Ecommerce Order Support Staffing Statistics for 2026

Order support starts after a shopper clicks buy. It includes payment and address questions, order changes, tracking, delivery exceptions, cancellations, returns, refunds, and marketplace messages. The work does not arrive evenly, and the amount attached to an order varies sharply by product category.

The ecommerce order support staffing statistics below come from merchant-platform data, retail surveys, company transaction data, a marketplace service rule, and federal labor data. They do not describe one common population. Each section states the geography, sample, reference period, and metric definition where the publisher provides them.

For operating guidance, see ecommerce customer service outsourcing, ecommerce virtual assistant services, and the ecommerce industry page.

Order support benchmarks at a glance

These figures should not be averaged together. Gorgias reports activity from stores using its platform. Shopify reports merchant sales rather than support contacts. NRF and Happy Returns report survey estimates. The differences make the sources complementary, not interchangeable.

Order status is a measurable share of the queue

Gorgias says the question "Where is my order?", often shortened to WISMO, accounts for 18% of incoming requests on average. Another Gorgias page says shipping-status requests can account for up to 30% of incoming tickets.

Those figures are not a contradiction. One is an average for a named WISMO intent, while the other is an upper bound for the broader shipping-status category. A team should not replace them with a midpoint. It should tag its own order-status contacts and report the count as a share of human-handled tickets, automated resolutions, and all support interactions.

The distinction affects staffing. A tracking lookup that a shopper completes through self-service creates no agent work. A delayed, lost, split, or incorrectly marked delivery can require carrier research, an order-management update, and a replacement or refund decision. Counting both as one "order-status ticket" hides the difference in handling time.

Tickets per order vary by product category

Gorgias Ecom Lab reports median tickets per 100 orders for ecommerce brands with about $10 million in annual GMV. Its March 2026 table ranges from 19 tickets per 100 orders for toys and games to 46 for electronics and vehicles and parts. Food and beverages recorded 20 per 100 orders, while apparel and accessories recorded 22 per 100.

Applied as labeled planning calculations, those medians produce very different queues:

The calculations multiply orders by the relevant Gorgias March 2026 median. They are examples, not reported headcounts. A store should substitute its own category, contact policy, channel mix, automation rules, and repeat-contact rate.

Gorgias also reports that the median ecommerce brand self-serves 45% of AI-touched tickets without human involvement, while the top quartile reaches 65%. That denominator is AI-touched tickets, not every contact or order. It should not be applied to the full forecast unless the store knows what share of its queue enters the same automation flow.

Response time changes the coverage requirement

In March 2026 Gorgias reported a 6.3-hour median first response at the $10 million GMV band across ecommerce categories. Category medians ranged from 1.6 to 9.1 hours, a spread that makes one cross-category target unreliable.

The separate Gorgias retail benchmark reports a 4-hour median first response and 14.1-hour median resolution. Its email results are slower at 7.9 hours for first response and 17.1 hours for resolution. These are platform medians, not promised service levels.

Marketplace work can carry an external clock. Amazon tells sellers to respond to buyer messages within 24 hours, including weekends and holidays. Amazon also says an automated acknowledgement does not count as a response. Stores selling through a marketplace therefore need coverage that protects the platform threshold even when their direct-store email queue uses business hours.

Track at least three clocks: first substantive reply, next promised update, and final resolution. Staffing only to the median can still leave an old tail of delivery investigations or refund cases. Interval forecasts should include the 90th percentile and the oldest open order case.

Seasonal volume creates two support peaks

Shopify merchants generated $14.6 billion in global sales during the 2025 Black Friday Cyber Monday weekend, up 27% from 2024. More than 81 million customers bought from Shopify-powered businesses, and sales peaked at $5.1 million per minute.

These are Shopify-wide transaction figures, not the growth rate of an individual store and not a ticket forecast. They show why a monthly support average can fail during a compressed selling event. The support peak can also lag the order peak as shoppers wait for dispatch, watch tracking, report delivery problems, and request returns.

The 2025 NRF and Happy Returns study estimated that 17% of holiday sales would be returned. Among the surveyed large U.S. merchants, 43% planned to hire seasonal returns staff and 49% planned more focus on third-party logistics partners. Peak planning should therefore cover both the purchase period and the returns period that follows it.

Returns and refunds create their own workload

NRF and Happy Returns estimated that 19.3% of online sales would be returned in 2025, compared with 15.8% of all retail sales. The research included 2,006 U.S. consumers who had returned an online purchase in the prior year and 358 ecommerce professionals at U.S. merchants with more than $500 million in revenue.

The large-merchant sample limits direct comparison with a small store, but it identifies real work. Support staff explain eligibility, create labels or authorizations, check receipt status, communicate refund timing, and escalate exceptions. The warehouse or logistics provider inspects and dispositions the item. Finance reconciles the refund. Fraud staff may review the claim because the same NRF study estimated that 9% of returns were fraudulent.

Returns volume alone is not enough for a support forecast. Measure contacts per return, agent minutes per contact, repeat-contact rate while a refund is pending, and the share that needs approval. Keep physical inspection time in a separate warehouse model unless the same team performs it.

A workload-based staffing model

No cited source establishes one universal number of order-support agents per thousand orders. The evidence supports a workload model instead:

forecast tickets = forecast orders × observed tickets per order

agent workload hours = forecast tickets × observed active minutes per ticket ÷ 60

required scheduled hours = agent workload hours ÷ productive share of scheduled time

Run the calculation by channel and interval. Email, live chat, social messages, marketplace messages, and phone calls have different concurrency and response clocks. Add separate workload lines for return intake, carrier follow-up, refund approval coordination, quality review, coaching, training, meetings, breaks, and absence.

Gorgias has published an older platform benchmark of one ticket handled per agent hour on average. Treat that as an external comparison, not a current productivity promise. Ticket complexity, automation, channel, and what the platform counts as handled can all change the rate.

Labor data helps with cost and ramp planning. The U.S. Bureau of Labor Statistics reported a national median wage of $21.53 per hour for customer service representatives in May 2025. It also says short-term on-the-job training commonly lasts two to four weeks. The wage covers customer service across industries, not ecommerce order support alone, and the training range is not a guarantee of full proficiency in a store's systems.

What to measure before adding capacity

A useful order-support dashboard connects demand, service, and staffing:

Use at least a normal week, a promotion week, the holiday order peak, and the post-holiday return peak as separate scenarios. A flexible ecommerce virtual assistant can handle documented tracking, return intake, and follow-up tasks. The retailer should retain clear approval limits for replacements, high-value refunds, fraud cases, and policy exceptions.

FAQ

What percentage of ecommerce support contacts are order-status questions?

Gorgias reports that WISMO questions account for 18% of incoming requests on average. A broader Gorgias shipping-status figure reaches up to 30%. The definitions differ, so stores should measure their own tagged share rather than merge the figures.

How many support tickets should an ecommerce store expect per order?

Gorgias Ecom Lab reported 19 to 46 tickets per 100 orders across its published vertical medians at the $10 million GMV band in March 2026. Category, contact policy, automation, and channel mix can move a store outside that range.

How many order-support agents does an ecommerce business need?

Calculate forecast contacts by channel, multiply them by observed handling time, and divide by productive capacity. Then add interval coverage and shrinkage. A fixed agents-per-order ratio ignores category, response promises, returns, marketplace rules, and seasonal peaks.

When should seasonal order-support staffing start?

Coverage should be ready before the order surge and remain through the return and refund wave. Shopify's 2025 transaction data shows concentrated BFCM demand, while the NRF study says merchants expected 17% of holiday sales to be returned.

Sources

The staffing decision should follow the store's measured workload. External benchmarks can test whether the forecast is plausible, but the final schedule has to reflect the store's orders, contact reasons, return rate, response promises, and productive capacity.