Cralgo
In-house team, agency, or Centre of Excellence?
A clear comparison of building in-house, working with an agency, and establishing a Centre of Excellence with applied intelligence — including when each is the wrong choice.
Knowledge Note
Choose by what you need to keep, not by what you need built: agencies rent capacity, in-house teams build capability slowly, a Centre of Excellence keeps direction and standards while capacity flexes.
One current Cralgo reading of this question — based on observed situations, repeated practice, and open to revision as context or evidence changes.
The current reading
- Capacity can be rented. Capability has to be kept somewhere.
- An agency reads well when scope is bounded and the outcome does not need to survive the contract.
- In-house reads well when the work is permanent and an operating model already exists to receive the hires.
- A Centre of Excellence is a structure for holding direction, standards and mastery so that delivery — internal or external — has something to be accountable to.
How the question is usually framed
As a cost comparison. Day rates against salaries, with a spreadsheet that makes one option look obviously cheaper.
That spreadsheet is usually incomplete, because it prices delivery and ignores what the organisation retains afterwards.
The real variable
Ask what must still be true in three years. If the answer is 'this specific project is live', an agency is efficient. If it is 'we can keep making good decisions in this domain without asking anyone', you are buying capability, and capability has to live inside a structure that outlasts individuals.
That is the whole argument. Everything else is implementation detail.
Three models, compared honestly
Agency Fully in-house Centre of Excellence + pods Buys you Capacity and speed Permanent capability Retained direction and standards, flexible capacity Time to value Weeks Two to four quarters Weeks for delivery, quarters for capability Fails when The outcome becomes business-critical Hiring outruns the operating model The internal owner is nominal and cannot say no Leaves behind Deliverables A team, and whatever it learned Standards, playbooks and people who can apply them More useful for Bounded scope, early channels, surge work Permanent, differentiating work at real volume Organisations that need to keep the capability but cannot hire the whole chainThese are not loyalties. Most organisations use all three at different times, and the mistake is using one past its stage.
What a Centre of Excellence actually is
Not a department, and not a review board. It is where direction, standards and mastery are held for a domain — so that anyone delivering in that domain, employee or partner, is accountable to something written rather than to someone's memory.
Delivery teams — internal or external — then work against those standards. When the team changes, the standard does not. That is the mechanism by which capability becomes institutional rather than personal.
The common failure of each
With agencies, the failure is accountability drift: you hold them to opinions instead of standards, so every disagreement becomes a negotiation. Write the standard first and the relationship improves immediately.
With in-house, the failure is hiring into a vacuum. Six good engineers without decision rights or standards will produce six defensible and incompatible approaches.
With a Centre of Excellence, the failure is nominal ownership. If the internal owner cannot decline work or reject output, you have built a label rather than a structure.
Choosing between them
Separate your current work into what must be kept and what merely must be delivered. The first list determines the model; the second determines the resourcing. Most organisations have never drawn that line.
If one model is still enough
If you are validating rather than scaling, rent everything and keep the learning in your own head. Structure before product-market fit is a cost with no return.
What this reading rests on
observed situations, repeated practice
Why it stays contextual
Technology × Organisations — the answer moves with the people and the environment, not only the technology.
What this note is not
A universal answer, a best-practice list, or evidence. It is a reusable reading that should be tested against your own situation.
What remains open
- How much retained capability survives when the people who held the standards leave.
What would change this reading
- Bounded external delivery that leaves durable capability behind with no internal owner of standards.
Related reading at Cralgo
Centres of Excellence The operating model PlaybooksThese informed the reading. They do not prove it.
Related Research
Questions Cralgo is currently exploring Open research questions, not conclusions.Continue the reasoning
Who should own your D2C technology? What a product and technology operating model actually is Do we need a CTO yet?This is a current reading, not a final answer.
Back to KnowledgeExplore related Research Work with CralgoWe identify emerging patterns and opportunities through research and Signals, and take them into real-world application through technology strategy and leadership, AI and data, digital commerce and consumer platforms, product and engineering, and technology-led growth and transformation.
From emerging patterns to real-world application.
Emerging Patterns → Research & Signals → Collaborate → Technology & Application → Better Outcomes
Where we apply this thinking
Technology Strategy & Leadership · AI & Data · Digital Commerce · Consumer Platforms · Product & Engineering · Technology-led Growth & Transformation
Psychology × Technology × Organisations → Outcomes — the research lens Cralgo explores through.
Explore Signals · Collaborate with Cralgo
Signals · What is a Signal? · Open Signals · Threads · Papers · Reflections · Field Notes · Questions · What we are learning · Research · Technology · Open Explorations · Collaborations · Evidence · About Cralgo · Beyond · Connect
Founding
Cralgo was founded in 2025 by Anil Kabir Kumar. Legal entity: Cralgo Innovations (OPC) Private Limited, incorporated 27 June 2025.
The three lenses
Psychology — how people perceive, decide and behave. Technology — what the system makes possible or constrains. Organisations — what conditions shape action. Outcomes emerge from their interaction.
Research — the questions underneath the work
Collaborative investigation with organisations, universities, practitioners and the people a question affects. Commissioned research is disclosed as such.
Beyond — questions Cralgo carries on its own account
Beyond describes the origin of a question: one Cralgo carries independently, or around a public-purpose context.
- Beyond
- Explore
- Beyond Ecosystem
- AI and public purpose — a provisional context inside Beyond
- Cralgo Now — a public-interest initiative about timing and attention
Working together — the belief behind all of it
Listening before concluding, room for different perspectives, useful disagreement, and evidence that still matters.