Information architecture · six directions · one product

Six ZopNights, one dataset

The same cloud estate — the same $412,800 of spend, the same 147 opportunities, the same 4,218 resources — organised six different ways. Each one borrows the structure and the design language of a platform that solved navigation at this scale. Open them side by side and the choice makes itself.

spend $412,800/mo resources 4,218 opportunities 147 clouds AWS · Azure · GCP rails 9 – 19 rows
Vercel01 · 9 rows

Scope-first

One rail. A scope control above it decides what every page means.

Buys you
  • The shortest rail the product could have
  • Four duplicate surfaces stop existing
  • Your filter survives the page change
Costs you
  • A quarter of engineering — it changes the API contract
  • Nine words can undersell a thirty-capability platform
Open the product →
Stripe02 · 11 rows

Fewest links

Four daily nouns on top. Everything else is a product you find once and pin.

Buys you
  • The most scannable rail of the six
  • Products doubles as merchandising
  • Shortcuts make depth free after the first visit
Costs you
  • “Products” is a fiction unless they are separately enableable
  • A four-item top undersells the platform on first impression
Open the product →
Linear03 · 12 rows

Yours first

What is assigned to you sits above what belongs to the company.

Buys you
  • The only direction that answers “why open this on a Tuesday?”
  • The team block is the scope model expressed as navigation
  • A command palette carries the real traffic
Costs you
  • ZopNight has no personal object model to hang it on
  • The largest net-new product surface of the six
Open the product →
Datadog04 · 19 rows

Three zones

Search and memory on top, capabilities in the middle, raw data underneath.

Buys you
  • Handles breadth without hiding it
  • Puts the substrate low, where it belongs
  • Recents and pinned mean depth costs nothing twice
Costs you
  • The longest rail here — it needs flyouts to stay usable
  • Per-user state the product does not keep today
Open the product →
Vantage05 · 15 rows

The peer’s map

Group by FinOps verb. Every cloud and platform is a filter, never a shelf.

Buys you
  • Proven in market at exactly this job
  • Kubernetes becomes an integration, not a shelf
  • Cheapest of the six to reach — nav configuration only
Costs you
  • You inherit a competitor’s positioning
  • ZopNight *acts*; here that becomes one item inside Optimization
Open the product →
FinOps Framework06 · 16 rows

The practitioner taxonomy

Domains, capabilities and maturity — named and ordered by the industry standard.

Buys you
  • Every word is one a FinOps buyer already owns
  • The phase spine makes the loop visible on every screen
  • Maturity per capability gives the practice somewhere to go
Costs you
  • “Quantify” is a hard word to put in a sidebar
  • Half the users are engineers who do not speak this
Open the product →

How to read these. Every variant contains the same capabilities and the same numbers — nothing is added or removed between them. What changes is which shelf a thing sits on, what it is called, and how the product looks while you use it. If a variant feels better, that is the information architecture doing its job, not the data.

The spread in rail length is itself the finding. Nine rows and nineteen rows are both correct answers to different questions: nine works when a scope control absorbs the variation, nineteen works when search and pinning absorb the length. What fails is the middle — a long rail with neither. That is where ZopNight sits today.