550 surveys were mailed out in late February, 1992. 102 responses were received by 4 April, an 18.5% return rate. 4 surveys were subsequently received but are not reflected in the data. I attribute the improvement in response (from 8% last year to 18.5% this year) to the CGDC Board's decision to include business reply envelopes. Not all surveys were completely filled out.
135 individual rankings of 50 companies were included with the returned surveys. Only 16 of the companies received more than 2 rankings, thus being included in our publisher information.
The week before DevCon, over 200 phone survey calls were made to gather a limited data set that would allow us to judge the match between returned surveys and the population at large. 48 responses were gathered.
At the 1992 DevCon, I presented basic developer and publisher information in a lecture. Notes from that lecture are in the 1992 DevCon Proceedings, but some of that information will be repeated here.
This was a major issue at the 1991 DevCon. To address it we added the phone survey this year and included Business Reply envelopes to encourage response. Our response rate was up greatly, and the phone survey allowed us to compare the self-selected sample of those who returned surveys to the population at large.
With 102 returned surveys, the developer data is even better than last year. By comparing the phone data with the returned survey data, we see that the distribution of job functions is similar:
| 1992 | Phone | 1991 | |
| Programmer | 55% | 46% | 46% |
| Graphic Artist | 15% | 13% | 11% |
| Music/SFX | 6% | 4% | ** |
| Designer | 46% | 44% | 51% |
| Manager | 44% | 31% | 23% |
| QA | 10% | 2% | 4% |
| Other Job | 17% | 13% | 11% |
| Mean Age | 34.7 | 33.1 | 35.5 |
| Mean Years Active | 7.4 | 7.4 | n/a |
| Mean Education | BA | BA | n/a |
| Support Floppy | 89% | 81% | 81% |
| Support Cartridges | 42% | 44% | 32% |
| Support Coin-Op | 3% | 4% | 7% |
| Support CD | 35% | 35% | 11% |
There are some inconsistencies between the 1992 Survey and the phone results. Notably the percentage of Managers and QA people in the two samples. But overall the similarity is impressive. Our developer data is quite good.
As last year, the limited number of responses for most publishers limits the quality of the publisher data.
| Publisher | Number of Rankings | |
| 1992 | 1991 | |
| Electronic Arts | 20 | 20 |
| Broderbund | 12 | 12 |
| Accolade | 8 | 10 |
| Disney | 6 | 6 |
| Maxis | 6 | 2 |
| Sierra | 6 | 5 |
| Three-Sixty | 6 | 1 |
| SSI | 5 | 6 |
| Konami | 4 | 3 |
| Sega | 4 | 5 |
| Spectrum | 4 | 4 |
| Acclaim | 3 | 1 |
| Activision | 3 | 9 |
| New World | 3 | 1 |
| Software Toolworks | 3 | 0 |
| Virgin | 3 | 4 |
There's no clear trend up or down in the responses for these publishers. I believe the nature of the industry is that we will never get as good data on publishers as we do on developers, because most developers have limited experience with multiple publishers. The exceptions are EA, Broderbund, and possibly Accolade. Their numbers are based upon larger numbers of responses.
Absolutely. Our sampling on developer data is much better this year. The publisher data suffers from the same limitations as last year, so except for the Big 2 or 3, consider the publisher data somewhat suspect. If you want to evaluate publishers, think of the survey data as collecting opinions from people. Then go collect some additional opinions from people who'll tell you why they think this publisher or that publisher is good or bad.
The development community divided itself sharply since the 1991 Survey. The number of people describing themselves as ÒIndependentÓ dropped sharply, accompanied by rises in the number of ÒCompany PrincipalsÓ and ÒEmployeesÓ.
| 1992 | 1991 | |
| Principals | 54% | 40% |
| Independent | 18% | 46% |
| Employee | 30% | 5% |
This realignment caught me entirely by surprise. Next year's survey will be more representative of this year's population distribution.
88% of survey respondents work full-time in the industry, 9% part-time, and only 3% are hobbyist or wannabes. I know there are more of the last category out there, and I suspect they are not returning surveys because 1) they don't think we want their input; or 2) because they don't hear about the conference and register early enough to catch the survey mailing.
Of the 102 respondents, 13 were female, 86 male. The mean level of education is a 4 year (Bachelor's Degree). Virtually every respondent had spent at least some time in college. The mean number of years in the industry is 7.4.
Mean age was 34.7 (34.3 for males, 36.0 for females). Ages ranged from 22 to 60 (see graph, below).
| programmer/engineers | 56 |
| designer | 47 |
| manager | 45 |
| graphic artist | 15 |
| quality assurance | 10 |
| music/sound fx artists | 6 |
| other | 17 |
Write-ins included Writer, Art Director, Producer, Marketing, Publisher, Financier, and others.
| (in $1,000s) | Mean | Median |
| All | 51.7 | 40.0 |
| Male | 50.0 | 40.0 |
| Female | 47.6 | 40.0 |
| Manager | 59.0 | 52.5 |
| Graphic Artist | 53.1 | 40.0 |
| Designer | 48.7 | 39.0 |
| Programmer | 45.7 | 35.5 |
| QA | 42.9 | 30.0 |
| Music/SFX | 40.0 | 35.0 |
A smaller percentage of developers are performing conversions this year, but far more are part of companies that publish and/or distribute. I believe this reflects the trend away from small independent groups to larger groups and more employees in the industry.
| 1992 | 1991 | |
| Create Originals | 94% | 88% |
| Do Conversions | 30% | 40% |
| Publish Products | 34% | 18% |
| Distribute Products | 32% | 12% |
| Create Tools | 32% | n/a |
While we did not collect actual costs of product development, the percentages of development cost spent on various categories were as follows:
| 1992 | 1991 | |
| Design | 13% | 17% |
| Programming | 37% | 43% |
| Graphic Art | 24% | 16% |
| Music/SFX | 7% | 6% |
| QA | 7% | 8% |
| Management/Admin | 12% | 12% |
| Floppy | 14 |
| ROM Cart. | 11 |
| Hand-held | 8 |
| CD | 14 |
| Online | 10 |
| Floppy | 6 |
| ROM Cart. | 7 |
| Hand-held | 5 |
| CD | 7 |
| Online | 4 |
Mean time to sign a contract from walking in the front door is 132 days. Last year it was 75 days. The median is 90 days. There's still very wide variation in the length of time it takes. One report was 2 years and still trying, one was 20 minutes.
Mean royalty percentage on originals is down slightly from last year, but still in the 12% range. Mean advances for originals are up considerably, from the $50K range last year to the $90K range this year. Mean conversion advances are upon from the $30K range last year to the $70K range this year.
Percentage of original product development deals which include required sales quotas for the publisher/distributor to maintain a license: 37%
Percentage of original product development deals which leave copyright with the developer: 57%
Percentage of original product development deals which leave source code with the publisher: 72%
Percentage of conversion deals which include royalties to the converting developer: 63%
Only 12% of developers routinely use an intermediary, and 17% sometimes. 51% never do. The most popular intermediary is a lawyer, followed by an agent.
Following is a chart of publisher ratings. Note that only the mean values of Tools & Technical Support and Ease of Negotiation are below the nominal average value of 5.5. Many of the remaining mean values are above 6.5.
Generally speaking, the individual publisher ratings indicate that developers are collectively more pleased with publishers. There are a variety of possible explanations that come to mind, but whatever the reason, publishers deserve credit for the progress they have made.
Next year the survey will be better targetted toward the population we've become, with questions more suited for the majority of respondents. If you have questions about or for the survey, I can be reached at oldsurveys@enginesofmischief.com. (address good as of 2005)