Stop the bleeding first
Before digging out, make sure new tickets aren't making it worse. Turn on an auto-reply that says you are behind and when to expect an answer. Customers forgive delay far more readily when they were told about it.
Look at the shape of the pile
Export or filter the queue and answer four questions.
- How many tickets are open, and how many are older than your target?
- Which topics make up most of them?
- Which are waiting on you, and which on the customer?
- Which are at risk: angry, high value or about to breach a target?
Close what is already finished
Backlogs are padded with tickets that are done. Pending tickets where the customer never replied can be closed with a polite note and reopen if they write back. Duplicates can be merged. This alone often shrinks the number by a quarter or more, and the real size is easier to plan around.
Answer by risk, then by age
Start with the at-risk tickets, then the oldest. Use macros for the top topics and answer in batches by topic, because the second answer to the same question takes a fraction of the first. Set a time box per person and rotate so nobody spends all day on the hardest ones.
Be honest with the people still waiting
If a ticket will take longer, say so and say when. A short, specific message beats silence. Offer a quick workaround if there is one.
Find why it grew, and fix that
Once it is down, look at the topics. A flood of the same question is a missing article, a confusing page or a product problem. A flood on one day is a staffing or notification gap. Write the fix down and assign it, or the backlog returns.