"Can't we do something about this system?" — Even with such voices echoing across the company, years go by without anyone touching it. This is a common sight in the day-to-day operations of small and medium-sized enterprises.
Data transcription that accounting spends 30 minutes manually entering every morning, inventory information that sales cannot check on the go, and operating procedures that newcomers still cannot master after three months. Even when everyone knows the root cause is the "antiquated business system," it is only natural to freeze when told that "rebuilding it will cost tens of millions of yen."
Yet in reality, methods exist to improve just the usability without rebuilding the system from scratch. This article explains five practical approaches to improving only "the parts users touch" without altering the backend.
"Hard to use" is the cost of inaction — Losses legacy system UX issues inflict on businesses
"It's running, so it's fine" — this is a common justification heard for continuing to use legacy systems. However, "running" and "easy to use" are entirely different matters.
In a 1,000-person survey conducted by baigie in 2024, "unintuitive operations" and "complex input fields" ranked high among dissatisfactions with business systems. Such UX problems steadily erode corporate earnings as hidden costs.
Concrete losses generated by difficult-to-use systems:
| Loss Category | Specific example | Severity of Impact |
|---|---|---|
| Increased work hours | Double data entry, excessive screen transitions, poor searchability | Up to 90 minutes lost per employee per day (according to Smashing Magazine) |
| Increase in human error | Input errors, transcription errors, missed checks | Occurrence of rework hours plus customer support costs |
| Lower retention rates for new hires | Takes time to learn, manuals rely on tribal knowledge | Increased hiring and training costs |
| Delayed decision-making | Data extraction takes time, lack of real-time capability | Opportunity loss, falling behind competitors |
| Employee dissatisfaction and turnover | "Why are we forced to use a system like this?" | Decline in engagement |
These do not appear directly on financial statements, but it is not uncommon for them to amount to annual losses in the millions or tens of millions of yen. UX issues are not something that can be dismissed with "just put up with it"; they must be recognized as an executive management issue.
No full replacement required for legacy systems — The option of UX-layer improvements
People often think "old system = rebuilding is the only option," but a full replacement carries significant risks and costs.
| Comparison item | Full Replacement | UX-Layer Improvement |
|---|---|---|
| Cost | 20 million to over 100 million yen | Around 1 million to 5 million yen |
| Duration | 1 to 3 years | 1 to 6 months |
| Business disruption risk | High (data migration and parallel runs required) | Low (existing system remains operational) |
| Internal resistance | High (operations change completely) | Low (can be improved incrementally) |
| Loss upon failure | Devastating (investment cannot be recovered) | Limited (reversible) |
| Speed of realizing results | Slow (no results until completion) | Fast (immediately noticeable even with partial improvements) |
A special feature article in Smashing Magazine in April 2026 also recommended "revamping the frontend" as the first step in legacy system improvement. This approach improves only the UI layer that users interact with daily—such as screens, input forms, and report outputs—without touching backend logic or databases.
The essence of this mindset is not to "make the entire system perfect," but to "first fix only the parts where users are struggling." From a return-on-investment perspective, studies also show an average return of $100 for every $1 invested in UX improvement.
Five approaches to legacy system UX improvement
1. Revamping the UI frontend (keeping the backend as is)
The simplest and most effective method is the approach of rebuilding only the frontend (screens) using modern technology. It integrates with the existing backend via APIs or database connections, renewing exclusively the screens operated by users.
Specific improvement examples:
- Splitting complex input forms into step-by-step formats
- Enhancing search capabilities (fuzzy search, filtering)
- Making screens responsive so they can be viewed on tablets and smartphones
- Improving accessibility through better font sizes and contrast
Application conditions: The backend must be relatively stable and capable of retrieving and updating data. Even if no API exists, there are cases where direct database connections or screen scraping can handle it.
Estimated cost: 1.5 million to 5 million yen / Duration: 2 to 6 months
2. Adding dashboards and portals
This method involves building a separate dashboard that provides an at-a-glance view of information required for daily work while leaving existing system screens intact. It consolidates data scattered across multiple systems into a single screen, eliminating the hassle of logging into various systems to look for information.
Specific improvement examples:
- KPI dashboards providing a single-screen view of sales, inventory, and order statuses
- List displays of tasks awaiting approval
- Alert features (notifications for out-of-stock items, delivery delays, etc.)
Application conditions: Means of retrieving data from each system (APIs, CSV export, DB connections, etc.) must exist.
Estimated cost: 1 million to 3 million yen / Duration: 1 to 3 months
3. Automating operations with RPA and macros
This approach automates repetitive operations using RPA (Robotic Process Automation) or macros without altering the appearance of the screens. It delegates tasks that do not require human effort—such as data transcription, report generation, and routine email sending—to machines.
Specific improvement examples:
- Automating data transcription from core systems to accounting software
- Automatically generating daily morning sales reports and sending them via email
- Automating input checks for order data and providing immediate error notifications
Application conditions: Work with many routine, repetitive operations. Screen interactions must be standardized (unsuitable for tasks where operational steps vary each time).
Estimated cost: 500,000 to 2 million yen / Duration: 2 weeks to 2 months
4. Exposing APIs and adding mobile-friendly frontends
This method exposes legacy system data and features as APIs, making them accessible from mobile and web apps. It satisfies needs such as checking inventory or performing approval processing on smartphones while out of the office.
Specific improvement examples:
- Sales representatives checking inventory and delivery dates immediately on smartphones
- Managers processing expense and proposal approvals while on the go
- On-site workers submitting work reports via tablets
Application conditions: The backend must have an interface that can be turned into an API (or an intermediate layer can be constructed). Security requirements must be clarified.
Estimated cost: 2 million to 6 million yen / Duration: 3 to 6 months
5. Integrating AI assistants (introducing natural language input)
Attracting attention as a 2026 trend is the approach of embedding AI assistants into existing systems to enable operation via natural language. Simply typing "Show me last month's sales for the Tokyo branch" allows the AI to extract data from the system and reply.
Specific improvement examples:
- Data search and report generation in chat format
- Input assistance (auto-completion based on historical data, flagging input errors)
- Operation guides (AI guiding users on what can be done on the current screen)
Application conditions: Access privileges to databases must be organized. Rules for handling confidential data must be well-defined.
Estimated cost: 2 million to 8 million yen / Duration: 2 to 6 months
Selection flowchart for business system UX improvement approaches
The optimal solution among the five approaches varies across three axes: "backend condition," "budget," and "challenges to solve." Please refer to the table below to select the approach best suited to your company.
| Decision Criteria | Approach 1: UI Revamp | Approach 2: Dashboard | Approach 3: RPA | Approach 4: API Enablement | Approach 5: AI Integration |
|---|---|---|---|---|---|
| Backend stability | Must be stable | Does not matter | Does not matter | Requires a structure that can be API-enabled | Must allow DB connections |
| Budget | Medium to large | Small to medium | Small | Medium to large | Medium to large |
| Duration | 2 to 6 months | 1 to 3 months | 2 weeks to 2 months | 3 to 6 months | 2 to 6 months |
| Best-fit challenge | Poor usability, outdated screens | Scattered information | High volume of routine work | Need to use outside the office | Search and aggregation take time |
| Internal IT structure | Development partner required | In-house development possible with BI tools | In-house implementation possible with RPA tools | Development partner required | Development partner required |
Decision criteria when in doubt:
- "Want to feel results first" → Start small with Approach 3 (RPA)
- "Screens are fundamentally hard to use" → Approach 1 (UI revamp) is most effective
- "Just want to see information quickly" → Approach 2 (dashboard) is fastest
- "Want to use outside the office as well" → Approach 4 (API enablement + mobile)
- "Learning how to use it is overwhelming to begin with" → Simplify operations with Approach 5 (AI integration)
How to run a legacy system UX improvement project — 3 steps to avoid failure
Even if the approach is decided, results will not materialize if the project is handled incorrectly. Here are three steps to avoid failure in legacy system UX improvement.
Step 1: Identify pain points through frontline interviews
The first step in improvement is to listen to the voices of people actually using the system. Very often, assumptions made by executives or IT departments diverge from the frustrations of frontline workers.
Five questions to ask during interviews:
- What operation takes the most time in your daily work?
- Is there anything that makes you think, "If only we didn't have to do this task"?
- What is the first thing that new hires struggle with?
- What causes hassle when exchanging data with other departments?
- Are there any improvements where you think, "I wish it worked like this"?
The key point is that simply interviewing 5 to 10 frontline staff members for 30 minutes each brings 80% of the points to improve into view. Large-scale surveys are unnecessary.
Step 2: Produce quick wins with minimal improvements
Among the challenges identified through interviews, start with the ones of highest frequency and lowest improvement cost. Rather than aiming for global optimization, creating one success story first is critical.
For example, if the challenge is that "daily morning data transcription takes 30 minutes," simply automating it with RPA can save over 10 hours of labor per month. This success builds internal momentum that "it is okay to make improvements," providing the driving force for subsequent steps.
As introduced in Five Success Stories of SME DX, the common denominator among successful companies is the approach of "starting small, showing results, and then expanding."
Step 3: Incrementally expand the scope of improvement
Once results are achieved with the initial quick wins, expand the scope of improvement step by step. Attempting to change everything at once risks confusing the frontline and actually degrading productivity.
Example of a recommended roadmap:
- Months 1–2: Automate routine work through RPA adoption (quick win)
- Months 3–4: Visualize information by building dashboards
- Months 5–6: Revamp the UI of key screens (starting with the most frequently used ones)
- Month 7 onwards: Consider mobile compatibility and AI integration
By taking steps in this manner, you can steadily accumulate results while spreading out investments. As explained in our article on How to Choose a Web Production Company, selecting a partner capable of handling such incremental improvements is also an important element of success.
Conclusion — Start by changing "what people touch"
The longer legacy system problems are left unaddressed, the larger the losses grow. However, there is no need to "rebuild everything."
Key takeaways from this article:
- Hard-to-use systems generate "hidden costs" on the order of millions of yen annually
- Before doing a full replacement, consider improving only the UX layer
- Choose from five approaches (UI revamp, dashboards, RPA, API enablement, AI integration) to fit your company
- Proceed in three steps: frontline interviews → quick wins → phased expansion
What is vital is not aiming for perfection, but changing things starting from "what users touch every day." Even small improvements will reliably transform frontline productivity and satisfaction.
Much like the philosophy behind entry form optimization (EFO), simply redesigning with the "experience of the person entering data" as the starting point dramatically improves operational efficiency.
If you are facing challenges with legacy system UX improvement or business system usability, start by listening to frontline voices. If you are unsure where to begin, having UI/UX experts assess your current situation is also an effective first step.
