--case-castleUX metrics · B2E · 2025 – now
CASTLE.
Internal tools, measured by friction.
CASTLE is how we measure internal tools at MANGO: six UX pillars built for software people have to use, not software they choose. Each challenge picks its pillars in discovery and reads them again after the build, so the squad knows whether the solution moved the needle, and what to do next.How we measure internal tools at MANGO: six UX pillars for software people have to use. Chosen in discovery, read after the build, so squads know if a solution moved the needle.
- Every SCALE squad
- Proven on a real tool
- On stage
- Role
- Creator · designed and implemented it
- Team
- SCALE, Experimentation & Growth · MANGO Online
- Timeline
- 2025 – now
- Status
- Live in every SCALE squad
Fig. 1 · UX metrics · B2E
My role
I created CASTLE: I designed it, implemented it and keep refining it with the people who use it.
Born from MANGO Online's internal tools. Today it's the measure phase of the Agentic Product Framework.
- 01
Research into why consumer metrics fail on internal tools, and what really changes when people can't opt out.
- 02
The six pillars, grouped in three families: efficiency, hidden value and reliability.
- 03
Where it lives in a challenge: pillars chosen in discovery, read again after the build.
- 04
Rollout with the squads, and follow-ups with users to keep refining each pillar.
TeamDesigned and implemented by me, validated and reviewed together with the PO Lead, who also shared the stage with me at ProductTank Barcelona.
Context
MANGO Online runs on internal tools like AURA, used every day across teams and markets.
In SCALE, every challenge needs to know whether it worked. For internal tools, nobody had an answer.
- Mandatory use
People can't leave the tool, so adoption and retention say nothing.
- No standard
Consumer frameworks like Google's HEART assume people choose the product.
- Measurable today
Every pillar had to work with what squads can already run: analytics, sessions and short surveys.
Problem
How do you measure a tool people can't choose not to use? The usual answers break. Adoption is guaranteed, retention is guaranteed, and delight isn't the point.
So the question that mattered, did this change make the work better, had no answer.
- Wrong metrics
Consumer metrics gave confident answers to the wrong questions.
- Invisible friction
Extra clicks, workarounds and cheat sheets that nobody counted.
- Design as taste
Without a measure, UX debates stayed opinions instead of operating costs.
Key decisions
- 01
From choice to mandatory efficiency.
HEART asks whether people choose a product. In internal tools they already use it every day, because they have to. CASTLE moves the question to how well the work gets done: effort, time and errors.
trade-off: A new framework, so it had to be explained before anyone could use it.
- 02
Defined in discovery, read after the build.
The metric isn't picked once the solution exists. Each challenge chooses its pillars in discovery and takes a baseline; after Create we read the same pillars. That's how we know if the solution moved the needle, and whether to persevere, pivot or stop.Each challenge chooses its pillars in discovery and takes a baseline; after Create we read them again. Then: persevere, pivot or stop.
trade-off: Discovery takes a little longer, because the baseline comes first.
- 03
Friction as cost, not taste.
Seconds lost on a task people repeat all day add up across a whole team. CASTLE turns that friction into time and cost: a conversation leadership prioritises, instead of a debate about taste.
trade-off: It needs honest data from each team, so the numbers stay inside the company.
- 04
Priorities depend on the product.
Not every pillar weighs the same. In internal tools, task efficiency and cognitive load come first. In B2B software, learnability and advanced features matter more, because customers can leave.In internal tools, efficiency and cognitive load come first. In B2B, learnability and advanced features.
trade-off: Six pillars, but a challenge only tracks the two or three that matter.
Craft
Six pillars, three families.
Pick a pillar to see what it measures, why it matters in an internal tool and what to look for.
01 / 06 · Efficiency
Cognitive load
The mental effort a task takes.
In an internal tool
Overload multiplies errors and drags efficiency and satisfaction down.
Look for
Printed cheat sheets for basic tasks.
Pairs with T · Task efficiency
01
DISCOVER
Pick the pillars that matter for the challenge, and take the baseline.
C · T · baseline02
CREATE
Build the solution with that metric in mind.
03
MEASURE
Read the same pillars again, against the baseline.
C · T · after04
DECIDE
Persevere, pivot or stop, with evidence.
↺ Next iteration, starting from what we learned.
Outcome
Every SCALE squad
CASTLE is the measure phase of the Agentic Product Framework, used by 100% of SCALE's squads.
Proven on a real tool
An order tool customer service agents use all day, with everything on screen at once. Progressive disclosure, smart defaults and inline help brought cognitive load and errors down.
On stage
Shared at ProductTank Barcelona with the PO Lead: believe in internal tools, spread the word, win executive support and grow ambassadors, so they join the digital product flow.
Reflection
If people can't leave a tool, friction is the thing worth measuring.