Managing Acceleration: Complexity and Trade-offs
Two firms running the same project do not pay the same price for the same six months. Here is what sets the steepness of your own curve, what to do when complexity is the thing driving it, and the single number a 1989 practitioner source offers for pricing a compression decision.
Five factors that set the steepness of your curve
R4 treats the shape of the time-cost curve as general and its gradient as specific. Why cost rises at all, and why it rises at an increasing rate, is covered separately in why costs increase when projects accelerate. What follows is about how much.
THE FIVE FACTORS INFLUENCING THE COST OF ACCELERATION
| Factor | Direction | Mechanism R4 gives | Caveat R4 attaches |
|---|---|---|---|
| Technological difficulty — projects pushing the state of the art | Steeper curve, more severe penalties | Two reasons. Projects closer to the state of the art involve more uncertainty, so information gaps are more severe when compression increases task overlap. They also involve more tasks where it is uncertain which of several approaches will work, so compression means funding more approaches across more technical tasks. | — |
| Firm size | Larger firms pay more | More coordination and consultation may be required before any major decision. Accelerating an important development project would almost certainly require substantial coordination and planning; the same decision might be taken in a small firm on the basis of a single conversation or meeting. | The effect depends on organisational factors. A highly flexible organisation with an extremely responsive R&D support staff may mitigate it substantially — but other things equal, size carries a penalty. |
| Relevant prior experience | Less experience, dearer acceleration | Three sub-mechanisms. Firms that have run similar projects know which parts are most cost sensitive under compression; experience may reduce the cost of increased overlap between tasks; and experienced firms may choose more wisely among the approaches to uncertain technical tasks. | — |
| Project scale | Large projects, greater penalties | Large and expensive projects obviously cost more per unit of acceleration in absolute terms, but even relatively the penalty may be greater: large projects may already sit in a region of more severely diminishing returns to added personnel at normal staffing, and they suffer more from network compression as the task network densifies. | Complexity is a related but distinct driver of time sensitivity, treated separately below. |
| Professional workers in the labour mix | High professional content, smaller penalties | When a project is accelerated, labourers must be paid more in overtime even if no additional workers are hired. Professional workers are usually expected to contribute whatever time is necessary to complete the task, without additional compensation. | This considers only wages and salaries. It leaves out any negative effects from declines in morale or organisational effectiveness. |
Close paraphrase of R4's five factors. Four of the five are largely fixed in the short run, which is what makes this a capability question rather than a willpower question.
Complexity, and the remedy R4 names
Complexity is treated as a driver in its own right. R4 reviews data quoted from a consulting firm's quarterly, which was itself reviewing others' work, showing substantial cost overruns associated with very complex projects. All three figures below are externally cited — quoted by R4 from that review, not measured by R4, and third-hand by the time they reach this page.
- A large computer manufacturer's mainframe product line development spent two to five times its original estimate — externally cited figure, quoted from a consulting review.
- Offshore oil development estimates often fell 100 to 150 percent short — externally cited figure, same source.
- A survey of 44 plant design and construction projects in synthetic fuels and chemicals showed significant cost overruns in more than three fourths of them — a finding of another study, cited from the same source.
The competitive-strategy layer
R4's five factors are meant to be read comparatively. The observations, it says, suggest to R&D managers the conditions under which one is more likely to be successful in competitive rivalries.
Reading your position against a rival's
The conclusion R4 draws is that knowledge of the relative time-cost tradeoff of the firm versus its competition should be an important consideration in competitive strategy. Racing on speed is presented as a capability question — a position on a curve set by experience, scale, size and technological difficulty — not a matter of resolve.
Estimating the cost penalty — the article's headline metric
R4 makes results from different studies comparable by converting each into one quantity: the percentage increase in cost that results from a 1 percent decrease in project duration. That is a cost-duration elasticity, and it is what the article is usually quoted for.
PERCENTAGE INCREASE IN R&D COST FOR A ONE PERCENT REDUCTION IN PROJECT DURATION
| Project type and source | Value | Provenance |
|---|---|---|
| Hardware projects — an econometric study of chemical, electrical and machinery R&D projects | 1.75 | Finding of that one study, recomputed by R4 into the common metric |
| Software projects — a software engineering economics study | .88 | Finding of that one study, recomputed by R4 into the common metric |
| Software projects — a software cost estimation study | 2.00 | Finding of that one study, recomputed by R4 into the common metric |
Each value is computed for a point about 10 percent above the minimum possible completion time. The two software estimates disagree by more than a factor of two and straddle the hardware figure, so no blanket claim about software's compressibility survives this table.
Worked example — pricing a six-month acceleration
R4 closes on a hypothetical two-firm comparison. Every figure in it is invented for the illustration; none of it is measured.
R4's hypothetical, step by step
Set the baseline
Two firms, each with an R&D project originally planned to run five years at a development cost of $100 million. Hypothetical figures.
State the compression
Each contemplates accelerating the project by six months.
Convert to a percentage
Six months on a five-year schedule is a 10 percent reduction in duration.
Apply the elasticity band
At 1 to 2 percent cost increase per 1 percent duration reduction, a 10 percent reduction implies a 10 to 20 percent cost increase.
Read the result
An estimated total increase in R&D project cost on the order of $10 to $20 million. A worked illustrative result, not a benchmark.
Adjust for who you are
R4's own qualifier: if one of the two firms were larger, or had less relevant developmental experience, or used relatively less professional labour on the project, the cost penalty could be greater.
Decision rules, and what the number leaves out
R4's operating rules for an acceleration decision
- Before entering a speed race, work out the size of the cost penalty a contemplated acceleration will produce. R4 makes this the precondition for a meaningful strategic calculus.
- Know your time-cost curve relative to your competitors'.
- Adjust the elasticity for where you sit: penalties are greater near the absolute minimum completion time and less severe at durations well above it.
- When crashing a critical path, take the cheapest task first, then the next cheapest — and expect each further increment to cost more.
- Expect to fund parallel approaches if a successful one must be found quickly, including approaches serial search would never have needed.
- Do not add technical people faster than they can be absorbed; the training penalty scales with the ratio of new to experienced people.
- To avoid complexity-related penalties, disaggregate major development projects and introduce features incrementally.
- Expect greater relative penalties if the project is near the state of the art, large in scale, run by a large organisation, or in an unfamiliar field.
- Do not take the accountant's number as the cost of acceleration — the loss to other projects from pulling talent away will not appear in it.
One last framing point. An acceleration decision is a purchase of time at a rising price, so it belongs in the same conversation as the rest of the project's economics — the discounting conventions in the financial frame for R&D management, the treatment of skewed cost outcomes in risk and uncertainty in R&D financial analysis, and the question of whether the earlier launch is worth what it costs, which is where an option frame such as real options valuation does its work.
What to carry forward
- The curve's shape is general; its gradient is yours. Technological difficulty, firm size, relevant experience, project scale and professional-labour content set how steep it is.
- Four of those five are largely fixed in the short run, which makes racing on speed a capability question rather than a matter of will.
- Complexity has a structural remedy: disaggregate the project and ship features incrementally. Managing the same large project harder is not the answer R4 gives.
- The elasticity is the usable output — cost percentage per duration percentage — but it is a point estimate about 10 percent above minimum duration, and it rises as you approach that minimum.
- The $10 to $20 million in the worked example is an illustration built from invented inputs. Reproduce the method, not the number.
Frequently asked questions
Can I use 1 to 2 percent as a planning rule?
As an order of magnitude, with the conditions attached. It is R4's own synthesis of three point estimates — 1.75 for hardware, .88 and 2.00 for software — each computed for a point about 10 percent above the minimum possible completion time. It is not a measured constant, it does not hold at other places on the curve, and two firms running the same project may face different values.
Why would being a larger firm make acceleration more expensive?
R4's mechanism is coordination overhead. A major decision such as accelerating an important development project would almost certainly require substantial coordination and planning, where the same acceleration might be decided in a small firm on the basis of a single conversation. The effect is conditional: a highly flexible organisation with a responsive R&D support function may mitigate it substantially.
Is a professional-heavy project genuinely cheaper to accelerate?
On the wages line, yes — labourers must be paid overtime, professional workers are usually expected to contribute whatever time the task needs without additional compensation. R4 states that the comparison considers only wages and salaries and excludes any negative effects from declines in morale or organisational effectiveness, so the apparent saving is narrower than it looks.
What should I do if complexity is what is driving our overruns?
The remedy R4 reports is structural: some companies have disaggregated major development projects and spread them over time, introducing new features incrementally rather than all at once. That reduces project scale and network density, which are two of the factors that make the curve steep, rather than trying to run the same large project more tightly.
How do I turn this into a number for my sponsor?
Follow the worked example: express the contemplated compression as a percentage of planned duration, apply the elasticity band to the planned development cost, then adjust for your position on the five factors and for how close you already are to minimum duration. State that the figure excludes the opportunity cost of the people being reassigned, because R4 excludes it and says it will not appear in accounting figures.
Do the software figures mean software is easier to compress than hardware?
No, and that is the interesting thing about the table. The two software estimates are .88 and 2.00, which sit on either side of the hardware figure of 1.75. One credible estimate says software is roughly half as expensive to accelerate as hardware and another says it is more expensive, so any blanket claim about software's compressibility is unsupported by this evidence.
References and source attribution
- R4 - why costs increase when projects accelerate. Practitioner review and synthesis article in a journal for research and technology management, March-April 1989; 3 printed pages; 7 references; one table and one figure. Sections used here: the five factors influencing the cost of acceleration, the complexity argument, the competitive-strategy discussion, the elasticity table and the closing hypothetical.
- The econometric study of chemical, electrical and machinery R&D projects, and the two software cost studies whose results R4 recomputes into a common metric, are cited within R4 and were not supplied to this library.
- The consulting firm's quarterly review from which R4 quotes the mainframe, offshore oil and plant-construction overrun figures was not supplied to this library. Those figures are recorded here as R4 quotes them, at third hand.
- Eleven copyrighted journal articles on R&D project management, supplied as a reading set for a literature review and profiled for this library. Front matter, abstracts, framework sections, tables and figures were read; article bodies were not reproduced, and all content here is paraphrase. The set is a reading list, not a systematic survey of the field.
- Supplied teaching source for this library (research methods and research process materials). Used here for page conventions and voice only; it does not treat schedule compression.
Suggested questions for Ask KEVOS
- Score our project against the five factors and tell me whether our curve is steep or shallow.
- Run the worked example with our own planned duration and development cost.
- How much closer to minimum duration are we than the point where the published elasticities were measured?
- Draft the case for disaggregating our release into incremental features instead of accelerating it.
- Compare our position with a competitor that has run three similar projects and we have run none.
- What should I state as excluded when I present this acceleration cost?