Support Team Productivity Benchmarks for Ecommerce (2026)

Tickets per agent per day: the core benchmark

The most-used support team productivity benchmark is tickets resolved per agent per day. For ecommerce, the typical range is 40-120, and where you land depends on three things: the channel mix, how simple your tickets are, and the tooling your agents work in. Email-heavy teams cluster at the low end; chat-heavy teams with strong templates push the high end.

That single number is imperfect. It doesn't weight a 20-second WISMO reply against a 25-minute lost-package dispute, and a team can inflate it by closing tickets fast and re-opening them later. But it's still the most practical figure for capacity planning, and it's the one number nearly every support tool reports, so it's the right place to start a benchmark conversation.

Read the rest of this guide as a way to make that number honest. Channel context, the metrics underneath it, and the AI multiplier all change what 'good' means for your store. A team at 60 tickets per agent per day can be overperforming or underperforming — you can't tell from the number alone.

One more framing note before the numbers. Benchmarks are most useful as a sanity check against your own trend, not as a target to hit. A store selling $400 furniture with white-glove returns will never match a $25 apparel brand on raw tickets-per-day, and it shouldn't try. The point of a benchmark is to tell you whether your team sits roughly where comparable operations sit, and to flag when a sudden drop signals a tooling problem rather than a people problem.

Productivity benchmarks by channel

Channel is the biggest single driver of throughput, because each channel has a different concurrency ceiling. A phone agent handles one customer at a time. A chat agent juggles two to four. An email agent works a queue with no live customer waiting, so handle time is the only constraint. That structural difference is why the same person can resolve 35 phone tickets or 110 chat tickets in the same shift.

Live chat earns its higher numbers through concurrency, but the ceiling is real: most agents max out around three simultaneous chats before reply quality and customer wait both slip. Pushing past that to hit a productivity target usually backfires in CSAT. Email throughput, by contrast, is almost entirely a function of average handle time — cut AHT and tickets-per-day rises in near-direct proportion.

The metrics behind tickets-per-agent

Tickets per agent per day is a summary statistic. The metrics that actually move it are average handle time, utilization, occupancy, and first-contact resolution. If you only track the headline number, you'll chase it the wrong way — usually by pressuring agents to close faster, which tanks resolution quality and quietly creates repeat contacts. Track the inputs and the output takes care of itself.

Two of these are easy to confuse. Utilization is the share of paid hours an agent spends on support work at all; occupancy is the share of available work-time spent actively handling contacts versus waiting for them. A team can be highly utilized and poorly occupied — busy all day, but with thin or bursty volume that leaves gaps between tickets. Knowing which one is low tells you whether to fix scheduling or staffing.

First-contact resolution is the quiet multiplier in this group. Every ticket that needs a follow-up is effectively two tickets of work for one customer problem, so a team stuck at 60% FCR is generating 15-20% more total volume than a team at 75% solving the same issues. Push FCR up and tickets-per-agent rises even though nobody is typing faster — you've simply stopped creating the second contact. That's why FCR and repeat contact rate belong on the same dashboard as raw throughput; they explain most of the variance between two teams that look identical on paper.

What actually drives agent productivity

Past channel mix, a short list of factors reliably predicts whether your agents sit at the top or bottom of the benchmark range. None of them require AI, and the first two return the most for the least effort. If your tickets-per-agent number looks low, start here before you blame the team or add headcount.

The reason these matter so much is that the work is bimodal. A large share of ecommerce tickets are near-identical repeats — where's my order, can I return this, when does my refund land — and the rest are genuinely hard. Productivity lives or dies on how fast an agent clears the repetitive majority, because that's the volume. Shaving even ten seconds off a WISMO reply, multiplied across hundreds of them a week, dwarfs anything you'll gain optimizing the rare complex case.

How AI changes team productivity

AI doesn't make a human agent close email faster — it removes the easy tickets from the human queue entirely. That's the key mental shift. An AI agent that resolves WISMO lookups, return requests, refund-status questions, and product queries on its own isn't adding to anyone's tickets-per-day; it's covering a whole second queue in parallel, 24/7, while your team sleeps.

The table below shows what that looks like for a small team. The 'rightsized' row is where many stores land: the same or higher total contacts covered with fewer humans, because AI absorbed the automatable majority. Note that the two-agent AI team in the example covers more total contacts than the three-agent team with no AI — and the remaining humans spend their day on genuinely complex, high-value work instead of pasting tracking links.

There's a quality dividend hiding in that table too. The tickets AI deflects best are also the most repetitive ones — WISMO, refund status, return eligibility, basic product specs — which are exactly the tickets that burn out human agents. Hand those to AI and your people spend their day on the interesting, ambiguous, relationship-building cases. Teams that make this shift often report the productivity gain shows up twice: once in raw coverage, and again in lower agent turnover, which quietly removes the single biggest hidden cost in support — constant rehiring and retraining.

AI as a copilot for the human queue

Deflection handles the tickets that never reach a person. The second productivity gain is on the tickets that do. For the complex, escalated, or judgment-heavy cases that land in the human queue, AI augmentation — drafting replies, surfacing the relevant order and policy, summarizing long threads — typically lifts human throughput by 20-35%. The agent edits and approves instead of researching and writing from scratch.

The two gains stack, and they compound at peak. During BFCM or a viral product moment, AI absorbs the volume spike on the front line while augmentation keeps your humans moving on the overflow. The result is a flatter staffing curve: you don't have to hire and train seasonal agents for a six-week surge you'll unwind in January.

Augmentation also narrows the gap between your newest and most experienced agents. A 30-day hire who gets an accurate draft and the full order context inline can close a ticket nearly as fast as a veteran, because the research and recall that used to separate them is now done by the system. That doesn't eliminate the need for training — judgment on the hard cases still takes time — but it means a new agent contributes real throughput in week one instead of week five.

1. 1Draft generation — AI proposes a complete, on-brand reply the agent can send or edit in seconds.
2. 2Context surfacing — order status, shipping carrier, prior tickets, and the relevant return policy appear inline, no app-switching.
3. 3Thread summarization — a three-line recap of a 40-message escalation so the agent doesn't re-read it.
4. 4Suggested actions — the agent confirms a refund or return that AI pre-filled within your policy caps.
5. 5Real-time QA — tone and accuracy checks flag a reply before it goes out, protecting CSAT under speed pressure.

Team sizing math with AI

Once AI is in the mix, headcount planning stops being 'volume divided by throughput' and becomes 'residual human volume divided by throughput.' That residual is what's left after deflection, and it's usually about half. The framework below consistently shows that mid-size Shopify stores doing 2,000-8,000 monthly contacts can be well-served by 1-3 human agents plus AI, versus the 4-8 the raw volume would otherwise demand.

1. 1Total your monthly contact volume across every channel — chat, email, social DM, and phone.
2. 2Set a target deflection rate; 40-55% is realistic for most ecommerce stores in the first few months.
3. 3Compute residual human volume = total contacts x (1 - deflection rate).
4. 4Divide residual volume by your tickets-per-agent-per-day, then by ~20 working days, for the base headcount.
5. 5Add a 10-15% complexity buffer, because the tickets AI escalates are the harder, slower ones.
6. 6Worked example: 4,000 contacts x 50% deflection = 2,000 human tickets. 2,000 / 20 days = 100/day. At 80 tickets/agent/day that's 1.25 agents — so one full-timer plus part-time help, or two agents with room to grow.

How to measure productivity without gaming it

The fastest way to ruin a productivity program is to reward tickets-closed in isolation. Agents will optimize for it — closing tickets prematurely, splitting one issue into several, or rushing replies that generate a follow-up tomorrow. Every one of those moves the headline number up and the actual customer experience down. Pair throughput with a quality guardrail so the two can't drift apart.

Track productivity as a small basket, not a single stat. Watch tickets-per-agent alongside CSAT, first-contact resolution, and repeat contact rate. If throughput rises while FCR holds and repeat contacts stay flat, that's real improvement. If throughput rises while repeat contacts climb, you've just moved work into next week and called it a win.

Be careful comparing yourself to published benchmarks, including the ones in this guide. Every range here assumes a roughly typical ecommerce ticket mix, and yours probably isn't typical in some way — a subscription brand drowns in billing and cancellation tickets, a furniture store lives on freight and damage claims, a supplement brand fields a wall of pre-sale questions. Normalize for your mix before you decide your team is slow. The most honest benchmark you have is your own number three months ago.

• Pair throughput with quality — tickets/agent only counts if CSAT and FCR hold steady.
• Measure resolved, not touched — a 'reply sent' is not a 'problem solved.'
• Segment by ticket type — a WISMO-heavy day and a returns-heavy day aren't comparable raw numbers.
• Watch repeat contact rate — rising repeats mean throughput gains are borrowed from the future.
• Benchmark against your own trend first, the channel ranges second — your ticket mix is unique.

Productivity mistakes to avoid

Most productivity problems aren't lazy agents — they're structural. These are the patterns that quietly cap a team well below its benchmark, and what to do instead. If your numbers feel stuck, you'll usually find the cause on this list before you find it in anyone's performance review.

• Chasing throughput with speed pressure instead of removing friction — fix the tooling and the templates first.
• Pushing chat concurrency past three — the extra chat costs you CSAT on all of them.
• Leaving order data in a separate tab — the context-switch tax is invisible but constant.
• Treating AI as a replacement for the whole team instead of a parallel queue plus a copilot for the rest.
• Staffing for peak volume year-round instead of letting AI absorb the spike and flattening the curve.
• Comparing your raw tickets-per-agent to a competitor's without normalizing for channel and ticket mix.

Where Bookbag fits

Bookbag is the parallel queue this whole guide describes. It's an AI agent built for ecommerce that resolves WISMO lookups, processes returns and exchanges, issues refunds within your rules, and answers product questions on its own — across website chat, email, WhatsApp, Instagram, and Messenger, 24/7. On the tickets that still reach a human, it drafts the reply and surfaces the order context, so your team works the augmented queue described above instead of starting cold.

It connects natively to Shopify, WooCommerce, and BigCommerce, and most stores are live in under a day: connect the store, import your help docs, drop in the widget snippet. Pricing is flat monthly plans with a message-credit allowance and a spend cap you set — no per-resolution fee, so deflecting more volume doesn't inflate your bill. That matters for a productivity play, because the entire point is to cover more contacts without your costs scaling one-for-one.

If you're weighing options, compare the ecommerce-native, action-taking approach against general chatbot builders before you commit.