Four Myths About New Product Failure
R7 closes by naming four received beliefs and setting its own data against each. Three of the four are not about how often products fail at all — they are about where new-product performance comes from, and the answers point away from the R&D budget.
How R7 frames its four myths
R7 is a practitioner-journal report of a firm-level survey, published in 1983 and adapted from a longer academic marketing-management article of the year before. Its concluding section is organised as four popular beliefs, each paired with what the survey found. The paper calls them myths; the more useful reading is that each is a claim the data is capable of testing.
The first myth is the failure-rate claim itself, and it belongs to the sibling page. The disputed 50-to-90 percent statistic, R7's own success, kill and failure rates, the definitions they rest on and the measurement model that produced them are set out in measuring new product success rates. This page concentrates on the three myths about determinants.
THE FOUR MYTHS AS R7 STATES THEM
| The belief | What R7 reports finding instead | Where the evidence sits |
|---|---|---|
| 1. New product failure rates are alarmingly high | Effectiveness is considerably higher than previously reported. Most industrial products in the sample did succeed | Covered on the sibling page — the disputed statistic and the denominator |
| 2. Every firm is unique — those data do not apply to us | The nature of the firm did not have much impact on performance. R7 reports remarkable consistency in effectiveness and efficiency regardless of size, industry or ownership | Performance cross-tabulated against industry, ownership and annual sales |
| 3. R&D spending drives new-product performance | No direct link between spending and effectiveness. Success, failure and kill rates were not tied to R&D inputs. Spending did produce more new-product sales, but only above a threshold and with rapidly diminishing returns | Performance against five R&D-spending bands drawn as approximate quintiles |
| 4. Technological strength is the dominant determinant | Technological and production strengths related to the magnitude of output. Marketing variables were the ones that related to effectiveness and efficiency | Correlations between eight self-rated resource strengths and six performance measures |
Framing and findings from R7 — results of one survey of 103 industrial firms, not norms for new-product programmes.
Myth two: our industry, our size and our ownership make us different
R7 crossed its six performance measures against three firm characteristics — industry, ownership and annual sales. The intended reading of the resulting table is the column of negative significance tests running down it.
FIRM CHARACTERISTICS AGAINST NEW-PRODUCT PERFORMANCE — STUDY FINDINGS
| Grouping | Range of success rates | Range of sales from new products | Range of sales per R&D dollar | Statistically significant? |
|---|---|---|---|---|
| Industry — five categories | 54.5% to 65.5% | 26.8% to 41.6% | 3.69 to 13.93 | No on every effectiveness and magnitude measure. Yes on the efficiency ratio (analysis of variance p = 0.008; cross-tabulation p = 0.021) |
| Ownership — three categories | 47.9% to 63.5% | 28.2% to 35.7% | 3.16 to 16.70 | No on every measure, including efficiency |
| Annual sales — five bands | 50.5% to 65.4% | 24.5% to 41.5% | 1.84 to 23.90 | No on every effectiveness and magnitude measure. Yes on the efficiency ratio (p = 0.006; correlation 0.862, p = 0.001) |
| All firms | 59.37% | 31.89% | 8.30 | — |
Ranges reconstructed from R7's cross-tabulation; findings of that one survey. The success-rate ranges are wide, but no between-group difference was significant on any effectiveness measure.
The single exception is the sales-to-R&D efficiency ratio, which rose strongly with firm size — from 1.84 in the smallest band to 23.90 in the largest — and differed by industry. R7 reads that as an artefact of firm size rather than a performance signal: what varies is the arithmetic of the ratio, not the quality of the programme.
Treat that rejection carefully. R7 shows that three coarse firm descriptors did not separate high from low performers in its sample, which removes an easy way of dismissing external data. It does not establish that no firm-level characteristic matters — only that these three did not, on these measures, in 1983.
Myth three: more R&D spending buys better new-product performance
R7 sorted firms into five R&D-spending bands, expressed as a percentage of sales and chosen to break the sample into roughly equal-sized groups. The bands are quintiles of the sample, not natural breakpoints in the business, and that matters when reading them as thresholds.
PERFORMANCE BY R&D SPENDING BAND — STUDY FINDINGS
| R&D as % of sales | Success % | Killed % | Failed % | Programme rating | Sales from new products % | Sales per R&D dollar |
|---|---|---|---|---|---|---|
| 0 to 0.46 | 59.5 | 19.5 | 21.1 | 1.75 | 29.8 | 29.55 |
| 0.47 to 0.94 | 62.7 | 20.5 | 16.8 | 1.56 | 22.5 | 5.86 |
| 0.95 to 2.08 | 56.6 | 22.3 | 21.2 | 1.91 | 28.0 | 3.79 |
| 2.09 to 4.39 | 62.0 | 22.1 | 15.9 | 1.85 | 35.7 | 2.14 |
| 4.4 and above | 56.7 | 25.1 | 18.3 | 1.80 | 42.9 | 0.91 |
| All firms | 59.37 | 21.92 | 18.71 | 1.78 | 31.89 | 8.30 |
| Significant? | No | No | No | No | Yes (p = 0.060; correlation 0.402, p = 0.001) | Yes (p = 0.001; correlation −0.231, p = 0.010) |
Findings of one 1983 survey. The programme rating is a self-assessment on a −5 to +5 scale. Bands are approximate sample quintiles, stated as such in the original.
Read the table in two halves. The four effectiveness columns on the left do not move: spending more on R&D changed neither the success rate, the kill rate, the failure rate nor the self-rated programme performance. The two columns on the right move sharply, and in opposite directions.
The magnitude relationship is essentially flat across the three lowest bands. Output begins to increase only when R&D spending exceeds roughly 2 percent of sales: the heaviest spenders drew 42.9 percent of sales from new products against about 26 percent for firms spending under 1 percent. The efficiency spread across the same bands is roughly thirty-two-fold, and monotonic.
What R7 concludes from the spending result
Myth four: technological strength determines programme success
The fourth myth is tested with correlations between eight self-rated corporate resource strengths and the six performance measures. Respondents rated their firm against domestic competitors on a −5 to +5 scale, where +5 meant much stronger. The pattern of blanks in the resulting table is the finding.
CORPORATE RESOURCE STRENGTHS AGAINST PERFORMANCE — CORRELATIONS, STUDY FINDINGS
| Resource strength | Success rate | Kill rate | Failure rate | Programme rating | Sales from new products | Sales per R&D dollar |
|---|---|---|---|---|---|---|
| Financial | n.s. | n.s. | n.s. | n.s. | n.s. | 0.191* |
| R&D | n.s. | n.s. | n.s. | 0.219* | 0.242** | n.s. |
| Engineering | n.s. | n.s. | n.s. | 0.312*** | 0.310*** | n.s. |
| Market research | 0.324*** | −0.177* | −0.293*** | 0.479*** | n.s. | 0.223* |
| Management | 0.287** | −0.242** | n.s. | 0.338*** | n.s. | 0.211* |
| Production | n.s. | n.s. | n.s. | 0.325*** | 0.176* | 0.219* |
| Salesforce and distribution | 0.265** | n.s. | −0.254** | 0.300*** | n.s. | 0.233** |
| Advertising and promotion | 0.359*** | −0.243** | −0.262** | 0.392*** | n.s. | 0.249** |
n.s. = not significant. One asterisk = 0.05, two = 0.01, three = 0.001. Pearson correlations on cross-sectional data from one 1983 survey; no causal identification is claimed anywhere in the paper. Resource strengths are firm self-assessments relative to self-nominated competitors.
The three technological rows — R&D, engineering and production — are not significant on any effectiveness column. They never predict the success, kill or failure rate. What they predict is the self-rated programme evaluation and the quantity of new-product sales.
The three marketing rows do the opposite. Market research, salesforce and distribution, and advertising and promotion are the only strengths besides management that touch the effectiveness columns at all, and they carry the expected signs: positive against successes, negative against kills and failures. The strongest single coefficient in the table is market research against the programme rating.
Where the sampled firms rated themselves strongest
- R&D skills and people — mean 2.00
- Engineering skills and people — 1.87
- Management skills and people — 1.80
- Financial resources — 1.60
- Production resources and skills — 1.26
Where they rated themselves weakest
- Salesforce and distribution — 1.18
- Marketing research skills and people — 0.68
- Advertising and promotion skills — 0.26
- R7's reading: industrial firms are often found wanting in their commitment to a market orientation
- The weakest areas are almost exactly the strengths that predicted effectiveness
One result cuts across the technological-versus-marketing split. Management strength was significant against the success rate, kill rate, programme rating and efficiency ratio — everything except the quantity of new products. That gives general management the broadest footprint in the table, and it is the result R7 comments on least.
The decision rules R7 draws from the four
R7's conclusions section converts the four myths into guidance. The rules below are the paper's own. Each lever is chosen to match an objective, which is the practical consequence of separating effectiveness, magnitude and efficiency in the first place.
R7's guidance, as the paper states it
- Treat technological prowess as the lever for the quantity of output, not for the hit rate
- Treat marketing strength as the lever for success rate and for returns on R&D spending
- Balance any increase in R&D expenditure with a greater focus on new-product market research and launch effort
- Do not expect a larger R&D budget to raise the proportion of products that succeed
- Do not use industry, firm size or ownership as grounds for dismissing external new-product data
- Count kills separately from commercial failures when you assess your own programme
How much weight one study can carry
R7 disputes a received statistic on the grounds of sampling, measurement and non-comparable definitions. That critique applies to its own conclusions with the same force.
WHAT LIMITS EACH OF THE THREE DETERMINANT MYTHS
| Limit | What it constrains | Consequence for reuse |
|---|---|---|
| All relationships are Pearson correlations and analyses of variance on cross-sectional data | Every myth-2, myth-3 and myth-4 result | No causal claim is supported. 'Marketing strength drives success' is not what the table shows; co-movement is |
| Resource strengths and the programme rating are self-assessed on a −5 to +5 scale against self-nominated competitors | The whole of myth four, and the programme-rating column throughout | A firm that believes it is marketing-led may rate itself so and rate its programme well. Common-method association is not excluded |
| The unit of analysis is the firm and its programme | All four myths | Nothing here supports a decision about an individual project |
| Firms were drawn from a standing panel known to be active in new products | All four, and myth two most directly | 'Firms do not differ' is a statement about a set already filtered on the behaviour being measured |
| The spending bands are sample quintiles, not business-meaningful breakpoints | Myth three, and the roughly 2 percent threshold | The threshold is a property of where this sample split. Do not quote it as a spending target |
| Very large standard deviations on the effectiveness measures | Every mean in the paper | The success rate's deviation is nearly 25 percentage points. The means describe the sample, not any firm in it |
| The survey was conducted for a 1983 publication | All four myths | Market conditions, product cycles and accounting practice have all moved since |
Limits as stated or evident in R7. The consequence column is this library's synthesis; the paper does not set out such a table.
One further boundary is structural. R7 was supplied to this library inside a reading set assembled for a literature review on R&D project management — not a systematic survey of the field. Any apparent consensus between R7 and a neighbouring paper is a property of that reading list — see eleven R&D management papers compared.
Where a project-level view is what you need, the material is elsewhere: what distinguishes successful R&D projects and sources of uncertainty in R&D projects. For the budget question in financial rather than survey terms, see the financial frame for R&D management.
What to carry forward
- Three of R7's four myths are about determinants, not about failure rates. Those three are the part of the paper that is not covered by the sibling page.
- Industry, firm size and ownership did not separate performers in this sample. That removes an easy dismissal of external data; it does not prove no firm characteristic matters.
- R&D spending moved output volume, not hit rate. Effectiveness measures were flat across every spending band while per-dollar efficiency fell roughly thirty-two-fold.
- Technological strengths correlated with how much came out. Marketing and management strengths correlated with how much of it succeeded.
- The sampled firms rated themselves strongest on R&D and engineering and weakest on advertising and market research — the reverse of what predicted effectiveness in their own data.
- Everything here is cross-sectional correlation from one 1983 survey, on self-rated measures, at firm level. It cannot support a causal claim or a project decision.
Frequently asked questions
Does R7 show that cutting the R&D budget is safe?
No. It shows that in its sample the success, kill and failure rates did not vary with R&D spending, while the quantity of new-product sales did. Cutting spending would be expected to reduce output volume without improving the hit rate. R7 explicitly argues for more new-product commitment, not less.
Is the roughly 2 percent of sales figure a target we should aim at?
No. It is the point in this sample above which new-product output started to rise, and the spending bands were chosen to split the sample into approximately equal groups rather than at business-meaningful breakpoints. Treat it as a description of one 1983 dataset, not as a budget rule.
Why does R7 say technological strength does not matter when R&D strength correlates with the programme rating?
Because the programme rating is a different construct from the effectiveness rates. R&D, engineering and production strengths correlated with the self-rated programme evaluation and with the quantity of new-product sales, but with none of the success, kill or failure rates. R7's claim is specifically about effectiveness.
Can we use the correlation table to decide where to invest in capability?
Only as a prompt, not as evidence. The coefficients are cross-sectional correlations on self-rated strengths, so a firm that considers itself marketing-led is rating both the cause and the effect. Use them to challenge an assumption that capability equals technical capability, then test the question with your own data.
How does this page differ from the one on measuring success rates?
That page covers the disputed 50-to-90 percent failure statistic, the definitions and denominators that decide any such rate, R7's own success and kill and failure figures, and the three-construct measurement model. This page covers the three myths about determinants and the guidance R7 draws from them.
If the results are only correlations, what is the paper actually good for?
It is good for removing unexamined beliefs. The claims that failure rates are known to be very high, that firm characteristics excuse you from external evidence, that R&D budgets buy hit rate, and that technological strength decides programme success were all in circulation without support. R7 is evidence that they were not established, which is a different and more defensible job than establishing their opposites.
References and source attribution
- R7 — most new products do succeed. Practitioner-facing report of a firm-level survey study in a journal for research management, November–December 1983; 6 printed pages; 13 references; four numbered tables and no figures. Adapted with permission from a longer academic marketing-management article published the previous year. Sample: 103 industrial firms, effective response rate 69 percent, drawn from a standing sample of almost 200 industrial firms active in new products maintained since 1974. Sections used here: the four-myth conclusions section, the tables relating performance to firm characteristics, to R&D spending and to corporate resources, and the stated limitations.
- The prior strategic-planning and profit-impact database programme cited within R7 (R&D spending positively related to return on investment for dominant firms only; product quality related to profitability), and a study of industrial firms in another country measuring new-product strategy against competitive advantage, technological innovation and patent protection. Both are figures and findings R7 quotes from elsewhere; they were not supplied to this library and R7 is not itself evidence for them.
- R6 — why R&D projects succeed or fail. Practitioner-facing report of a multi-firm research programme, November–December 1986. Used here only for the contrast between firm-level and project-level evidence.
- Eleven copyrighted journal articles on R&D project management, supplied as a reading set assembled by a student 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 or representative survey of the field.
- Supplied teaching source for this library (research methods and research process materials). Used here for page conventions, voice and the treatment of evidence claims; it does not treat new-product performance.
Suggested questions for Ask KEVOS
- Draft a one-page challenge to the assumption that raising our R&D budget will improve our new-product hit rate.
- List what would have to be true before this study's correlation table could guide a capability investment decision.
- Explain the difference between effectiveness, magnitude and efficiency to a finance director in three sentences.
- Write the counter-argument a sceptical executive would make to the claim that firm size and industry do not matter.
- Turn R7's four myths into four questions we can actually test against our own data.
- Summarise what this paper can and cannot evidence, for inclusion in a literature review.