Backlog – what is it?

The simplest definition of a backlog is the number of unresolved customer support requests a company has over time.

The actual definition of “unresolved” can vary from company to company but it usually includes all new and open trouble tickets. Sometimes pending tickets or those that are waiting for information from the customer are taken out of the backlog calculation. Sometimes tickets assigned to specific and solid resolution paths (e.g. feature not in the product) are also excluded.

In an ideal world, there would be a very small backlog and the number of new tickets opened would roughly be equal to the number of existing tickets you close, with the majority of tickets getting resolved within the normal response time. But in the real world, tickets volume can spike or grow too fast, and then backlog starts to grow, to become sometimes the unclimbable mountain of new and opened tickets that we have all faced.

And how can you handle it? How many customers are affected? Do you need more agent resources? Is it a product problem? Which tickets should I address first as a matter of urgency? These are the immediate questions you would like answers to before you start the triage process. But to triage the tickets, someone has to read them. Someone you don’t have since your backlog is growing.

What is the cost of Backlog?

Whichever definition you use, each customer in the backlog represents a “holding” cost to your company or brand. They represent those customers that STILL have a problem with some part of your offering and they are expecting a resolution, quickly. The longer you keep them waiting the greater the frustration and the bigger the holding cost. These customers represent potential detractors and can seriously damage your brand. Ignore them at your peril!


Managing and understanding the nature of backlog is therefore absolutely critical for both support teams and the C-Suite alike. However, this is a catch-22 situation. In other words, to triage and prioritize the backlog you have to understand what it consists of. To understand that you need to know what the tickets are about. To know that you need to open and read every ticket. But the reason you have a backlog in the first place is that you don’t have the resource to handle the uptick in volume ….and so it goes around.

The only sensible way of breaking this vicious circle is to automatically “process” this backlog once it has built up, and then apply similar triage rules to all tickets as they get raised, automatically categorizing every ticket as they come with granular and meaningful “reasons to call”. Then use these new parameters to identify clusters of problems that can be routed or handled more effectively. Enabling a faster and more efficient triage process. This is what an AI-based system can do.

Causes of backlog

Understanding the causes of backlog can also help the system identify the right routing decision or action to be taken.

· Increase in customer usage – general resourcing issue

· Increase in numbers of products or services launched – resource planning

· Technical issues with infrastructure – which system? What problem? Routing to IT

· Partner issues – e.g shipping partners, IT partners, fulfilment partners.

· Product issues – which product? What problem?

· Knowledge or training issues with agents

Being able to breakdown the backlog volume with this level of granularity is invaluable and directly improves prioritization and triage without human resource investment. The direct benefits of improving resolution times and customer score can be directly measured.

How AI helps prioritize and triage backlog.

AI can automatically surface insights such as “50% of the ticket spike was due to an order problem specifically caused by a shipping partner issue”, or “40% was due to a newly released product in one region”.

When comparing to the traditional Agent tagging, it appears that categorization by agents sometimes leaves 70% of customer tickets un-tagged. And when a tag is indeed applied, it is nowhere precise or reliable enough to provide actionable trend or insights. AI, on the other hand, will analyze ALL tickets and will provide an automated, granular, multi-tagging capability which surfaces underlying customer issues un an unprecedented manner.

Was Donald right?

There is a lot of hype and expectation around AI but the world is fast changing and the benefits of these technologies are starting to bear fruit. Hopefully, from this short blog, you can start to see the value of using automatic trouble ticket topic discovery and analytics to help with Backlog triage.

Donald Rumsfeld sort of had it right when he talked about known knowns, known unknowns and unknown unknowns:

· There are known known’s, e.g. you have a backlog of 20,000 tickets and its growing.

· There are known unknowns, you know that a large chunk of this backlog will be related to something you know, either it’s a problem about a product, services or maybe your order, delivery and shipping systems. But this is currently hidden in the ticket so it remains unknown and you can’t use it to triage. With AI you can bring that to the surface at triage time, automatically and in real-time.

· There are unknown unknowns. AI has the ability to discover problems that you did not know of, either because customers describe them with very different wordings than the ones you would use internally, or because the low ticket volume at start kept hiding them as a growing trend.

If you would like to see how such systems can provide benefit to your support systems and triage process, please contact us or click here to schedule a demo.