Bimodal IT Distorts the Portfolio Decision It Was Meant to Improve
Simplicity is valuable only until it conceals the exposure that determines the decision.
Executive Summary
A portfolio committee is presented with two proposals. The first is a twelve-month renewal of a core transaction platform, costed at £9.4 million, supported by a detailed plan and classified as the controlled mode of delivery. The second is a new mobile service, costed at £1.2 million for its first release, sponsored by the commercial director and classified as the fast mode. The committee debates them under different rules, with different evidence and different expectations of certainty. It appears sensible. It is also the point at which portfolio discipline begins to fracture.
The attraction of bimodal IT is understandable. By 2016, organisations are being asked to modernise ageing systems while responding quickly to mobile customers, software-as-a-service competitors and an accelerating shift toward public cloud services. One delivery rhythm cannot sensibly govern every kind of work. The mistake is to convert that valid observation into two institutional lanes: one presumed safe and slow, the other presumed innovative and fast.
Once those labels enter the portfolio, they cease to describe delivery conditions and begin to determine status, funding, scrutiny and career attention. Work in the fast lane is allowed to learn but can escape scrutiny of dependencies and total cost. Work in the slow lane receives control but can be excused from testing whether its original case still holds. The portfolio then compares unlike evidence and makes distorted choices.
This paper recommends one portfolio with differentiated policies, not two modes. Every initiative should be assessed against four conditions: uncertainty, reversibility, operational criticality and coupling. Those conditions should determine funding increments, assurance intensity, architectural freedom and the evidence required at each decision. Speed becomes an outcome of exposure and learning, not a badge attached to a category.
The practical recommendation is to replace mode-based governance with a common decision architecture:
- One investment thesis and one set of portfolio outcomes for all work.
- Small, staged commitments where uncertainty is high, regardless of whether the work is called digital or core.
- Stronger controls where failure is difficult to reverse, operations are critical or dependencies are dense.
- Regular reallocation based on evidence, with the same right to stop a slow programme as a fast experiment.
The question is not whether all work should move at one speed. It should not. The question is whether pace should be governed by an organisational label or by the actual exposure of the decision. Only the latter produces a coherent portfolio.
The Right Diagnosis, Institutionalised Badly
The diagnosis behind bimodal IT is sound. A team replacing an uncertain customer interaction should not be forced to produce the same evidence as a team changing the ledger at the heart of month-end reporting. The first needs short cycles, direct user feedback and permission to discard weak ideas. The second needs reconciliation, controlled migration, service continuity and proof that failure can be contained.
The serious case for two modes therefore deserves respect. Its advocates are not arguing for fashion over discipline. They are trying to protect exploratory work from processes designed for predictable engineering, while protecting critical systems from enthusiasm masquerading as assurance. In organisations where every project must promise fixed scope, fixed cost and fixed time before discovery has begun, a protected fast lane can be a useful act of institutional self-defence.
The difficulty is that a temporary remedy quickly becomes an operating settlement. The two modes acquire different sponsors, governance forums, suppliers, measures and reputations. The distinction that began as a response to uncertainty becomes a distinction between kinds of people and kinds of technology. “Digital” attracts discretionary money and executive attention. “Core” becomes the place where unavoidable cost is managed. The portfolio no longer asks which risks a proposal contains; it asks which lane the proposal belongs in.
That substitution matters because the label is only loosely related to the risk. A mobile service can carry severe data, conduct and reputational exposure. A change to a core system can be technically reversible and economically modest. A public cloud service may reduce infrastructure risk while increasing contractual concentration and integration dependency. A packaged-system upgrade may look predictable until data conversion exposes years of inconsistent definitions. The category simplifies the conversation precisely where the portfolio needs discrimination.
Bimodal IT confuses a useful difference in delivery conditions with a permanent division in investment governance.
How the Two Lanes Distort Investment
The distortion does not usually appear as an explicit policy. It emerges through four mechanisms.
Different standards of evidence
Fast-lane proposals are commonly funded on the promise of learning. That is appropriate at the outset, but learning can become a rhetorical exemption. Sponsors present user interest, early usage or release velocity without connecting those measures to margin, service cost or strategic outcome. The initiative remains “experimental” long after the organisation has accumulated operating commitments.
Slow-lane proposals suffer the opposite problem. They present detailed schedules, resource estimates and risk registers, which create the appearance of evidence without proving value. A plan containing 4,000 lines can be precise about activity and silent about whether the investment remains worthwhile. Because the work is necessary or regulatory in character, its scope is treated as indivisible and its benefits are rarely revisited.
Asymmetric permission to change course
The fast lane is expected to pivot. The slow lane is expected to deliver. That sounds logical until evidence changes. When an exploratory service fails to find demand, a pivot may preserve spend rather than protect value. When a core renewal discovers that only three of eleven interfaces require replacement, adherence to the original programme may destroy value rather than demonstrate control.
The right to change direction should follow new evidence, not mode. Yet many portfolio forums grant one lane excessive freedom and the other excessive commitment. Both errors are expensive.
Hidden transfer of dependencies
A fast team can release a front end in weeks only because identity, customer records, payments, security monitoring and service support already exist. The apparent speed is partly borrowed from the slow lane. Conversely, a core programme may claim enterprise-wide benefits that depend on changes to customer journeys owned elsewhere. The business case records the dependency, but the funding decision does not price it.
This creates a recurring pattern: the fast initiative reports its direct cost while the core estate absorbs integration, data and support work; the core programme reports its own milestones while dependent channel changes wait in another backlog. Both look healthier than the portfolio is.
Status becomes a substitute for strategy
Once “fast” becomes synonymous with strategic and “slow” with maintenance, selection is no longer neutral. Sponsors reframe proposals to qualify for the prestigious lane. Necessary resilience work is described as digital enablement. Modest channel improvements are elevated into transformation. Meanwhile, maintenance that preserves economic capacity is treated as reluctant overhead.
A portfolio should distinguish growth, efficiency, resilience and obligation. Bimodality instead tempts it to distinguish excitement from gravity.
A Worked Portfolio: The Cost of the Label
Consider a composite portfolio in late 2016. It contains 22 active initiatives and a £48 million annual change budget. Nine initiatives sit in the controlled lane and thirteen in the fast lane. The portfolio pack reports that the fast lane is delivering every six to eight weeks, while the controlled lane has missed three milestones. The apparent conclusion is to transfer a further £4 million toward fast work.
A closer examination changes the decision.
The thirteen fast initiatives have direct budgets totalling £14 million. They also require £3.6 million of integration, test-environment and service-readiness work carried inside four controlled programmes. Only five have explicit measures of customer or economic outcome. Four have passed their original six-month discovery horizon without a renewed investment decision. One mobile service has attracted 38,000 registrations but only 6,400 monthly users; its business case assumes 25,000 monthly users by year end. Its team continues to report release frequency as the leading measure.
The nine controlled initiatives have budgets totalling £34 million, including the transferred £3.6 million. One £8.8 million infrastructure renewal is 11 weeks late, but its delay is largely caused by the need to certify interfaces for three fast services added after approval. Another programme, a £6.2 million data consolidation, still carries all fourteen source systems in scope even though analysis shows that six account for 91 per cent of the operational errors. Its gate review tests schedule recovery, not whether the scope should be reduced.
When costs and dependencies are reassigned, the fast lane consumes £17.6 million, not £14 million. When outcomes are examined, only five of thirteen initiatives can show evidence strong enough to justify their next increment. When controlled work is made divisible, the data programme can release £1.7 million by concentrating first on the six material sources. The sensible portfolio action is not a £4 million transfer from slow to fast. It is to stop two weak experiments, narrow one core programme, explicitly fund the shared integration capacity and redirect £2.3 million toward the proposals with the strongest current evidence.
Nothing in that decision requires all teams to work alike. It requires the portfolio to compare them under a common economic logic.
“Two speeds may describe delivery, but they cannot be allowed to define value.”
The Alternative: One Portfolio, Four Conditions
The replacement for bimodal governance is not uniform governance. It is a common set of decisions whose policies vary with exposure. Four conditions provide a more reliable basis.
| Condition | Question for the portfolio | Consequence for governance |
|---|---|---|
| Uncertainty | How much remains unknown about demand, solution and delivery? | Higher uncertainty requires smaller funding increments and more frequent evidence. |
| Reversibility | How easily can the decision be undone without material loss or disruption? | Low reversibility requires earlier assurance, rehearsal and explicit exit design. |
| Operational criticality | What happens to customers, finance or service continuity if the change fails? | Higher criticality requires stronger controls, proving and contingency. |
| Coupling | How many systems, suppliers, processes and initiatives must move together? | Dense coupling requires dependency ownership and portfolio-level sequencing. |
These conditions are deliberately independent. An initiative may be highly uncertain but easily reversible. Another may use familiar technology yet be almost impossible to reverse once data conversion begins. A third may be operationally modest but tightly coupled to six other releases. The combination, not a label, determines the treatment.
For example, an uncertain but reversible customer proposition should receive a small discovery allocation, rapid user testing and a clear stop date. Its architecture should be sufficient to protect data and future options, not designed for an imagined mature service. A well-understood but irreversible ledger conversion should receive staged migration, parallel proving and independent reconciliation. A routine interface change touching seven programmes should receive portfolio dependency management even if its own cost is small.
This model preserves the insight that work needs different rhythms while removing the institutional hierarchy between them.
A Common Decision Architecture
A portfolio can apply the four conditions through five linked decisions.
Define the investment thesis
Every proposal, regardless of delivery approach, should state the outcome sought, the evidence that would support it, the assumptions that could invalidate it and the next decision date. This is not a demand for false certainty. It is a demand for explicit uncertainty.
A useful thesis answers:
- What change in customer, operational or financial behaviour justifies the investment?
- Which assumptions are evidence and which are judgement?
- What is the smallest commitment that can test the critical assumption?
- What continuing cost or dependency is created if the initiative succeeds?
Set the commitment size
Annual approval should establish an envelope, not guarantee full expenditure. The portfolio should release money in increments matched to uncertainty and reversibility. A £10 million initiative may begin with a £400,000 discovery or design commitment; a £700,000 mandatory change may warrant full funding if its scope and route are genuinely clear.
The discipline is not “fund small things.” It is “make the next commitment no larger than the evidence supports.” This applies equally to experimental services and core renewal.
Match assurance to exposure
Assurance should test the mechanism of failure, not the prestige of the mode. High operational criticality may require service rehearsal, reconciliation, recovery tests and senior acceptance. Dense coupling may require an integrated release plan and named dependency owners. High uncertainty may require direct observation, prototypes and commercial tests.
A checklist applied to every initiative creates bureaucracy. No assurance applied to favoured work creates fragility. Exposure-led assurance does neither.
Make dependencies economically visible
Shared integration, environments, data work and service transition should not disappear into the budgets of core programmes. The portfolio should identify which initiatives consume that capacity and price the consequences of sequencing.
This does not require elaborate cost allocation. A monthly dependency ledger can record the supplying initiative, consuming initiative, required date, accountable owner and estimated capacity. Its purpose is to expose choices. If five fast initiatives all require the same identity service in the same quarter, the constraint is a portfolio decision, not a delivery inconvenience.
Reallocate on a fixed rhythm
Quarterly portfolio review is often frequent enough to change material allocations and infrequent enough to examine evidence properly. More uncertain initiatives may face monthly continuation decisions within their agreed envelope. Critical programmes may retain stable teams while their scope is adjusted at quarterly points.
Reallocation must include four legitimate outcomes:
- Continue because evidence and exposure remain within tolerance.
- Increase because evidence has strengthened and the next commitment is justified.
- Narrow or redirect because learning has changed the most valuable scope.
- Stop because the thesis has failed, the obligation has changed or a better use of capital exists.
The stop decision is essential. Without it, incremental funding merely divides a foregone conclusion into smaller cheques.
What Must Be Standard — and What Must Differ
A coherent portfolio standardises decisions, not methods.
The following should be common to every initiative:
- The strategic outcome and accountable sponsor.
- The investment thesis, assumptions and next decision date.
- Full cost, including material dependencies and continuing service cost.
- The four-condition assessment.
- Evidence required for continuation.
- A visible decision record showing why money was committed, changed or stopped.
Delivery practices may then differ. Teams facing uncertain demand may use short iterations, prototypes and frequent releases. Teams changing critical records may use formal design authority, controlled environments, rehearsals and staged cutover. Some initiatives will combine both: exploratory work at the customer boundary and highly controlled change beneath it.
The strongest objection is that this model is more demanding than two simple modes. It is. Portfolio judgement cannot be reduced to a pair of labels without losing information. The answer is not to create a scoring industry around the four conditions. A short, evidence-based discussion is sufficient if leaders are willing to make choices. Simplicity is valuable only until it conceals the exposure that determines the decision.
A second objection is organisational capacity. Many portfolio offices can administer gates but cannot evaluate uncertain propositions; many digital teams can test propositions but resist full-cost accountability. Moving to one portfolio exposes those capability gaps. That discomfort is not a reason to preserve the split. It is evidence that the split has allowed each side to avoid learning the other’s discipline.
Implementation Without Another Reorganisation
The change does not require dismantling delivery teams or renaming every function. It requires changing the decisions around them.
- Freeze new mode classifications. Existing labels may remain temporarily for reporting, but no new proposal should receive governance solely from a mode.
- Reassess the active portfolio. For each initiative, record uncertainty, reversibility, operational criticality and coupling. Do this in a working session with sponsors, finance, architecture, operations and delivery leads.
- Reconstruct full cost and dependency. Move shared work into view, even if accounting treatment remains unchanged. Identify capacity bottlenecks for the next two quarters.
- Set the next evidence point. Every initiative receives a dated continuation decision and a statement of what evidence will be examined. Programmes previously treated as fully committed should be included.
- Pilot one common review. Select a mixed group of six to eight initiatives and make continuation, scope and funding decisions using the common architecture. Record where evidence was insufficient.
- Retire the separate portfolio forums. Specialist technical or service assurance may remain, but investment decisions should return to one accountable portfolio body.
Roles should also be explicit. The sponsor owns the investment thesis and outcome. The delivery lead owns credible execution evidence. Finance owns the integrity of full cost, not the defence of the original budget. Architecture and operations advise on reversibility, criticality and coupling. The portfolio chair owns comparison and reallocation across initiatives.
Within two quarterly cycles, leaders should expect to see fewer debates about category and more about evidence, exposure and opportunity cost. If the same initiatives continue with the same funding after every review, the new language has merely decorated the old process.
Recommendation
Organisations should retire bimodal IT as a portfolio doctrine while preserving its original insight about different delivery conditions. They should govern all technology-enabled change through one investment portfolio, using uncertainty, reversibility, operational criticality and coupling to determine funding cadence, assurance and architectural freedom.
The recommendation rests on a practical judgement. The greatest portfolio failures rarely come from choosing the wrong delivery method in isolation. They come from committing too much before uncertainty has reduced, failing to see shared dependencies, protecting sunk commitments and comparing initiatives under incompatible standards of evidence. Two modes make each of those failures easier to conceal.
The alternative is neither one speed nor one method. It is one economic logic with policies fitted to the decision at hand. That is more demanding than a fast lane and a slow lane, but it is also more faithful to how value, risk and dependency actually behave.
A portfolio earns coherence when every initiative, however fashionable or foundational, must justify its next claim on scarce capital.