Prioritize customer service requests by the cost of waiting, not by who the customer is. Put deadline-bound problems, changes to unfulfilled orders, and money in motion first, everything due today second, and the rest in order received. VIP status breaks ties inside a band. It should never jump a band.
A priority system that ranks customers will always look fair on paper. It just answers the loyal customer asking about a restock before the first time buyer whose wrong address ships in an hour.
A customer emails at 9:05 on a Tuesday morning: the shipping address on the order she placed last night is her old apartment. Your fulfillment cutoff is noon. In the same queue sit a repeat customer asking when a color comes back in stock, a wholesale inquiry, a return label request, and four “where is my order” messages. If your helpdesk sorts by loyalty tier, the address change waits behind the repeat customer. If it sorts by arrival time, it waits behind everything that came in overnight. Either way, by the time someone reads it, the box is on a truck, and a two minute fix has become a reship, a refund, and a customer who will not order again.
That is the gap this guide closes. Most priority setups I see on Shopify stores were built from a helpdesk vendor’s best practices list: VIPs first, repeat customers next, social channels bumped up, review threats escalated. Each rule sounds reasonable. Together they rank people, when the thing that should be ranked is how fast a ticket gets more expensive. I call that the cost of waiting, and once you sort by it, most of the rules you have today either become tiebreakers or come out entirely.
If you are a two person team answering everything yourselves, the first two sections are the whole system. If you run five or more agents, spend your time on the stage section and the weekly checks.
Most ticket priority systems rank the wrong thing because they sort by the customer’s value to the business, when the only question that matters at the moment of triage is how much the problem costs for every hour it waits. A loyal customer asking about a restock loses nothing if the answer comes this afternoon. A new customer with a wrong address on an order that ships at noon loses the order.
Value-based rules creep in for an understandable reason. Repeat customers really are worth more, and the helpdesk makes it easy to tag them. The trouble is what the tag does in a busy queue. A VIP rule moves every message from that customer up, including the ones that could wait a day, and it pushes every message from everyone else down, including the ones that cannot. On a launch day or during a carrier delay, that leaves low urgency VIP questions sitting above urgent problems from people who have not yet earned a tag.
The review threat rule is the clearest example of a rule that ranks the wrong thing. Escalating any message that mentions a bad review teaches the queue that threatening is the fastest way to be heard. The customer who writes politely about a broken item waits longer than the customer who writes the same thing with a one star warning attached.
Cost of waiting fixes this by asking about the problem instead of the person. Some problems are cheap to solve now and expensive later: an address change before the label prints, a size swap before the pick, a cancellation before the charge settles into a shipment. Some are tied to a date the customer cannot move: a gift for Saturday, an event, a subscription renewal tomorrow. Those come first no matter who sent them. Everything else can be answered in a sensible order, and that is where customer value earns a place, as a way to choose between two tickets that are equally urgent.
Three questions set a ticket’s priority: does the customer face a deadline they cannot move, can the order still be changed because it has not been fulfilled, and is money or a public complaint moving right now? A yes to any of them puts the ticket in the first band. An agent can answer all three from the first lines of a message and the order sidebar.
Use the answers to sort every ticket into one of three bands. The first band, Now, is anything with a yes: a delivery date the customer has told you about, a change request on an unfulfilled order, a duplicate charge, a chargeback warning, a public post about a safety problem. The second band, Today, is anything that needs a human decision but does not get worse by the hour: a return outside the window, a damaged item with photos, a wholesale inquiry, a question blocking a purchase. The third band, Queue, is everything else, answered oldest first: restock questions, general product questions, feedback, partnership pitches.
The second question deserves the most attention, because it is the one most priority systems miss. Shopify lets you edit an order after it is placed and before it is fulfilled, adding or removing items, changing quantities, and refunding or invoicing the difference. Once the order is fulfilled, the same request becomes a return, a reship, or a refund. That window is invisible unless someone checks fulfillment status during triage. If your warehouse or 3PL picks in the morning, the Now band has to be cleared before the pick, not by end of day.
Three bands is deliberate. Helpdesks offer more levels, and teams use them, but every extra level is another judgment an agent has to make on every ticket. With three bands the triage call takes seconds, the Now band stays small enough to clear first thing, and disagreements are easy to settle: if nobody can name what gets more expensive by waiting, it is not Now.
When two tickets land in the same band, break the tie. Oldest first is the default. Customer value is a reasonable second tiebreaker, and so is channel, if a live chat customer is still on the page. What a tiebreaker never does is lift a ticket into a higher band. The one exception is not really an exception: a high value subscriber who writes that they are about to cancel is money moving now, and that ticket is Now on its own merits.
Philippe Lehoux learned to triage by deadline running a business where the deadline was the product. Before he co-founded Missive, the shared inbox and team communication platform, he ran an online store selling printed name badges for conferences, and he told me about it on episode 406 of the eCommerce Fastlane podcast. As a Canadian company shipping into the United States, packages sometimes got held at customs, and as he put it, “the conference still going to be at the same time even if it’s stuck at the border. So we had to deal with those issue urgently.”
That is the cost of waiting rule in its plainest form. A stuck badge order and a question about badge sizes might come from the same customer on the same day, and only one stops being solvable on a fixed date. In cost of waiting terms, the stuck package had to surface first, whoever sent it.
His description of how a team grades tickets is about difficulty, which decides who answers rather than when, and it sits neatly underneath the bands. Done by a person, that grading means deciding that “those one would be handled by that team because they’re easy, those ones are harder or those one are extremely hard and need to be dealt just by senior people.” He was also candid about the cost of doing it by hand: it is a job “which takes a lot of time, usually in the morning because you don’t want to miss the really important tickets.” That morning sort is exactly the step most small teams skip, and it is why the address change in the opening example gets missed.
Missive now lets teams put a prompt inside a rule so AI does that first pass, and Philippe gave stuck packages as an example of what such a rule could surface instantly. But he drew a clear line on replies. On the long tail of customer questions, he said, AI “often creates a lot of mistakes,” and Missive’s approach is that “right now it’s more about augmentation than automation.” The AI can draft; an employee reviews and presses send. His view of letting it reply on its own: “I believe it’s a technology people can hang themselves with if they go the automation route.” He added that the space is moving fast enough for that to change, so treat it as one founder’s position about his own product’s limits, stated in mid 2025. It is still the right default for triage: let software sort and draft, and keep a person on the call for anything in the Now band.
VIP tags, channel rules, and review threat rules belong below the bands, as tiebreakers or routing, never as reasons to move a ticket up. Once the three questions decide the band, each of these older rules either finds a narrower job or gets deleted.
VIP and repeat customer tags make a good tiebreaker inside a band and a good routing rule if you have a senior agent who knows your best customers. Keep them for that.
Channel rules are trickier, because channel speed is real: a shopper in live chat may still be on the product page, while an email can wait an hour. The problem is what customers do when they wait. I admitted my own version of it to Philippe on that same episode: “you send an email and you don’t get an answer back in like you know, in a half an hour. Next thing you know, you’re sending a DM through to see if the social media person’s gonna pick it up.” A rule that answers DMs faster than email trains every customer to send both, and now one problem is two tickets, answered by two people, sometimes with two different answers. If you want channel rules, pair them with merging. Gorgias auto-merge, for example, combines a customer’s open tickets from eligible channels into one thread inside a window you set, and the customer support platforms worth comparing are the ones that tie conversations to the person rather than the channel.
Review threat rules should go. Route any public complaint about safety or a defect into Now because of what it is, a problem that is spreading, not because the customer used the word review. Everything else that mentions a review is handled in whatever band its actual problem belongs to.
The right setup depends on how many people answer customers: one person runs the bands from a morning sort and a few tags, two to four people need three helpdesk rules and one view, and five or more need AI doing the first sort with a person reviewing it. The bands do not change across stages. The machinery around them does.
If you are the founder answering everything, use what you have. Shopify Inbox lets you search conversations by order ID, open the customer’s profile from the conversation to see their orders, assign conversations, and it applies topic labels such as Shipping or Checkout automatically. That is enough. Sort once each morning before your pick, answer anything touching an unfulfilled order first, and resist buying a helpdesk to fix what is really the order you read messages in.
Once two to four people share the inbox, write the bands into the tool so nobody has to remember them. In Gorgias, as of October 2026, tickets carry a priority field with four levels (Low, Normal, High and Critical), rules can set it automatically, views can filter on it, and auto-assignment hands out tickets from Critical down with ties going to the oldest. Map Now to Critical, Today to High, and Queue to Normal, and leave Low for filtered noise. Most helpdesks have an equivalent field or tag. The three rules to write first are the three questions: unfulfilled order plus a change keyword, a stated date, and a billing or chargeback keyword. Then build one view that shows only the Now band, and make clearing it the first job of every shift.
At five or more agents, keyword rules start to miss things, and AI triage earns its place. Missive’s AI rules read a message and can apply labels, add a summary note, or create a reply draft for review, and other helpdesks offer similar intent detection. Use it to sort and draft, with the bands written into the prompt. Keep a person reviewing anything the AI places in Now, and check a sample of what it placed in Queue each week, because the mistakes that matter are the urgent tickets it filed as routine. For the predictable questions that never needed a human in the first place, the work is upstream: automate the predictable order and sizing questions so they never reach the bands at all.
For a typical two agent store heading into peak season, the reset is subtraction first: collapse the priority tags into three bands, add one rule for unfulfilled order changes, merge duplicates, and remove the VIP and review threat escalations. Here is how I would work through it.
This is a labelled composite, built from the patterns I hear from Shopify merchants at this size, not a single real store, and it carries no observed metrics. Picture a home and apparel brand doing about $1.5M a year. Two agents share a helpdesk connected to Shopify. Over three years they have added nine priority tags: VIP, repeat customer, Instagram, review threat, pre-sale, wholesale, influencer, urgent, and order issue. The warehouse picks at 11am. The agents open the helpdesk and start at the top.
The first thing I would look at is what happened to change requests on unfulfilled orders last month, because that is the most expensive ticket this store can miss. If address and size changes regularly arrived before 11am and were answered after the pick, that is the problem to fix before anything else. So the first change is one rule: any message from a customer with an unfulfilled order that mentions an address, size, swap, or cancel goes to Now, and Now is cleared before 11am.
Next, collapse the nine tags. Urgent, order issue, and pre-sale questions that are blocking a purchase today become the three questions. VIP and repeat customer stay, demoted to tiebreakers. Wholesale and influencer become routing to whoever owns those relationships, at Today priority. Instagram stops being a priority and becomes a merge rule on whichever channels the helpdesk can merge, so a DM and an email from the same customer end up in one thread. Review threat is deleted.
Then the subtraction that matters most before peak: do not add an AI agent to answer customers in the next six weeks. A store this size gets more from clearing the Now band every morning than from a new tool that has not seen a peak season with this team. If they want AI before the rush, use it to draft replies for the agents, and revisit after January with real ticket data. The pre-peak support audit is the place to decide which ticket types should go to automation once the bands are working. At $5M with five agents, the call changes: AI triage into the bands becomes worth testing, because no person can sort that queue by hand every morning.
Your priority system is working when the Now band is small, cleared before the deadline it exists to protect, and when change requests on unfulfilled orders stop turning into returns. Three weekly checks tell you whether that is true, and none of them needs a reporting tool beyond your helpdesk and Shopify admin.
First, the size of the Now band. If it holds a large share of all tickets, the questions are being answered too generously and the band has stopped meaning anything. My working rule, which is a starting point rather than a benchmark, is that if more than about one ticket in five lands in Now, tighten the rules before adding people. Read ten Now tickets from the week and ask of each: what got more expensive by waiting? Any ticket where the honest answer is nothing belongs in Today.
Second, missed windows. Pull the orders from the week where the customer asked for a change and the order shipped anyway. Each one is a Now ticket the system failed to surface in time. Drive this number toward zero; it says more about your triage than any average response time. If the misses cluster around a single cause, such as a carrier delay week, look upstream: cutting where is my order tickets with branded tracking shrinks the Queue band so the urgent tickets are easier to see.
Third, duplicates and repeat contacts. Count conversations where the same customer wrote twice about the same problem, on any channel. A high count means the first reply was too slow or did not finish the job. Fast answers are not the same as finished ones, and the support problem Shopify brands notice too late is exactly that gap. Review the three checks for fifteen minutes each week, change one rule at a time, and leave the bands alone long enough to see whether the change worked.
A small Shopify store should prioritize customer service tickets by the cost of waiting, putting deadline-bound problems, changes to unfulfilled orders, and billing problems first, today’s decisions second, and everything else in the order it arrived. If one or two people answer every message, you do not need helpdesk rules to do this. Do a single sort each morning before your warehouse picks orders, answer anything that touches an unfulfilled order before the pick, then work oldest first. Shopify Inbox lets you search by order ID and open each customer’s orders from the conversation, which is enough to answer the three triage questions. Add a helpdesk and written rules once more than one person is triaging, so the sort no longer depends on who opens the inbox first.
VIP customers should not always get answered first; their status should break ties between equally urgent tickets, not move routine questions ahead of urgent ones. A loyal customer asking about a restock loses nothing if the answer arrives this afternoon, while a new customer with a wrong address on an order that ships at noon loses the order. If your helpdesk lets a VIP tag raise every message, the queue fills with low urgency questions sitting above problems that get more expensive by the hour. Keep VIP tags for tiebreaking inside a band and for routing to an agent who knows your best customers. Remove any rule that lifts a VIP ticket into a higher urgency band on status alone.
The most important ticket to answer first is usually a change request on an order that has not been fulfilled yet, such as a wrong address, a size swap, or a cancellation. Shopify lets you edit items and quantities, or update the customer’s details, on an order that has not been fulfilled, so the fix takes a couple of minutes while the window is open. Once the order is picked and shipped, the same request becomes a return, a reship, or a refund, with shipping costs and an unhappy customer attached. That is why the Now band has to be cleared before your warehouse or 3PL picks orders, not at the end of the day. Deadline-bound problems and billing issues such as duplicate charges sit in the same first band.
AI can automatically sort customer service tickets into priority bands, but a person should still review anything it marks urgent and spot check what it marks routine. Tools such as Missive’s AI rules can read a message, apply a label, add a summary note, or draft a reply for review, and other helpdesks offer similar intent detection. That first sort saves the slow morning triage that small teams often skip. The risk is the urgent ticket the AI files as routine, which is why the review matters. Missive co-founder Philippe Lehoux described his company’s stance on the eCommerce Fastlane podcast as more about augmentation than automation. For most teams under five agents, keyword rules for the three triage questions do the job without AI.
A customer support team should use three priority levels in practice: one for problems that get more expensive by the hour, one for decisions due today, and one for everything else. Helpdesks often offer more. Gorgias, for example, has four levels, and you can reserve the lowest for filtered noise such as auto-replies. Every extra level adds a judgment call an agent has to make on every ticket, and under pressure those calls drift until most tickets sit near the top. Three bands keep triage fast and disagreements easy to settle: if nobody can name what gets worse by waiting, the ticket is not urgent. If more than about one ticket in five lands in the top band, tighten your rules.
