The first sign is usually a shadow app
Somebody in finance builds an approval tracker in a spreadsheet. A regional office pays for its own intake tool. A team starts pasting records into a chat assistant because the platform they were given can’t answer a question that spans two systems. None of this shows up in the platform’s own reports. It shows up later, as data nobody can reconcile and work nobody can see.
These are shadow apps, and more and more often they are shadow agents. People build or buy them when the system of record is too hard to change or too hard to use. Each one is a decision someone made to get their work done outside the shared process.
When enough of them appear around one platform, you have a platform trap: a system bought to consolidate work that now holds it back. It usually starts with good reasons (fewer vendors and lower cost). Over a few years the customizations pile up, changes slow down and the cost per useful change rises.
How to score this assessment
Answer each question for one platform: the ITSM, ERP, CRM or case system that most of your requests run through. Score each from 1 to 5.
- 1
- Never
- 2
- Rarely
- 3
- Sometimes
- 4
- Often
- 5
- Always
There are 14 questions in four groups. Add up each group, then add the four group totals. The lowest possible score is 14 and the highest is 70. You can print this page and write your scores in the boxes.
Ask the people who run the platform to answer the operational questions. Ask department heads for the organizational ones, and finance for the cost questions. Scores that differ a lot between groups tell you something too.
Operational: how hard is it to change?
These four questions measure how much of the platform team’s week goes to keeping things as they are.
| # | Question | Score |
|---|---|---|
| 1 | Does your platform team spend more time keeping the platform running than improving what people use? | Score from 1 to 5 |
| 2 | Is the backlog full of “quick form changes” that still need an administrator or a developer? | Score from 1 to 5 |
| 3 | Do you put off changes because nobody is sure what they will break upstream or downstream? | Score from 1 to 5 |
| 4 | Does every upgrade feel like a small re-implementation? | Score from 1 to 5 |
| Section total: ____ out of 20 | ||
Organizational: are people working around it?
A high score here means the shadow apps already exist. Question 5 is the one to answer most honestly.
| # | Question | Score |
|---|---|---|
| 5 | Have departments started building their own workarounds outside the platform? | Score from 1 to 5 |
| 6 | Are core business needs waiting on the vendor’s roadmap? | Score from 1 to 5 |
| 7 | Do stakeholders roll their eyes when they are told to use the platform as it comes, out of the box? | Score from 1 to 5 |
| 8 | Is the platform team overloaded with requests that people should be able to handle themselves? | Score from 1 to 5 |
| Section total: ____ out of 20 | ||
Financial: what does it cost to stay?
Pull the license report and last year’s consulting invoices before you answer.
| # | Question | Score |
|---|---|---|
| 9 | Are you paying license fees for users who rarely log in? | Score from 1 to 5 |
| 10 | Are you hiring more consultants just to manage the platform’s complexity? | Score from 1 to 5 |
| 11 | Have you stopped calculating the platform’s real return because the costs are too tangled to separate? | Score from 1 to 5 |
| 12 | Do you keep hearing “it’s too late to change now”? | Score from 1 to 5 |
| Section total: ____ out of 20 | ||
AI: who decides what your agents can see?
These two questions are new since the original version of this assessment. They ask whether AI is following the same path as the shadow apps.
| # | Question | Score |
|---|---|---|
| 13 | Are teams adopting their own AI assistants because the AI built into the platform can’t see data held in your other systems? | Score from 1 to 5 |
| 14 | Is your plan for AI agents waiting on one vendor’s AI roadmap? | Score from 1 to 5 |
| Section total: ____ out of 10 | ||
What your score means
Add the four section totals. Each band below ends in the same build vs buy decision, made at a different pace: keep the system of record, and build the experience and workflow that cross your systems on top of it.
| Total score | What it means | Build vs buy recommendation |
|---|---|---|
| 14–28 Low risk | Most answers are “never” or “rarely.” The platform still changes when the business needs it to, and few people work around it. | Keep the platform and keep it close to standard. The next time a request spans it and another system, build that process outside the platform’s core so the customizations don’t start to pile up. |
| 29–49 Warning signs | The averages sit around “sometimes.” Workarounds exist in a few departments, and the backlog is growing faster than the team can clear it. | Keep the platform as the system of record. Find your highest-scoring section, pick one workaround from it and rebuild that as a governed app on a build platform that reads from and writes to the system. Measure it, then do the next one. |
| 50–70 In the trap | Most answers are “often” or “always.” The platform is costing you speed and money, and departments have mostly stopped waiting for it. | Replacing the platform outright usually means a multi-year migration that ends with a new platform to customize. Keep the records where they are. Move the approvals and cross-system workflows that change most often onto a build platform, starting with the shadow apps that cost the most. |
The section totals matter as much as the overall score. A high operational score with low organizational and financial scores points to an upgrade and backlog problem. A high organizational or AI score means work has already left the shared process, which is harder to bring back the longer it runs.
Four organizations that kept their systems and built on top
In each case the systems of record were doing their jobs. What slowed the work down was the manual effort between them.
- Missile Defense Agency. MDA moved routing and approvals for more than 9,000 users onto Kinetic with no core system replaced. Routing cycles fell from five days to two. (MDA case study)
- USDA. The National Information Technology Center kept vSphere, Active Directory and its cloud platforms and built a self-service catalog on top. Its proof of concept was built in four days, and server provisioning fell from three weeks to 30 minutes. (USDA case study)
- U.S. Army INSCOM. INSCOM replaced an 85-page paper recruiting packet with a portal deployed in a DoD IL5 environment with CAC sign-in. Recruiter admin time fell from about 25 hours a week to about 12.5. (INSCOM case study)
- GreenState Credit Union. GreenState kept the systems its HR, IT and security teams already ran and built one onboarding workflow across them, cutting manual onboarding work by 85%. (GreenState case study)
What you can build next
If your score landed in the middle or top band, these two pages are the next step.
- How Kinetic’s experience, workflow and integration layers fit together, including how AI builds inside Kinetic through MCP.
- How Kinetic compares with custom builds and low-code tools, if you are weighing your options.
Frequently asked questions
What is a platform trap?
It is a platform bought to consolidate work that now slows that work down. Changes take months, upgrades feel like re-implementations, and departments build their own tools around it. The platform still holds important records, which is why replacing it is rarely the answer.
We scored over 50. Should we replace the platform?
Usually not. Replacing a system of record means migrating years of data and retraining everyone, and the replacement starts collecting its own customizations on day one. Keep the records where they are and move the processes that keep changing onto a platform built for them.
How does AI change the picture?
Teams now adopt AI assistants the same way they once adopted spreadsheets, and an assistant that can only see one system gives partial answers. In Kinetic, AI builds forms, portals and workflows through MCP, and can create records and start workflows within the signed-in user’s permissions. The workflows themselves run the same way every time.
How long does the first app take to build?
It depends on the process and the systems involved. USDA’s National Information Technology Center built a working proof of concept of its self-service catalog in four days, on top of its existing infrastructure systems.
What happens in a working session?
It is a free, two-hour session. We map one of your processes with you, from the first request to the last system update, and send you a written plan for building it.
Owner: Kinetic Data. Last updated .
Book a working session to map the process behind your highest-scoring question.