Support Backlog

Type of metric

Snapshot: size and age at a moment

Grows when

Arrivals per hour exceed resolutions per hour

Usually excluded

Pending on the customer, resolved, closed

Critical age band

Over 24 h since the customer's last message

Burn-down estimate

Backlog ÷ (resolved/h − arrivals/h)

1. 1

   Count what is waiting on you

   Include new and open conversations where the last message is from the customer. Exclude pending (waiting on the customer), resolved and closed chats.

2. 2

   Bucket by age

   Show the backlog in age bands — under 1 hour, 1–4 hours, 4–24 hours, over 24 hours. The last band is past the WhatsApp service window and needs templates.

3. 3

   Watch inflow versus outflow

   Plot new conversations and resolutions per hour on the same chart. The gap between the lines is the backlog's rate of change.

4. 4

   Estimate burn-down time

   Backlog ÷ (resolutions per hour − arrivals per hour) gives the hours to clear it at current pace. If the divisor is zero or negative, the backlog will not clear without changes.

5. 5

   Act on the cause

   Add capacity for a temporary spike, automate repeated questions for a structural rise, and fix the upstream issue — a delayed shipment, a broken payment link — that is generating contacts.

• Watching size without age

  A backlog of 30 looks fine until you see ten chats are two days old. Show the oldest waiting conversation and the age bands next to the count.

• Counting pending chats as backlog

  Including chats waiting on the customer inflates the number and hides the real queue. Filter on who owes the next message.

• Clearing by closing

  Mass-resolving old chats makes the number drop and the customers return angrier. Reply to old chats first — with a template if the window has closed — and resolve only when the issue is solved.

• Adding staff for a structural rise

  If backlog grows every week, not just during a spike, more agents only postpone the problem. Look at contact reasons and automate or prevent the top ones.

The backlog is the number supervisors can act on today. It tells them whether to pull agents from other work, extend a shift, switch on a bot for common questions, or send a proactive notice to cut incoming volume. On WhatsApp it also has a deadline attached: a conversation waiting more than 24 hours since the customer's last message falls outside the service window, and only a paid template can restart it, so old backlog costs money as well as goodwill.

What is a support backlog?

It is the number of open conversations waiting on the support team at a given moment, together with how long they have been waiting. It shows how much work is queued, not how fast each chat was handled.

What is a healthy support backlog?

One whose oldest waiting chat is well within your SLA and that clears by the end of each day. The right size depends on volume; age and trend matter more than the raw count.

How do you reduce a support backlog quickly?

Add temporary capacity, answer the most common question with automation or a proactive message, and work oldest-first. Then fix the cause, or the backlog returns.

How is backlog different from first response time?

First response time is a speed measure for each conversation after the fact; backlog is the size of the waiting queue right now. A growing backlog usually predicts worse FRT soon.

• Work the backlog oldest-first unless a priority rule says otherwise.
• Alert when any chat is close to 24 hours since the customer's last message.
• Report backlog at fixed times — start of shift, midday, end of day — to see the trend.
• Use agent concurrency limits; overloading agents slows handling and grows the backlog.