our minds to the technology. By working the other way round we may end up with less than optimum solutions. We should understand the differences between options before committing to a choice, because it is expensive to switch software platforms in midstream. data lakes, hadoop, nosql, and graph databases all have their places, provided the solution you buy is scalable.

Clean Up Data First

The golden rule is not to automate anything before you understand it. Know the origin of your data, and if this is not reliable clean it up before you automate it. Big Data projects fail when executives become so enthused by results that they forget to ask themselves, ?Does this make sense in terms of what I expected??

Beware First Impressions

Big Data is just that. Many bits of information aggregated into averages and summaries. It does not make recommendations. It only prompts questions and what-if?s. Overlooking the need for the analytics that must follow can have you blindly relying on algorithms while setting your business sense aside.

Hire the Best Brains

Big Data?s competitive advantage depends on what human minds make with the processed information it spits out. This means tracing and affording creative talent able to make the shift from reactive analytics to proactive interaction with the data, and the customer decisions behind it.

If this provides a d?j? vu moment then you are not alone. Every iteration of the software revolution has seen vendors selling while the fish were running, and buyers clamouring for the opportunity. Decide what you want out first, use clean data, beware first impressions and get your analytics right. Then you are on the way to migrating successfully from CRM to Big Data.

Contact Us

  • (+353)(0)1-443-3807 – IRL
  • (+44)(0)20-7193-9751 – UK
Are Target Operating Models strategic compasses?

The short answer is they usually are, because every organisation needs a road-map of where they are going. Target operating models can be complex documents with illustrative details including project management structures, special tools, implementation procedures and management metrics. They can also be simple statements, as for example Winston Churchill?s promise that ?we shall fight them on the beaches, on the landing grounds and in the fields? which gave Britain the strategic direction it needed.

Many initiatives unfortunately fail because managers are ?too busy? to bottom on what their target operating model should say, or simply don’t believe in paperwork. As a result, promising initiatives may blunder off course or die a slow death without them really noticing. We cannot manage what we cannot measure, which is where the management metrics fit in. One of my favourite quotes is ?if you don’t know where you are going any road will get you there? which is what the Cheshire Cat said to Alice in Wonderland when she got lost.

The author blundered through life without a plan because there was no one else with his particular brand of imagination. The current business climate is different because everybody is trying to ramp up, and investors want to know exactly what is going to happen to their money and by when. Hence a target operating model can be indispensable throughout a change or product cycle.

The benefits of having a measurable operations / technology plan can produce powerfully tangible results if the organisation follows through on it. Built-in metrics with milestones are powerful tool for management, and, when they map through to the company financial plan almost irreplaceable as cash-flow forecasters.

Other benefits may include:

  • Shorter times to market and greater agility when launching new ideas
  • Reduced investor risk through a predictable process that’s readily monitored
  • A stable operating environment where there is consensus on direction
  • Greater likelihood of delivering on time and leading to repeat orders
  • A more cost-effective process, with less risk of loss of quality and money

Although it dates back a few years the Wills UK and Ireland Retail model still provides an excellent benchmark of a target operating plan that worked. The strategic goals were exceptionally clear, and they brought in a proven project manager to help them drive the program forward.

We have delivered advanced business management services to many of our clients, and believe you will find our personalised approach time-efficient and effective too.

Ready to work with Denizon?