August was one of those months.
During August, 87.20% of support problems were resolved within our target time, compared with 93.18% in July. Summer holidays inevitably had an impact on the team, although we're pleased that 95.20% of our high-priority problems were still resolved on time.
But this month, we'd like to talk about something a little more interesting than percentages.
What if the best support ticket is one that never needs to be raised?
Over recent months we've been developing our own AI-powered Helpdesk system, and one of the areas we're particularly excited about is its ability to help us look beyond individual support tickets.
 |
| Our Root Cause Analysis stats in our new AI-powered Helpdesk system |
Resolving a customer's immediate problem is obviously important. But sometimes that ticket is a symptom of something else.
Perhaps several customers have encountered variations of the same issue. Perhaps a process is unnecessarily confusing. Perhaps there's something in Alpha Tracker that could be improved so that the question simply doesn't arise again.
Our new system helps us identify those patterns.
Rather than treating every support request as an isolated event, it can help flag tickets that may warrant a Root Cause Analysis. We can then investigate the underlying issue, decide what needs to change and — importantly — track that work through to completion.
That last part matters.
It's very easy to say "we ought to fix that properly". It's rather more useful to have a system which makes sure somebody actually does.
Turning support into product improvement
Our new Root Cause Analysis dashboard is already giving us a different perspective on the workload coming into the helpdesk.
It currently identifies around 83 support tickets per month as potentially preventable through the root-cause work already under way. We've also recorded 56 problems fixed at source.
That might mean a software change, an improvement to documentation, a change to a process or another permanent resolution.
The goal is simple: don't just answer the question more quickly — remove the reason for somebody having to ask it again.
AI is particularly useful here because of the volume of information involved. Humans are very good at investigating an individual issue, but spotting recurring themes across hundreds of support conversations is exactly the sort of task where AI can provide some additional help.
The decisions remain with our team, of course. The AI helps us find the patterns; our support and development teams decide what they mean and what we should do about them.
August's support figures
For completeness, here are our latest results:
| Month | Problems resolved on time | High-priority resolved on time | Problems solved |
|---|---|---|---|
| March | 92.67% | 97.93% | 996 |
| April | 93.21% | 97.92% | 869 |
| May | 94.27% | 96.07% | 803 |
| June | 94.43% | 98.33% | 952 |
| July | 93.18% | 97.41% | 982 |
| August | 87.20% | 95.20% | 742 |
Across those six months, the team has resolved 5,344 support problems.
August's overall result isn't where we'd like it to be, and we'll be working to bring that back up as the holiday period comes to an end.
At the same time, though, we're increasingly measuring support success in another way too: how many future problems can we prevent?
If our new helpdesk helps us answer today's support ticket faster, that's good.
If it helps make sure tomorrow's support ticket never happens at all, that's even better.