Guide · 5 min read · Updated
From prototype to product: how R&D work is run
The aim of R&D is to reduce uncertainty: to learn early and cheaply whether an idea will work. This guide summarises a way of working that starts from a question and reaches a product, along with common mistakes.
1. Write the question and the hypothesis
Good R&D starts with a question: "Is it possible to make this forecast with this data?", "Can this architecture carry this load?". Turn the question into a hypothesis that can be confirmed or refuted, and write down up front what result will make you say "it worked".
2. Build the smallest working example
The goal is not a pretty demo but learning. Reduce the scope to the smallest piece that will answer the question. Getting a result in a short time (days or a few weeks) prevents long progress in the wrong direction.
3. Pick the right kind of experiment
- Short spike: a quick answer to a single technical question.
- Prototype: seeing how an idea feels to a user.
- Proof of concept (PoC): showing that it is technically possible.
- Pilot: trying with real users and data for a limited time and scope.
4. Measure and document the result
Whether it succeeds or fails, record the result together with the data and conditions used. A negative result is valuable information that prevents repeating the same attempt.
5. Turning a prototype into a product is a separate job
Because a prototype is written fast, it may fall short on tests, security, scalability and maintenance. When carrying a successful idea into a product, rewriting it with a clean architecture is usually cheaper in the long run.
Common mistakes
- Starting an experiment without writing the question.
- Mistaking a prototype for a product.
- Setting no success criterion.
- Keeping the scope too wide.
- Not recording negative results.
Conclusion
VeriSet A.Ş. researches new technologies, tests them with prototypes and builds solutions that can be used in the future. For more, see our R&D area page.
More guides
7 questions before starting an AI project
Most AI projects struggle not because of the model itself but because of questions left unanswered at the start. This guide covers seven questions to ask before writing code and why each one matters.
Read the guideData technologies · 6 min readHow to design a data pipeline
A data pipeline is a chain of automated steps that moves data reliably from source to target. A well-designed pipeline delivers the data that analyses and AI models need correctly, on time and in a traceable way.
Read the guideMobile technologies · 6 min readNative or cross-platform? Choosing the right approach for a mobile app
One of the first questions in mobile development is whether the app is built separately for each platform or from a single codebase. The right answer depends on your product's performance expectations, budget and how much it needs device features.
Read the guideWeb platforms · 7 min readWeb performance and Core Web Vitals: where to start
Core Web Vitals are three metrics Google uses to measure real user experience: loading speed, interaction responsiveness and visual stability. This guide summarises what the metrics mean and where to start improving.
Read the guideInfrastructure · 6 min readMonolith or microservices? An architecture choice guide
Microservices are popular but not right for every project. The choice of architecture should be made by team size, product maturity and operational capacity. This guide summarises four common approaches and when each fits.
Read the guideHave an idea in this area?
Briefly describe your goal and where you are today, and we will take a look together.