Do We Know What We Will Do This September?
This time last year the blog post I Know What You Did Last September looked at 2018 data and made predictions for September 2019. At that point I was confident that the September 2019 data would inform the September 2020 data… but back in March the world was turned upside down.
As an aside, when such calamitous events happen, the Internet fills with spurious claims of prediction. For example, some claimed Nostradamus predicted the 2020 pandemic but such claims are debunked at the fact checking website Snopes.
So before we look at predicting this year, we should take a backward look….
How Did The 2019 Predictions Turn Out?
At the start of last September some predictions were made based on 2018 data. The table shows the prediction and the reality.
|Prediction||Result?||What actually happened?|
|More Service Request tickets from students will be logged, particularly from undergraduate students.||Tickets from students rose overall by 60% to 8748 compared with a 22% rise for tickets from staff to 21732. When we filter to Service Requests we saw 77% rise from students compared with 21% from staff.|
|Many of these tickets will be handled by teams within University Secretary’s Group (USG).||In 2018 248 Service Requests identifiable as from students were handled by USG, this rose to 2864 in 2019. This evidences increased service support maturity – the call load for USG rose from 4201 to 10911.|
|As a consequence, the percentage of tickets arising from staff will fall.||Tickets from staff rose overall but dropped as percentage from 40% of calls in 2018 to 35% of calls in 2019.|
|Whilst the number of tickets for IT and Library will not change greatly, their percentage of the total tickets will fall.||Tickets overall rose in September 2019 to 61469 from 44313 in 2018 (38% rise) but calls for IT and Library only rose by 22%.|
|Consequently, the percentage of tickets that are incidents will fall from 16% to below 10%.||Incidents fell from 18% of calls 2018 to 11% in 2019.|
|Finance Helpline will overtake IS Helpline in student tickets logged.||This did occur but was not as clear cut. IS Helpline resolved more student calls without requiring escalation.|
Now the predictions were hardly earth-shattering insights but they did indicate that service data can be used with some degree of confidence and as result difficult resourcing decisions can be better informed.
What Are The Predictions For 2020?
Predicting overall ticket numbers is difficult. Using the 2019 August to September trend against the 2020 August data gives the following prediction by support grouping:
|Operator Group Org Unit||August, 2019||September, 2019||August, 2020||Model Prediction of Tickets for September 2020||Comment|
|Information Services||24,539||40,081||17,774||29,031||Likely to be an under-estimate as shift-left means tickets will be resolved at first touch by service desk(s)|
|University Secretarys Group||12,950||31,146||10,208||24,551||Likely to be an under-estimate as more teams are using, and more tasks are being recorded in UniDesk|
|Corporate Services||14,076||16,689||11,310||13,409||Expect this to be the most reliable estimate of the three support groups.|
For nearly six months service support has been provided with virtually no in person interactions. As a result, Quick Calls (recording rapid, transactional in person support tickets, completed at first touch) and most of telephone calls were no longer available. Almost all support became asynchronous through either email or self service portal.
This may have encouraged some users to search for answers and as a consequence not need to raise a ticket.
Another major change is the availability of the student service desk EdHelp. Since EdHelp is accessed through Self-Service rather then email, the percentage of tickets where the user is identified will increase.
The increase in teams using the service management tool (5 of the 22 teams in University Secretary’s Group are new since last September) may also be a factor.
Predicting service demand is possible using data and information but requires additional contextual knowledge to allow interpretation using business context. Application of the simple but very valid Deming Cycle – Plan, Do, Check, Act – to predictive use of data to refine over time allows increased confidence. The aim is not to be absolutely right, but rather, less wrong.