National Capital FreeNet System Usage Report October 1995 Ian! D. Allen under contract to The National Capital FreeNet 1. Preface This report covers the usage of the system by members and guests over the past seven months. A sample set of seven hard-copy graphs is distributed separately with this report and referred to as G# where is 1 through 7. 1.1. Important conclusions The graphs and data all show a significant decline in the amount of time NCF's members spend online, without any corresponding increase in the number of active members. The decline is visible in all the graphs, starting at the end of April. Community interest in NCF, based on guest account activity, is also less than earlier in the year. Interest in NCF from the Web is still increasing; but, the increase is itself slowing down. Both NCF's Executive Director, Lisa Donnelly, and I think these trends indicate that NCF's popularity may have peaked and is now declining. Given trends in telecommunications, we are concerned that NCF is beginning to become increasingly unimportant in the local community, and immediate Board attention is needed to determine a strategy for handling this. Possible Board tactics include either drafting a plan for recovering the lost interest, or planning to downsize operations in 1996. The Finance, Planning and Development committees, along with the Board as a whole, must work together for the remainder of the year to carefully plot NCF's course. 1.2. Executive Summary The number of active users in NCF's user community stopped growing after April of this year and went into a decline until July. The number of NCF members has risen 18 percent after April (G#1); but, the number of active users of NCF has risen only 5 percent in the same period (G#2). The rate of online registrations and account valida- tions also decreased sharply after April and both continue to remain below levels set early in the year. The number of guest telnet connections and the system utilization by guests overall (G#7) both decreased after April. The rate of increase in access to NCF from the Web has also slowed since May. The overall use of the hardware by members (the number of online user-hours of service provided) decreased sharply at the end of April and remains 15 percent below levels set in March (G#4), despite having 10,000 more users to accommo- date. This drop corresponds with a sharp decrease after April in both members' individual session time (G#5) and members' total daily online time (G#3). Average daily system load on the machines handling modems peaked in early March and dropped again through late March and mid-April. Loads rose again slightly in August and dropped in September. 1.3. Time Line When attempting to infer causes for usage changes, the following NCF dates may prove useful: early March 1995 -- mandatory registration ID mid-March 1995 -- removal of binary news groups end April 1995 -- time limits Be wary of drawing conclusions based on these events alone; seasonal and market factors are also at work. Also consider the "percolation time" required for changes to be noticed and affect statistics. For example, since only half the active members use the system in any given week, any online change will take at least a week to be noticed by half the active NCF members. 2. Indicators The following observations were made by analysing graphs of the data that I post regularly to the NCF System Design news group. 2.1. Rate of online registrations falling (G#1) The number of people filling out online registration forms was growing exponentially until September of last year (1994), at which point the growth suddenly stopped and remained a constant 800 people per week for about 8 months. At the end of April, the growth in registrations fell sharply and, except for a brief spike at the start of Sep- tember, is now dipping below 550 people per week -- a drop of about 30 percent since April. 2.2. Rate of growth of new accounts slows (G#1) The number of people actually getting NCF accounts was growing exponentially until March/April, at which point the growth slowed to become currently linear or sub-linear. 2.3. More members use NCF less (G#3,4,5) The overall use of the hardware by members (the number of online user-hours of service provided) decreased sharply at the end of April and remains 15 percent below levels set in March (G#4), despite having 10,000 more users to accommo- date. Overall, NCF's current 48,810 members spend 15 per- cent less time using NCF than did the 38,490 members that were using NCF in March. This drop in system use correlates with a sharp decrease after April in both members' individual session time (G#5) and members' total daily online time (G#3). 2.4. Slow growth in NCF community (G#2) The number of active users of NCF stopped increasing at the end of April and declined throughout May and June, for a total drop of about 5 percent over the two months. The number of NCF members increased 6 percent during the same period. The decline in active users stopped in July, and reversed itself in August, for a total gain of about eight percent between July 1 and October 10. The number of NCF members increased 10 percent during the same period. Overall, the number of NCF members has risen 18 percent since May; but, the number of active users of NCF has risen only 5 percent in the same period. 2.5. Guest telnet connections decline (G#7) The number of telnet connections to NCF by guest users declined from April to August, dropping by about 30 percent. The total amount of system use by guests also dropped; guests don't spend as much time online as they did at the beginning of the year. Early September showed an increase of about 15 percent in guest telnet connections. Guest telnet connections since mid-September are declining again. 2.6. Interest in NCF via WWW slows The number of accesses of NCF from the Web was increas- ing exponentially until mid-year, at which point growth dropped to linear or sub-linear. NCF Home Page access counted for 6 percent of all hits last month. 2.7. System load moderate Average daily system load on the machines handling modems peaked in early March and dropped again through late March and mid-April. Loads rose again slightly in August and dropped in September. The load pattern has changed from the summer months where the busiest times occur during the afternoon, to the winter pattern where the load leaps up in late afternoon. 2.8. Pledges constant; donations steady (G#6) Except for a brief increase at the end of April, the average amount pledged by users at registration time remains a constant $7. Average registration donations were about $6-$7 before March of this year. They rose to around $10 during March, April, and May, then declined to a current average of about $8. -- Lisa K. Donnelly, Executive Director * National Capital FreeNet/Libertel de la Capitale nationale * Telecommunities Canada, Board of Directors (613) 241-9554 voice & fax (613) 241-2477