← All perspectives

Perspective / Software & CRM

Bespoke CRM economics: which capabilities should you build, extend or buy?

The strongest CRM decision is often made capability by capability. Compare packaged products, extensions and bespoke development through process fit, architectural boundaries, lifecycle cost and the commercial value of control.

10 min readReviewed

Hands comparing custom, modular and finished components on an engineering bench.
Conceptual editorial image.

The decision to build, extend or buy a CRM should follow the economics and distinctiveness of the capabilities the business needs. Treating the entire platform as one indivisible choice can hide better combinations: established software for common functions, an extension for a specific workflow and a separate service for genuinely differentiating logic. This article develops a capability-level decision model and an illustrative three-year cost comparison. It explains what executives should demand before approving either a packaged platform or a bespoke programme.

The executive view

  • Choose build, extend or buy at capability level using evidence about actual workflow fit.
  • Compare initial, recurring, transition and future-change costs on a consistent basis.
  • Treat control, operating capacity and exit requirements as part of the commercial decision.
01

Start with capability fit rather than a preferred product

Describe what the organisation must do: qualify enquiries, govern account ownership, calculate commercial terms, manage commitments, support customers and report performance. Separate essential behaviour from habits inherited from the current tool. Reproducing every existing screen can preserve unnecessary complexity.

Classify capabilities according to their commercial role. Some are common across many businesses; others support distinctive processes; a smaller set may create meaningful competitive advantage. Microsoft’s domain analysis guidance provides a useful vocabulary for distinguishing core, supporting and generic domains.

This classification does not produce an automatic build decision. A differentiating process may still be well supported by a configurable product. Equally, a seemingly routine capability may require custom work because of a difficult integration or operating constraint. Use the classification to focus investigation, then test the actual options against the required behaviour.

Translate the capability map into acceptance scenarios before supplier demonstrations. A sales opportunity with one contact and one product tells little about an account shared across subsidiaries, a quote with an expiring approval or a contract change after an order has entered fulfilment. Include exceptions that consume real operational effort today. Agree which behaviour the business is willing to standardise and which behaviour must remain distinct.

Distinguish capability value from familiarity. A team may describe a custom spreadsheet as indispensable because it resolves a genuine policy gap, or because no one has challenged its manual steps. Observe the work and identify its decision rule before choosing a platform feature. This avoids building a permanent system around accidental behaviour. It also improves a purchase evaluation: suppliers can respond to an explicit problem, rather than accepting a long list of screens that may not represent the best future process.

02

Give buy, extend and build comparable definitions

A packaged purchase still requires configuration, migration, integration, administration and a plan for changing the service. An extension adds custom behaviour to an existing platform or sits alongside it. A bespoke solution gives greater control over its design, while creating responsibility for engineering and operation.

Define the boundary of each option consistently. A purchase estimate excluding migration should not be compared with a bespoke estimate that includes it. Likewise, the cost of retaining a manual workaround must not disappear from the option that needs it.

Create a fit-gap model with three categories: satisfied directly, satisfied with an agreed adaptation, and unresolved. Document how each adaptation affects upgrades, support and data authority. A demonstration should test the business’s difficult exceptions rather than only the supplier’s preferred workflow. This exposes whether an apparently small configuration is really a second application hidden inside the platform.

Fit-gap evidence should include the quality of implementation, not a binary feature check. A platform may technically support an approval matrix but make it difficult to explain why a decision occurred. It may offer an export while omitting the relationships needed to reconstruct historical account ownership. Score the consequential scenarios with supporting evidence and the required operating workaround. Where a workaround is accepted, give it an owner and an estimated recurring cost.

Keep environmental parity visible. A demonstration tenant with unrestricted administration does not establish what an ordinary salesperson can do under production permissions. Test the intended roles, realistic data volumes and an integration failure. Record any paid modules or higher service tiers needed to make the scenario work. These details often alter the commercial comparison more than the advertised per-seat price. A comparable option is one that can deliver the agreed service under the intended operating conditions, rather than one that can reproduce a happy-path screenshot.

03

Protect the boundaries of distinctive business logic

Decide which rules belong in the CRM and which should remain under a separate business boundary. Pricing, eligibility, contract generation or fulfilment orchestration may be easier to govern when their authoritative logic is explicit rather than distributed across several workflows and plugins.

An extension can be commercially attractive if it preserves useful platform capabilities while isolating distinctive logic. It can also create a tightly coupled estate that is difficult to upgrade. Review API limits, transaction boundaries, identity, testing environments and the ability to observe failures.

For each critical rule, record the authoritative implementation and the route for changing it. If a platform change requires coordinated edits to the CRM, website and finance integration, the apparent convenience of configuration may carry significant future change cost. The architectural review should explain this exposure in business terms, including who must coordinate and how an unsuccessful change is recovered.

04

Worked example: three different cost shapes

Consider an illustrative business comparing three CRM options over three years. These hypothetical figures demonstrate the model; they are not supplier quotations, market benchmarks or KTE project prices.

OptionInitial delivery and transitionAnnual operation and licencesUndiscounted three-year total
Configure a packaged product£90,000£110,000£420,000
Extend the current platform£180,000£75,000£405,000
Build a bespoke capability£300,000£45,000£435,000

The totals combine initial cost with three years of recurring cost. They do not establish that one approach is better. Functional fit, adoption, transition risk and future change could reverse the conclusion.

The example shows why a low entry price or a low licence bill is insufficient. Each option needs a defined scope, a credible operating model and consistent treatment of the costs retained elsewhere in the business.

Discounting can change how the three options compare. Using a purely illustrative 8% annual discount rate, initial cost at time zero and recurring costs at the end of each of three years, the recurring-cost present-value factor is approximately 2.577. That gives present costs of about £373,500 for the packaged option, £373,300 for the extension and £416,000 for the bespoke capability. These are calculations from invented inputs, without tax, residual value or inflation; they are not investment advice or a recommended discount rate.

The first two options are effectively tied under those assumptions. A modest change in user growth, integration effort or required functionality could reverse their order. That tells the sponsor where more evidence is useful: resolving an uncertain £30,000 migration requirement matters more than refining a small present-value rounding difference. Add benefit timing separately. An option delivered sooner might create earlier value, while a rushed transition might increase disruption. Show those effects explicitly so that a sophisticated cost model does not become a substitute for a realistic implementation plan.

05

Model uncertainty and the cost of future change

Test the assumptions most capable of changing the recommendation. These may include user growth, transaction volume, platform pricing, the number of custom workflows and the availability of people who can maintain the solution. Separate costs that rise with usage from costs that change in steps.

Include a realistic change scenario. For example, assess the effect of introducing a new account hierarchy or an additional sales channel. Estimate which components need modification, the testing effort and the coordination required. Do not assume future change is free simply because the current option is labelled configurable.

The FinOps unit economics framework is a useful reference for connecting technology spending to value. Apply that thinking to meaningful business units and decision scenarios. A lifecycle model should show the range of plausible cost and benefit, not conceal uncertainty inside a single precise-looking total.

Use a small scenario set that tests different cost mechanisms. A doubling of licensed users, an increase in automated transactions and a new complex workflow are separate changes. One may trigger a platform tier; another may need extra queue capacity; the third may require development and regression work. Model the driver and its threshold rather than increasing every cost by a single contingency percentage.

Benefits also need accountable assumptions. Reducing quotation preparation from twenty minutes to twelve minutes across a hypothetical 1,500 monthly quotes releases 200 hours of task capacity. It does not automatically remove 200 paid hours or generate 200 hours of additional selling. Confirm adoption, scheduling and the business's ability to use the capacity. Compare error correction and approval quality as well as speed. A cheap system that produces faster but less reliable commercial commitments can damage the intended outcome. The investment case should state how benefits will be observed, who owns the process change and when the assumptions will be reviewed.

06

Assess control, exit and operating capacity

Control has value only when the organisation can use it. Owning source code does not provide deployment knowledge, suitable licence rights, access to environments or the ability to support the application. A bespoke option must explain how these obligations will be fulfilled over its expected life.

Packaged options need a different but equally serious exit review. Examine data export, identity dependencies, contractual rights, configuration portability and the effort of replacing integrations. Ask what the business would need to retain if the implementation partner changed while the platform remained.

GOV.UK’s purchasing strategy guidance provides an established reference for considering lifecycle cost and combinations of building and buying. For commercial buyers, use those ideas as decision inputs rather than government procurement obligations. The final option should fit the organisation’s operational capacity, not merely its appetite for initial investment.

An exit rehearsal is stronger evidence than a contractual statement alone. Export a representative account, its contacts, activities, attachments, ownership history and relationship identifiers into an agreed neutral form. Check which meaning is preserved and which information depends on platform-specific workflow state. Estimate the effort needed to reconstruct a usable record elsewhere, including access restrictions and history. Use fabricated or appropriately controlled test data for the rehearsal.

For bespoke software, perform the equivalent operational handover. A team other than the original developer should be able to provision an environment, restore a backup, deploy a known release and diagnose an integration error using the supplied materials. Source ownership, credentials, infrastructure definitions and documentation answer different parts of that requirement. Price the support arrangement that the business can actually sustain, including dependency updates and incident cover. This makes control measurable. It also exposes where a low licence cost has transferred responsibility to an organisation that has not yet funded the people or service needed to exercise it.

07

Approve a capability decision with a review trigger

The executive pack should contain the capability map, fit-gap evidence, architectural boundaries, lifecycle scenarios and transition requirements. Identify the assumptions that would cause the recommendation to change. A decision can be appropriate today without remaining appropriate after substantial growth or a change in business model.

Where uncertainty is material, fund a bounded investigation that tests the deciding assumption. That could be a difficult integration, a representative workflow or a migration rehearsal. Define the question and acceptance evidence before calling the exercise a proof of concept.

The strongest recommendation may combine options rather than select one label for the whole estate. The business should be able to explain why it buys a common function, extends a useful platform or builds a distinctive capability. That explanation is the commercial case for the architecture, and it should remain understandable to future owners and suppliers.

Further questions

Questions this raises

Is bespoke CRM always more expensive than buying a platform?

No universal answer is defensible. Initial development, recurring licences, integration, operating support and future change produce different cost shapes. Compare equivalent scope over an appropriate lifecycle and test the assumptions that could change the conclusion. Treat the current organisation's operating capacity as part of the comparison. A lower recurring bill can transfer patching, incident support and integration maintenance into internal work. A packaged platform can transfer other costs into modules, licence tiers or partner support. The useful question is which complete arrangement delivers the required capability with an affordable and sustainable operating obligation.

When is extending an existing CRM appropriate?

When the platform remains useful and custom behaviour can be added with acceptable coupling, upgrade impact and operating cost. Test important workflows and integration constraints rather than assuming a feature described as configurable will be straightforward. A sidecar service can isolate pricing or eligibility logic while preserving useful platform functions, but only if the ownership and failure paths are clear. Verify the available API behaviour, transaction limits, permissions and deployment process. Assess a realistic upgrade and an interrupted integration, since these conditions expose coupling that a normal demonstration may leave invisible.

What should a CRM proof of concept establish?

The specific assumption that could decide the investment: a complex workflow, integration boundary, migration constraint or operating requirement. It needs representative cases and agreed acceptance evidence, not simply a polished demonstration of basic screens. Retain the trial configuration and the difficult cases used to assess it. The final decision should explain the assumption tested, the observed result, the remaining uncertainty and how the evidence changed the recommendation. An attractive interface can aid adoption, but it cannot establish data completeness, recoverability or long-term platform fit without the corresponding technical and operational checks.

References

Sources & further reading

  1. Use domain analysis to model microservices 25 February 2026 (updated)
  2. Define your purchasing strategy 3 September 2026 (updated)
  3. FinOps Framework: Unit Economics

Connected capabilities

Software & CRMStrategy & digital transformation

From thinking to doing

What could this mean
for your business?

Your next chapter

What could your
business do better?

Tell us where you want to go, what’s getting in the way, or what you need to deliver. We’ll start there.

Start a conversation

Start a conversation

Tell us what
you have in mind.

Interactive preview. Your information stays in this page and is not sent.