Cost per ticket is your fully loaded support cost divided by tickets resolved, and it varies so widely by market, wage level, and channel that any single global benchmark is misleading. The useful move is calculating your own number with a simple framework, then watching what automation does to it.
Why is there no universal benchmark?
Because the denominator is universal but the numerator is local. A ticket is a ticket, but the cost of resolving one is dominated by labour, and fully loaded agent cost varies widely by market. A support hour in Cairo, Manila, London, and San Francisco are four very different numbers. Industry figures you see quoted are usually averages across markets and industries that may share nothing with your store.
What is reasonably stable across ecommerce, per industry reporting:
-
A typical human agent handles roughly 40-60 tickets per day
-
WISMO ("where is my order") commonly makes up 25-50% of ticket volume
-
Ticket types differ wildly in effort: a tracking question takes a minute; a damaged-item dispute can take days of back-and-forth
So rather than borrow a misleading average, compute your own.
How do you calculate your real cost per ticket?
The formula is simple; the discipline is in including everything.
Cost per ticket = (fully loaded support cost per month) / (tickets resolved per month)
Build the numerator honestly:
-
Wages and benefits for everyone doing support, including the founder's hours priced at a realistic rate. Founder time is the most commonly omitted cost in small-store math.
-
Tools: helpdesk, chat software, AI subscriptions, phone lines.
-
Overhead share: management time, training, workspace if applicable.
-
Cost of errors: goodwill refunds and reships caused by slow or wrong answers, if you can estimate them.
Build the denominator carefully:
-
Count resolved conversations, not raw messages (one customer thread with six messages is one ticket)
-
Count across all channels: email, Instagram, WhatsApp, Messenger, web chat. Social DMs are the most undercounted.
Run it for a normal month and your spikiest month. The spike number is the one that predicts pain.
What moves cost per ticket up or down?
| Variable | Pushes cost up | Pushes cost down |
|---|---|---|
| Ticket mix | Disputes, exceptions | WISMO, product FAQs |
| Channel | Phone, long email chains | Chat with live data access |
| First-contact resolution | Repeat contacts per issue | Answered fully the first time |
| Automation share | Everything handled by humans | Routine majority automated |
| Volume shape | Spiky (idle capacity or overtime) | Smooth |
Note the asymmetry hiding in that table: your cheapest tickets (WISMO, FAQs) are also your most frequent and most automatable. That is why automation moves the average so hard; it removes the tickets from the human ledger entirely rather than making each slightly cheaper.
How does automation change the calculation?
You split the ledger in two:
Automated tickets: cost is your AI subscription divided by AI-resolved tickets. Using ReplAi's public pricing as a transparent example: the Grow plan is $79.99/mo for 2,500 replies, and the Pro plan is $299.95/mo for 10,000. A store resolving thousands of conversations on a flat plan gets a per-ticket figure measured in cents, whatever the market's wage level. (Usage maps to credits internally, 100 credits covering roughly 50-150 replies depending on complexity.)
Human tickets: same formula as before, but the numerator now covers only escalations, and the denominator shrinks to the hard 20-30%. Counterintuitively, your human cost per ticket may rise, because only complex tickets remain. That is fine; the blended cost is what matters, and it falls sharply.
Across ReplAi's merchant cohort in Q1 2026, 80% of messages were fully automated; stores doing 1,000+ orders per month see 70-80% handled without humans and free up 4-8 hours of support time daily. Run those proportions through your own numbers and the blended cost per ticket typically drops to a fraction of the all-human figure. For the head-to-head version of this math, see AI vs human support agents.
What does cost per ticket miss?
Treating support purely as a cost centre misses the revenue on the other side of the same conversations. Support chats are sales chats: pre-sale sizing questions, stock checks, COD order requests. Answer them in seconds and they convert; baseline DM close rates around 3% reflect what slow replies do to intent.
In ReplAi's pilot cohort, DM sales rose +75% for the bottom quartile of merchants, +150% at the median, and +300% for the top quartile after automation, with fashion brand Bronze growing DM sales 13x in one month as the top outlier. A proper support ROI model has two lines, cost saved and revenue gained, and for chat-heavy stores the second line is often larger. Our chatbot cost guide covers the spend side in detail.
How does ReplAi fit into your cost model?
ReplAi gives you a clean, flat automated-ticket cost: plans from free (100 replies/mo) through Starter $24.99, Grow $79.99, and Pro $299.95, each including onboarding with a real person, covering Instagram, Messenger, WhatsApp, and WebChat with real Shopify actions (order tracking, returns, COD, catalogue, abandoned carts) at under 14 seconds average response, 24/7. Because pricing is per reply rather than per ticket or per seat, the meter tracks work the AI actually does, making the spreadsheet math honest.
Calculate your current cost per ticket this week, then book a demo or start on the free plan via pricing and recalculate after two weeks of live automation.
Frequently asked questions
What is a "good" cost per ticket for an ecommerce store? Lower than last quarter, on your own measurement. Cross-company comparisons break on wage markets, ticket mix, and channel mix, so external benchmarks mostly mislead. Track your own blended number monthly, split into automated and human-handled, and judge tools by how they move it.
Should I include founder time in support costs? Emphatically yes, and at opportunity cost, not minimum wage. Founder hours spent pasting tracking numbers are hours not spent on product and growth. Many "free" support setups turn out to be the most expensive once founder time is priced honestly.
How does per-ticket helpdesk billing interact with cost per ticket? Directly and unfavourably for small stores: your tooling cost scales with ticket count, so busy months raise both labour and software lines simultaneously. Models are analysed in Gorgias pricing explained; the short version is to know what unit your vendor meters and what a spike month does to it.
Does automation reduce headcount or redeploy it? For most small and mid-size stores, redeploy. The team stops doing the repetitive 70-80% and spends recovered time (4-8 hours daily at 1,000+ orders per month) on escalations, reviews, and proactive work. Stores like Bronze scaled DM sales 13x with zero new hires; the saving is usually the hire you never make.