How can organizations reduce costs while still delivering real value to their customers? This is a question I get asked frequently, and one that a client recently brought to me when they wanted a keynote for their solution architects. In this talk, I walk through the key principles and practical techniques for continuously delivering value while cutting unnecessary spending.
80% of Features Are Rarely or Never Used#
Let that number sink in. Studies consistently show that roughly 80% of features in an average software product are rarely or never used. On top of that, the maintenance cost of any feature is an additional 40 to 80% of the original development cost over its lifecycle.
Let me give you a quick calculation. If you develop 10 features at 100,000 Swiss Francs each, your total development cost is 1 million. Add maintenance at 40 to 80%, and you are looking at total costs between 1.4 and 1.8 million Swiss Francs. Since 80% of those features are rarely used, the potential savings in development alone are 800,000 Swiss Francs, plus 320,000 to 640,000 in maintenance savings. That is over 1 million Swiss Francs in total savings.
“Doing the wrong thing right is not nearly as good as doing the right thing wrong.” — Dr. Russell Ackoff
Building the Right Thing Right#
There are two dimensions to delivering value efficiently. Building the right thing is about effectiveness: aligning your solutions to real business and user needs. Building the thing right is about efficiency: applying solid software engineering practices to deliver quickly and maintain easily.
You are only successful when both come together. And if you think this is not your responsibility because “the business ordered it,” remember: if you run with the crowd, you share the blame. It is your company, and features that are never used still affect the bottom line.
From Projects to Products#
Part of the problem is that many organizations still think in projects instead of products. A project has a fixed scope, a fixed budget, and a set number of features to deliver. It maximizes output: the number of things produced. A product, on the other hand, focuses on outcomes: solving real customer problems and changing behavior.
When we move from project thinking to product thinking, we stop asking “How many features can we deliver?” and start asking “What problem are we solving?”
The Lean Startup Cycle#
The solution is to use the Lean Startup cycle. We acknowledge that we have many ideas, most of them not great, and we build minimal viable products (MVPs) to test our hypotheses. I recommend using the SAFe epic hypothesis statement format, which includes an elevator pitch, a business outcome with measurable benefits, leading indicators for early validation, and non-functional requirements.
Start with the customer and the problem to solve. Do not start with the technology and work your way back to the customer.
I walked through a practical example: a Swiss dentist company wanting a mobile app for scheduling appointments. Using paper prototypes, they discovered customers did not want another mobile app. They pivoted to a web application. By testing early and cheaply, they avoided building something nobody wanted.
Making Cost-Effective Decisions with MCDA#
When you have multiple options in your solution space, use the Multiple Criteria Decision Analysis (MCDA) technique. It is simple, powerful, and I use it nearly every week.
The steps are straightforward: define your objective, list the criteria with weights (cost effectiveness at 30%, scalability at 20%, etc.), list your options (monolith, SOA, microservices, serverless), rate each option, and calculate weighted totals. The key insight is that there is no silver bullet. Every solution has disadvantages that need to be mitigated. MCDA makes those trade-offs visible and deliberate.
Value Stream Mapping: Understanding the Whole System#
To identify and reduce waste, you need to look at your organization as a sociotechnical system. Processes, technology, and people must work together like a well-maintained factory. This requires whole-system thinking.
A value stream is a sequence of steps that delivers value to the customer. You have operational value streams (starting and ending with the customer) and development value streams that support them. By mapping your development value stream, you can measure process time, lead time, and the percentage complete and accurate at each step.
In one example I showed, the total lead time was 2,160 hours with only 7% of that time spent on value-adding work. By identifying bottlenecks (testing had 8 hours of process time but 336 hours of lead time) and applying automation, the future state could reduce lead time to 147 hours. That is a massive improvement in time to market, and faster time to market means faster value generation.
What About AI?#
Yes, AI can help across the value stream. But do not start with the technology. Start with the problem. If you can remove or simplify a process step, that is far better than replacing it with an AI. And remember: every AI agent you build will need maintenance, consuming that same 40 to 80% of development cost over its lifecycle. Always find the simplest solution first.
Key Takeaways#
- Build the right thing right: Focus on effectiveness first, then efficiency
- Use the Lean Startup cycle: Test hypotheses early with MVPs and paper prototypes
- Focus on outcomes, not output: Move from project thinking to product thinking
- Use MCDA for decisions: Make trade-offs visible and mitigate disadvantages deliberately
- Map your value streams: Understand how value flows, identify bottlenecks, and improve the whole system
- Start with the problem, not the technology: This applies to AI as much as anything else
- Remember the maintenance cost: Every feature, every AI agent needs ongoing care at 40 to 80% of its development cost