Three months ago you answered every Amazon message from your phone between packing orders. Now there are two of you, eBay and Shopify sales are live too, and messages are starting to slip through the cracks. That’s the exact moment a customer service workflow stops being optional and starts being the thing standing between you and a wave of late replies, bad feedback, and a burnt-out first hire.
This guide builds one from nothing, no existing tool, no existing process, just the six decisions that need to happen in order.
TL;DR
- Map every channel you actually sell on before picking a tool, most new sellers only need to track two or three.
- Set an SLA per channel using marketplace minimums as the floor, not the target.
- Decide who answers and when before automating anything, staffing gaps break workflows faster than bad tools.
- Pick a helpdesk built for your channel mix, then set routing rules once, not per ticket.
- Turn AI on for WISMO and tracking questions from day one, you don’t need a year of data to justify it.
- Start with a two-tier escalation matrix and a simple CSAT survey, refine both once you have real data.
Step 1: Map your channels
Before you touch any tool, list every place a customer can currently message you. For most growing sellers that’s Amazon Buyer-Seller Messaging, eBay’s messaging system, a Shopify contact form or email, and maybe a social DM inbox. Write it down, it’s usually shorter than it feels.
This list is what everything else in this guide gets built around. A workflow designed for five channels when you actually run two is wasted setup time, add channels as you add them, not preemptively.
Step 2: Define your SLA per channel
Amazon expects a response to 90% of buyer messages within 24 hours, that’s not a suggestion, it’s tied to your account performance rating. eBay penalizes slow replies through Top Rated status. Neither of those numbers is your target, they’re your floor. Set your own internal SLA a few hours tighter than the marketplace minimum on each channel, so a bad day doesn’t put you in breach.
Write the SLA down per channel, not as one blended number. A single combined target hides the channel that’s quietly falling behind while the others carry the average.
Step 3: Decide who answers, and when
This is the step most workflow guides skip, because they assume you already have a team. You probably don’t yet, or you have one person wearing five hats. Before picking any software, answer two questions: who owns support today, founder, a first hire, a contractor, and what hours are you actually covering, business hours only, extended, or something closer to always-on.
Get this wrong and no tool fixes it. A perfectly automated routing system still fails if nobody’s watching the inbox after 6pm and half your tickets come in overnight. Match your coverage hours to when your customers actually message you, check timestamps on your last month of tickets if you’re not sure.
Step 4: Pick a helpdesk and set up routing rules
Once you know your channels and coverage, choosing a tool gets simple: does it connect natively to the channels you listed in Step 1, and can it be running inside a week, not a month. A helpdesk built for marketplace sellers pulls order data in automatically, which matters more early on than any advanced feature you won’t use for months.
Set routing rules once you’re connected: which ticket types go to which person, what triggers an alert, what auto-tags by channel or keyword. Keep the first version simple, three or four rules covering your highest-volume ticket types. You can add nuance once you see where the real bottlenecks are, not before.
Step 5: Turn on AI for WISMO and tracking questions
Most new workflow guides tell you to wait on AI until your process matures. Skip that advice. WISMO and tracking questions are the easiest tickets to automate and usually the highest-volume ones from day one, you don’t need a year of ticket history to know a customer asking “where’s my order” needs their tracking number, not a person. Our guide to automating ecommerce customer support covers the full setup if you want to go deeper here.
Starting with AI on the simplest ticket type early means your first hire spends their time on the tickets that actually need a person, not repeating the same tracking-link reply forty times a day.
Step 6: Build a first-draft escalation matrix
You don’t need a five-tier escalation system on day one. Start with two: routine tickets your first hire or AI handles directly, and anything that needs the founder, a formal claim, a threat to leave negative feedback, an order over a set dollar value, an angry message that doesn’t fit a template.
Write down the trigger for tier two explicitly. “When it feels urgent” isn’t a rule anyone can act on consistently. “Any A-to-z Guarantee claim, any order over $200, any message containing the word ‘lawyer’ or ‘refund demand'” is. Expand to more tiers once you see patterns in what’s actually landing in tier two.
Step 7: Set up a first CSAT baseline
You can’t improve what you’ve never measured. Turn on a one-question CSAT survey (thumbs up/down or a 1-5 rating) sent immediately after each ticket closes, and let it run for a month before touching anything. That first month gives you a real baseline instead of a guess, and tells you which ticket categories need attention before you’ve spent a dollar fixing the wrong thing.
Key Takeaways
- Map your actual channels first, most new sellers need fewer than they think.
- Set SLA per channel using marketplace minimums as the floor, not the target.
- Decide staffing and coverage hours before picking a tool, this is the step most guides skip.
- Turn on AI for WISMO and tracking from day one, don’t wait for a mature process.
- Start with a simple two-tier escalation matrix and a one-question CSAT survey, refine both with real data.
Your action plan:
- List every channel customers can currently message you on.
- Set an internal SLA per channel, a few hours tighter than the marketplace minimum.
- Decide who owns support and what hours you’re covering, based on when tickets actually arrive.
- Pick a helpdesk that connects to your channels natively, and set 3-4 basic routing rules.
- Turn on AI for WISMO and tracking tickets, and launch a one-question CSAT survey.
FAQ
What is a customer service workflow?
A customer service workflow is the defined path a ticket follows from arrival to resolution: which channel it came from, who or what handles it, how it’s prioritized, and what happens if it can’t be resolved immediately. Building one from scratch means making those decisions deliberately instead of letting them happen ad hoc.
How do I set up customer service for my first online store?
Start by mapping every channel customers can reach you on, then set an SLA per channel using marketplace minimums as your floor. Decide who’s answering and when before choosing a tool, then connect a helpdesk built for your channel mix and set a handful of basic routing rules to start.
What is the right SLA for Amazon seller customer service?
Amazon requires a response to at least 90% of buyer messages within 24 hours, and falling below 95% compliance over a rolling 30 days affects your account performance rating. Set your own internal target a few hours tighter than that, so a busy day doesn’t put you in breach of Amazon’s own minimum.
Once this workflow is running, the next question is usually how to make it faster. Our guide to streamlining customer support workflows picks up from here, centralizing further, tightening automation, and layering in deeper analytics once you have real ticket volume to optimize against.
Want help mapping your specific channel mix to a workflow that fits? Book a Free Demo and we’ll walk through it with you.
