Growth and Underinvestment

Thinking Tools4.1 Root Causes and the Case for Going BackAlso known as Growth and Underinvestment Archetype

From Jay W. Forrester's market growth model (Market Growth as Influenced by Capital Investment, 1968), named as an archetype by Peter M. Senge (The Fifth Discipline, 1990), and modeled by John D. Sterman (Business Dynamics, 2000).

Describes how growth strains capacity, service slips, demand softens, and the firm reads the softening as a market ceiling and underinvests, locking in the limit.

A software firm lets support response slip, reads the slower sales as a full market, and stops growing

The firm hires support engineers only when response time is worse than its standard. At rest the standard is two days: response drifts to nearly two days, new sales halve, and the firm sees a ceiling. Hold it to one day and Play: hiring keeps up and sales recover. Illustrative.

Select a loop to trace it, or a box to read it.R1, Word of mouth. More customers tell more prospects, who become customers and tell others. No negative (−) links: an even count, so the loop reinforces.B1, Slow support. More customers slow the response, and slow support cuts new sales: this is the softening the firm misreads as a full market. One negative (−) link: an odd count, so the loop balances.B2, Capacity. Slow response sets off hiring, and months later the new engineers bring response back, but only down to the standard the firm holds. One negative (−) link: an odd count, so the loop balances.Customers. Firms on the subscription. About 2 percent leave each month at a one-day response, more as response slows.Response time. Days to answer a support ticket. One support engineer covers 100 customers at one day.New sales a month. Word of mouth from customers plus the sales team's own pipeline, cut sharply when prospects hear that support is slow.Support engineers. Ten at the start. New hires join after the months it takes to recruit and train them.Hiring. Engineers ordered each month: enough to close half the gap between today's staff and what the firm's response standard needs. While response is inside the standard, nobody is hired, however slow sales get.
2.0
3

New sales a month

0: 1501: 1562: 1503: 1434: 1355: 1276: 1207: 1138: 1079: 10310: 9811: 9512: 9213: 9014: 8815: 8616: 8517: 8418: 8319: 8220: 8221: 8222: 8123: 8124: 81010020030001234567891011121314151617181920212223240: 150: 1562: 150: 1434: 135: 1276: 120: 1138: 107: 10310: 98: 9512: 92: 9014: 88: 8616: 85: 8418: 83: 8220: 82: 8222: 81: 8124: 810100200300024681012141618202224

Response time, days

0: 1.01: 1.12: 1.33: 1.44: 1.55: 1.66: 1.77: 1.78: 1.89: 1.810: 1.911: 1.912: 1.913: 1.914: 1.915: 2.016: 2.017: 2.018: 2.019: 2.020: 2.021: 2.022: 2.023: 2.024: 2.00.00.51.01.52.001234567891011121314151617181920212223240: 1.0: 1.12: 1.3: 1.44: 1.5: 1.66: 1.7: 1.78: 1.8: 1.810: 1.9: 1.912: 1.9: 1.914: 1.9: 2.016: 2.0: 2.018: 2.0: 2.020: 2.0: 2.022: 2.0: 2.024: 2.00.00.51.01.52.0024681012141618202224

Month 24 of 24

Customers
1,994
New sales a month
81
Response time, days
2.0
Support engineers
10

R1, Word of mouth
More customers tell more prospects, who become customers and tell others. No negative (−) links: an even count, so the loop reinforces.
B1, Slow support
More customers slow the response, and slow support cuts new sales: this is the softening the firm misreads as a full market. One negative (−) link: an odd count, so the loop balances.
B2, Capacity
Slow response sets off hiring, and months later the new engineers bring response back, but only down to the standard the firm holds. One negative (−) link: an odd count, so the loop balances.
Customers
Firms on the subscription. About 2 percent leave each month at a one-day response, more as response slows.
Response time
Days to answer a support ticket. One support engineer covers 100 customers at one day.
New sales a month
Word of mouth from customers plus the sales team's own pipeline, cut sharply when prospects hear that support is slow.
Support engineers
Ten at the start. New hires join after the months it takes to recruit and train them.
Hiring
Engineers ordered each month: enough to close half the gap between today's staff and what the firm's response standard needs. While response is inside the standard, nobody is hired, however slow sales get.

An example

A company selling scheduling software to dental offices had a thousand customers, ten support engineers, and about 150 new sales a month. Its rule was to hire support only when the average response time passed two days. Response drifted from one day toward two, dentists heard about it from colleagues, and over two years new sales fell by almost half. The leadership team concluded that the dental market was nearly saturated and cut the next year's hiring plan.

A new head of customer success read the win-loss notes first. Most lost prospects mentioned slow support; few mentioned price or features. She drew the growth loop, the slow-support loop, and the hiring rule, and ran the same two years with the standard held at one day: about 3,900 customers instead of 2,000, with new sales still rising. The firm reset the standard to one business day, hired six engineers ahead of the numbers, and began reviewing new sales and response time side by side every month.

Why it works

The archetype hides because the firm's own decision rule is inside the loop. Growth strains capacity, service slips, demand softens, and the softening looks like evidence from the market. The rule for investing waits for demand to prove itself, so weaker demand means less investment, which keeps service poor, which keeps demand weak. Each step is reasonable on its own, and the result is a ceiling the firm built for itself.

Two things make it worse. Capacity takes months to add, so even a correct decision arrives late, and the standard that triggers investment tends to drift, since tolerating a slightly slower response each quarter is easier than hiring. Drawing the loop with the standard on it shows that the limit sits in a number the firm controls, and that the test is simple: did service slip before demand did?

The thinking behind it

Jay W. Forrester Jay W. Forrester, the MIT engineer and management theorist who founded system dynamics with Industrial Dynamics (1961), modeled organizations as stocks, flows, and feedback loops with delays. This archetype comes from his market growth model, set out in Market Growth as Influenced by Capital Investment (Industrial Management Review, 1968), in which a firm's own decisions about capacity shape the demand it then reads as the state of the market.

Peter M. Senge Peter M. Senge named growth and underinvestment as a system archetype in The Fifth Discipline (1990). In the archetype, growth nearing a limit could be carried further by investing in capacity, but the investment waits on proof of demand, and the performance standard the firm holds decides how long it waits. Where the standard slips, the investment never comes, and the limit the firm sees is partly of its own making.

John D. Sterman John D. Sterman modeled the structure in Business Dynamics (2000), the standard system dynamics textbook, where capacity takes time to acquire and the rule for acquiring it is part of the system under study. His beer game experiments, reported in Management Science in 1989, showed that managers misjudge feedback and delays of this kind, which is how a firm can mistake its own slow response for a saturated market.

How to run it

  1. Chart growth and service together. Put new sales, capacity, and a service measure such as response time on one page, month by month.
  2. Check which slipped first. If the service measure worsened before sales slowed, capacity is the likely constraint.
  3. Ask lost customers why. Win-loss notes and exit interviews show whether prospects left over the market or over the service.
  4. Write down the investment trigger. Name the number at which the firm adds capacity, and check whether it has drifted.
  5. Hold the standard. Set it to what customers need, and invest to it even while sales look soft.
  6. Allow for the lead time. Order capacity early enough that it arrives before service slips again.
  7. Watch demand respond. Track new sales against the service measure for two or three quarters to confirm the link.

Where it goes wrong

The defining error is reading a self-made limit as a market limit. Sales slow, the plan says the market is maturing, and the investment that would restore growth is cut. Spot it when the explanation for the slowdown never mentions the firm's own service numbers, and recover by checking win-loss notes against response time before anyone sizes the market again.

A second failure is a standard that drifts. Each quarter the acceptable response gets a little longer, nobody decides it, and the trigger for investing moves out of reach; write the standard down and review it. A third is overcorrecting by building far ahead of any evidence. The archetype argues for holding a service standard, and it gives no license to add capacity without limit.

When to use something else

  • Limits to Growth. When the limit is real and outside the firm's control, and the question is what tightens as growth continues.
  • Eroding Goals. When the main problem is the drifting standard itself, on any target, with or without growth.
  • Queueing Theory. When the task is sizing the capacity: how many engineers keep response under the standard at a given volume.

Where it sits in SOLVE eX

Go to the source

  • Market Growth as Influenced by Capital Investment, Industrial Management Review, 1968
  • The Fifth Discipline, 1990
  • Business Dynamics: Systems Thinking and Modeling for a Complex World, 2000

Back to the catalog