Predictability is what we sell.Code is the medium.
Most agencies sell speed, talent, or hype. We sell a
delivery system that does not surprise you. The proof is
a written set of commitments the client can point to in
procurement, and two live platforms running on it:
Ingersoll Rand's IRIS in China, and the Council of Europe
Development Bank's events platform.
Ingersoll RandCouncil of Europe Development BankFramerRamp NetworkPrimarySpaceliftRye201.vc
02·What this system protects
Four things the SDLC keeps off your desk.
Each one is a written commitment that procurement can
point to.
01
Scope clarity
Scope is approved in writing by the right decision-maker, with a date, before delivery planning starts. Out-of-scope items are named explicitly. The thing the client thought was included but is not gets called out before the contract is signed.
02
Budget control
Discovery puts an honest investment figure in front of the client before anyone quotes a number. A scope change after sign-off is a planning event, not a ticket edit, and its impact on date, budget, and dependencies is visible.
03
Delivery predictability
Every active task belongs to a version, milestone, sprint, or other visible plan. Implementation gets continuously checked against signed-off scope, so drift is caught early. Release readiness is based on evidence.
04
Decision traceability
Important decisions are written down and accessible to both sides. A "no" is also a written decision. Rejected ideas do not silently come back six months later as new chaos.
03·Content architecture
The content model is shaped during Discovery and signed
off in Definition.
Page types, reusable blocks, tenant configuration, and the
relationships between them are nailed down before the
first commit. On IRCO, that meant nine page types and
tens of reusable blocks for 20+ brand tenants in China,
modeled before the build, approved in writing, and
implemented against acceptance criteria.
The model is the part that is hardest to change later.
Discovery scales to the work. A small request might be
one well-written ticket. A big one gets a workshop, a
technical feasibility check, a UX exploration, and a
full PRD.
Page types, reusable blocks, and field-level localization
Tenant-aware collections for multi-brand rollouts
Navigation maps projected separately from the content tree
Acceptance criteria for the model live in one place both sides can pull up later
SCHEMA · BLOCKS
Hero Feature row CTA strip Article card Form embed Media gallery Pricing table Testimonial Logo strip
9 blocks · composablelocalized
04·Authoring
RBAC · PERMISSION MATRIX
Editor
Lead
Admin
Page draft
●
●
●
Publish
—
●
●
Image alt text
●
●
●
Media library
—
●
●
Tenant config
—
—
●
Redirects
—
●
●
field-level · per tenantaudit logged
Editors run the system after we hand it over.
On IRCO, permissions can be configured down to a single
button or field. A user can be allowed to edit only image
alt texts, without receiving broader access to the media
item or the rest of the page. Most role-based access
systems stop at collection-level permissions.
On the CEB events platform, we built a custom Participants
Manager inside the admin panel: bulk import from Excel,
filtering by invitation and registration status, mass
messaging to a filtered group, a submission approval
workflow for public events, and a change-history audit
trail on every participant.
Custom dashboards and page views, scoped per role
Role-based access tied to tenants, down to a single field
Publication workflows with manual and scheduled revalidation
Change history on every editable record
05·Provider contract
The frontend reads one schema. The provider sits behind
an adapter.
Ingersoll Rand's frontend is a complex Next.js
application, and IRIS 2.0 had been tightly coupled to
Contentstack. Swapping one CMS for another would have
solved the short-term migration while leaving the same
architectural problem in a new form. The decision was to
introduce a Content Provider Service inside IRIS,
validated with Zod. Adapters were built for both
Contentstack and Payload against that contract.
IRIS no longer needs to know whether a page came from
Contentstack or Payload. A new adapter implements the
same interface. That future-proofing was part of the
architecture we sold and delivered.
PROVIDER CONTRACT
PayloadContentstackSanityHeadless WP
ContentProvider · Zod-validated
IRIS · Next.js application
one schema · many providersv 2.0
06·The SDLC
The SDLC has seven phases.
They live in writing, visible to both sides, before
delivery starts.
00
Engagement Setup
Before delivery starts, we walk the client through how cooperation will work, what scope sign-off means, how new requests enter the plan, and what happens when scope changes. The cooperation model is written down and accessible to both sides later.
01
Discovery
A dedicated discovery team sits on every inquiry from day one. We refuse to start building from a vague request. For every request we agree on the problem being solved, why it matters, who it affects, the expected outcome, the known constraints, the known risks, and what is still unclear.
02
Definition and Sign-off
Scope is approved in writing by the right decision-maker, with a date, before delivery planning starts. Out-of-scope items are named explicitly. Acceptance criteria are written before the first commit. A scope change after sign-off is a planning event, not an edit to a ticket.
03
Scoping, Planning and Design
Approved scope becomes a visible plan with tasks, owners, priorities, dependencies, risks, and a target delivery window. Technical review is part of planning, not something that surprises us during build. Active scope and future ideas live in different places.
04
Development and QA
Code is reviewed before it gets merged. QA is part of delivery, not a final check after coding is finished. Implementation gets continuously checked against signed-off scope. Release readiness is based on evidence, not optimism.
05
Release
A release is a communication event, not only a deployment. The release outcome is verified after the deploy. Release notes spell out what was released, what changed for the user, what limitations remain, and what follow-up exists. Releases never introduce new scope.
06
Maintenance and Next Scope
Live projects do not decay into an unmanaged stream of urgent Slack messages. Defects, support requests, and new ideas are separated and routed differently. Support work does not silently consume the development capacity committed to planned scope.
07·Numbers from production
Numbers from the two engagements running on this system.
0
Downtime on the IRCO cutover
20+
Brand tenants on IRIS in China
5
Months from kickoff to IRCO launch
9
Page types in the IRIS content model
100%
Of invited CEB participants responded
~0
Support tickets on the CEB invitation flow
IRCO shipped the Payload CMS platform before the Oracle
Content Manager discontinuation deadline, on AWS China, with
24/7 monitoring. CEB sent hundreds of invitations for their
70-year anniversary with almost no support tickets on the
invitation flow, and now use the platform for every new event
they organise.
08·From the operators
"
[Pending — a 30–50 word quote from a content platform
client about the architecture decision, the editorial
rollout, or the post-launch stewardship. The slot below
holds attribution structure and is ready for sourcing
before publication.]
"
Name TBC
Role TBC, content platform team
Client name
Yes. Scope is approved in writing before development starts. Signed-off scope is fixed for the current cycle. Changes after sign-off require written approval and may affect timeline, cost, task breakdown, priorities, QA scope, and release plan.
02What happens if a new request shows up mid-flight?
It goes back through Discovery, unless it is an urgent production issue. The client helps prioritise new requests instead of treating everything as urgent. Developers do not absorb unclear scope by working overtime or guessing what the client meant.
03Which CMS do you build on?
We hold a top-tier formal partnership with Payload and ship the majority of enterprise platforms on it. Both IRCO and CEB run on Payload. Framer is the right answer for marketing surfaces where the client team needs to author visually; we hold the Framer Pro Enterprise Expert designation there.
04What happens after launch?
Defects, support requests, and new ideas are separated and routed differently. Support work does not silently consume the development capacity committed to planned scope. The next cycle starts with Discovery and Sign-off, not with momentum.
05How involved does my team need to be?
The client participates in Discovery and scope definition, helps prioritise new requests, and approves scope in writing before development starts. Delayed client decisions, feedback, access, or approvals may delay delivery. Both sides see and sign off on the cooperation model before anyone writes code.
NEXT STEP
We're booking enterprise content platform builds for 2026.
Thirty minutes on a call gives you a shape of the work and
an honest investment figure. We will also tell you, on that
same call, whether we are the right team for the work.