Most help desk software requirements docs ship missing the phone. That's the bug in this whole exercise.

You sit down to write the requirements doc. You list ticketing, channels, automation, reporting. You forget that during Black Friday week, 50 to 80% of your tickets are "where is my order" and a chunk of those come in by phone, not email. You shortlist three vendors. None of them actually run the phone. Six months later you're stitching a second tool onto the stack.

This guide is the doc you wish you'd written the first time. Sixteen requirements, organized into four buckets you can paste into Notion or a Google Doc today: five functional, four integration, three reporting, and four "most teams miss." It's built for Shopify and DTC brands evaluating tools like Gorgias, Zendesk, Help Scout, Richpanel, Freshdesk, and Reamaze, not for IT directors picking a service desk.

Hear what AI support calls sound like for your store. Just paste your Shopify URL and get sample calls in under 20 seconds, no email required. Listen to demo calls for my store.

Why most help desk software requirements docs ship broken

Two problems show up in every requirements doc I've reviewed.

The first is the feature matrix trap. You build a 40-row spreadsheet, send it to four vendors, and they all check every box. The matrix doesn't tell you how deep each box goes. A vendor that "supports voice" might mean a $0.40-per-minute call routing add-on with no AI. Another might mean a fully resolved AI call. Same checkbox. Different products.

The second is the channel-blindness problem. Most help desks were built to handle tickets that arrive as text. Email, chat, social DMs, web forms. Then somebody slapped a voice add-on on the side. WISMO ("where is my order") inquiries account for 20 to 40% of total ecommerce customer service volume in normal months, and climb to 50% or higher during peak shipping weeks, according to Salesforce's WISMO benchmark data. A meaningful share of those come in by phone, especially after-hours when chat is down.

If your requirements doc treats the phone as one of N channels, you're going to underbuy the channel where your angriest, highest-intent customers go.

Here's the frame this guide uses. Sixteen requirements, four buckets:

• Functional (5): the workflow your agents live in
• Integration (4): the connections to your stack (Shopify, shipping carrier, voice, helpdesk handoff)
• Reporting (3): how you measure the loop
• Most teams miss (4): the gaps SERP listicles skip

Each one has a "score it on" line so you can rate vendors 1 to 5 and total up.

5 functional help desk software requirements

These are the baseline. Every help desk on a 2026 shortlist will check most of them. The depth is what separates the winners.

Req 1. Ticket management

The job: ingest a contact, route it to the right person, give the agent every piece of context they need to resolve it, and close it out.

The depth questions:

• Sort, filter, tag, priority, merge, snooze: every helpdesk has these. Do they actually work in bulk?
• Conversation threading across channels: if a customer emails you Monday and DMs you Wednesday, does that show up as one thread or two?
• Agent view of full history: order history, past tickets, returns, lifetime value. All in the sidebar, no extra clicks.

Score it on: tag depth + bulk actions + cross-channel conversation merging.

Req 2. True omnichannel inbox

Most vendors claim omnichannel. Then you read the fine print and "voice" is an add-on, "social" is one network not all, and "SMS" is a forwarding setup.

What you actually need:

• Email and live chat: native, every helpdesk has this
• Social DMs (Instagram, Facebook, TikTok, WhatsApp): native, not via a third-party connector
• SMS: native two-way, not just forwarding
• Voice: this is the one most lists treat as a side dish. It's a main course for ecommerce.

Score it on: which channels are native vs add-on, and which "native" channels actually have full feature parity.

Req 3. Automation and workflow rules

The win here is removing the boring 30% of tickets so agents focus on the chewy 70%.

What strong automation looks like:

• Rule-based routing: by tag, language, order value, return reason, refund amount
• Macros and canned responses: with merge fields pulling order data
• AI-assisted draft: the AI writes a draft, the agent edits and sends
• Auto-tagging: AI reads the message and tags by intent (WISMO, return, exchange, product question)

Score it on: how many conditions you can chain into one rule, and whether AI tagging actually works on your traffic.

Req 4. SLA management

This is where most ops teams admit their helpdesk is just a glorified inbox. They have no SLAs running.

Real SLA management:

• First Response Time (FRT) targets per channel: under 1 hour for email, under 40 seconds for chat (the 2026 benchmarks from Helpable and Hiver, customer satisfaction peaks at 84.7% when chat FRT is 5-10 seconds)
• Breach alerts: agents and managers get pinged when an SLA is about to miss
• Per-tier SLAs: VIP customers get a different clock
• Channel-specific SLAs: voice gets its own targets (hold time, abandon rate)

Score it on: per-channel SLA control + breach alerts + VIP tiering.

Req 5. Knowledge base and self-service

The cheapest ticket is the one the customer answers themselves.

What good looks like:

• Public help center: branded, SEO-friendly, accessible without login
• Internal agent KB: macros and internal docs agents can search while on a ticket
• AI-suggested answers from KB: when an agent opens a ticket, the helpdesk surfaces relevant KB articles
• Self-service measured: deflection rate tracked, not just "we have a KB"

Score it on: deflection rate measured as a KPI, not just article count.

If you're on Shopify, Ringly.io handles the voice slice of your support stack so your helpdesk's KB does its actual job (deflection through chat and self-serve). Try it free for 14 days.

4 integration help desk software requirements

This is where vendor demos get slippery. Everyone says "we integrate with Shopify." What they mean varies wildly.

Req 6. Native Shopify integration (not an iframe)

Surface depth: a sidebar widget that shows order number and customer name. You can do better.

Real Shopify depth:

• Look up any order from inside the ticket: search by name, email, order number, phone
• Edit, refund, cancel, partial-refund: take action without leaving the helpdesk
• Subscription edit: pause, skip, swap, cancel from inside the ticket
• Tags and metafields: helpdesk reads them, you can write back to them
• Bi-directional sync: not just pull, also push (notes from helpdesk land back on the order)

Score it on: actions you can take inside the ticket without opening Shopify admin.

For more depth, Gorgias has the deepest bi-directional Shopify sync of any helpdesk on the market right now. Zendesk and Help Scout are catching up, but they're not there yet.

Req 7. Shipping carrier event data inside the ticket

Most helpdesks show a tracking link. That's a URL. It's not data.

What you actually want:

• Real-time carrier scan events: last scan location, last scan timestamp
• Exception codes: USPS "delivery exception," UPS "address correction needed," FedEx "weather delay"
• Delay reason in plain English: "the package is delayed at the regional hub" beats "see tracking URL"
• Automatic ticket tagging when a shipment is late or stuck

This is the difference between an agent saying "let me check the tracking" (2 minutes of dead air) and saying "your package is in Memphis, USPS had a weather delay, ETA is now Thursday" (5 seconds).

Score it on: real-time carrier event data attached to the ticket, vs a static tracking URL.

Req 8. Phone / voice integration that does more than route calls

A voice add-on that just routes calls and logs them is not a voice strategy. It's a phone bill with a UI.

Real voice integration:

• Inbound AI answer 24/7: the AI picks up, identifies the caller, pulls their order
• Resolution on the call: order status, return start, exchange, simple product questions answered without escalation
• Clean handoff when needed: AI passes to a live agent with full transcript and context

Most helpdesk voice add-ons fail this test. Gorgias voice, for example, is "basic, no AI for phone calls, can't answer from mobile app, geographic dialing restrictions," per Capterra reviews where Gorgias scores 4.7/5 across 600+ reviews. The voice piece is the consistent complaint.

Score it on: AI voice resolution rate + per-call cost.

Req 9. Helpdesk handoff path (when AI passes to a human)

If your phone is run by AI (Ringly, an in-house build, whatever), and your text channels are run by your helpdesk (Gorgias, Zendesk, whichever), you need a clean handoff.

What clean handoff looks like:

• Voice to text channel: AI takes the call, customer asks for a human, voice transfers to a live agent with the transcript and order context attached. If a human isn't available, the AI offers to send an email follow-up that lands as a new ticket in the helpdesk inbox.
• Email-in handoff: AI emails a summary into the helpdesk inbox, the ticket opens with full context, an agent picks it up
• Call transfer: AI patches to a live agent's phone number with a brief "this is what the customer needs" handoff
• Context preserved: transcript, customer ID, order number, intent

Note: this is not a "native integration" with Gorgias / Zendesk / Reamaze / Richpanel. It's call transfer or email-in. The ticket arrives like any other email. That's good enough for 95% of teams.

Score it on: handoff latency + context preserved (transcript, customer data, intent).

3 reporting help desk software requirements

Most teams either over-report (50 dashboards no one reads) or under-report (one CSAT score, nothing actionable).

Req 10. CSAT by channel

Aggregate CSAT is a vanity metric. CSAT broken out by channel is where the truth lives.

Why it matters:

• Your email CSAT might be 88%. Your chat CSAT might be 82%. Your phone CSAT might be 64%.
• That phone score is screaming at you. Aggregate hides it.
• Ecommerce industry CSAT average is roughly 82%, per SurveySparrow's 2026 benchmarks. If you're below that, you need to know which channel is dragging.

Score it on: per-channel CSAT + per-agent CSAT + per-issue-type CSAT (WISMO, return, exchange, product question).

Req 11. First Response Time and resolution time

FRT is the metric that correlates most directly with CSAT.

The 2026 benchmarks:

• Email: under 1 hour is excellent, under 4 hours is good, industry average is 12 hours
• Chat: under 40 seconds is the benchmark, satisfaction peaks at 84.7% with 5-10 second FRT
• Phone: under 30 seconds to first human voice. Customers abandon hold queues at 8 minutes on average.

Resolution time matters too, but FRT correlates more strongly with retention.

Score it on: real-time SLA dashboard + breach alerts + per-channel FRT tracking.

Req 12. Channel-level resolution rate

What percentage of contacts get resolved without escalation, broken out by channel.

Why it's the headline metric:

• A 70% chat resolution rate with a 30% voice resolution rate tells you exactly where your AI investment goes next
• It's the only metric that captures "we got better" without confounding from volume changes
• For voice specifically, you want resolution rate alongside abandon rate

Score it on: resolution rate tracked per-channel, including voice. If the helpdesk can't show you voice resolution rate, voice isn't really a channel for them.

4 help desk software requirements most teams forget

This is the section that separates a thoughtful requirements doc from a SERP-listicle copy-paste. These are the four reqs that don't show up in most lists, and they're the four that cost the most to retrofit later.

Req 13. Phone overflow and after-hours coverage

What it is: when your phone team is at capacity or off the clock, what happens to the call.

What most teams do today: voicemail. Customer leaves a message. 60% never call back. You lose the order. The same problem hits inbound on WISMO calls specifically, where the customer is anxious and won't wait.

What you actually need:

• 24/7 inbound coverage: an AI picks up after hours, weekends, holidays
• Overflow routing: when all agents are busy, the call doesn't ring forever, it gets answered
• Callback scheduling: if AI can't resolve and no agent is on, schedule a callback
• Measured abandon rate: track it, it's a leading indicator of revenue loss

24/7 ecommerce phone support is the cheapest way to keep this from happening.

Score it on: 24/7 coverage path + abandon rate tracked.

Req 14. Voice channel deflection rate

You track chatbot deflection. Why don't you track voice deflection?

Deflection rate = % of calls resolved without a human ever touching the call.

The benchmark that matters: across 50+ Shopify brands running Ringly, the AI resolves 73% of inbound calls autonomously at roughly $0.42 per resolved call (vs $7-$16 per call for human agents). That's the order of magnitude you should expect from a real voice AI in 2026.

Score it on: voice deflection % + cost per resolved call. If the vendor can't give you those two numbers, voice isn't a serious part of their product.

Req 15. Escalation to a live agent in under 10 seconds

The handoff from AI to human is where most setups fall apart. The customer says "give me a human," and then there's 45 seconds of hold music and the agent picks up with zero context.

What good escalation looks like:

• Sub-10 second transfer: AI to live agent in under 10 seconds of customer wait time
• Full context preserved: transcript, customer ID, order number, intent label all attached
• Smart routing: VIP customers, angry tone, refund requests above $X, all routed appropriately
• Fallback to email-in: if no agent is available, AI sends the conversation to the helpdesk inbox so it doesn't die

Score it on: transfer latency + context preserved + fallback path.

Req 16. Voice channel coverage (the requirement most lists omit)

Read any "top 13 help desk software requirements" article. Voice is either missing entirely or it's bullet point 11 with one sentence next to it.

The phone is where your angriest, highest-intent, most-likely-to-churn customer goes. WISMO inquiries during peak season hit 50-80% of total support volume per Salesforce's WISMO data, and a meaningful share come in by voice especially after hours.

Most helpdesks log calls. They don't resolve them. That's a different product category, and it needs to be a discrete requirement in your doc.

Ringly.io: AI phone support for Shopify brands

Ringly is AI phone support for Shopify brands. Phone support, without the payroll. Instead of growing your support headcount every time call volume goes up, the AI takes the routine inbound calls so your team can focus on the work that actually moves revenue.

The AI answers inbound calls 24/7. It finds orders in your Shopify store, processes returns and exchanges, answers product questions from your knowledge base, and rescues abandoned carts via outbound follow-up. Across 50+ brands, the AI resolves 73% of calls autonomously at roughly $0.42 per resolved call. Calls that need a human escalate cleanly to Gorgias, Richpanel, Reamaze, or whatever helpdesk you already run.

Plans: Grow $349/mo (1,000 minutes), Pro $799/mo (2,500 minutes), Enterprise custom. 14-day free trial on Pro. Live in under an hour. 65% resolution guarantee: if the AI resolves under 65% of your calls in 90 days, we refund the last 3 months.

Score it on: native voice resolution + clean helpdesk handoff + per-resolved-call economics.

How to turn this into an RFP scorecard

You don't need a downloadable template. You need three columns in a spreadsheet.

Total score per vendor = sum of (weight x vendor score).

Some quick weighting hints based on what I've seen DTC ops leads do:

• Weight 5 the reqs in the "most teams miss" bucket if you sell physical product and field phone calls. Phone overflow, voice deflection, fast escalation, voice coverage. These are where stack costs balloon if you get them wrong.
• Weight 4-5 Req 6 (native Shopify) and Req 7 (carrier data in the ticket). These are ecommerce-specific and your team will live in them daily.
• Weight 3 the baseline reqs (ticket management, omnichannel inbox, automation, SLAs, KB). Every serious helpdesk does these. The weighting separates good from great.
• Weight 2-3 the integration / handoff reqs depending on whether you're running a single helpdesk or a stacked setup.

Run the scorecard on three vendors. Whichever totals highest wins your shortlist. Whichever scores lowest on a weight-5 req drops out automatically.

Ready to score the voice column? Try Ringly.io free for 14 days and you'll have real resolution data for your own store inside 7 days.

Frequently asked questions

What's the difference between help desk software and service desk software?

Help desk = customer-facing support (returns, order questions, refunds). Service desk = internal IT support, usually ITIL-flavored, for employees raising tickets about laptops and Wi-Fi. If you're a Shopify or DTC brand, you want help desk software. Service desk vendors will sell to you, but the workflow won't fit.

Do I need a separate phone tool if my help desk has voice?

Probably yes. Most helpdesk voice add-ons (including Gorgias voice, Zendesk Talk, Help Scout's phone integration) route and log calls. They don't resolve them. If you want the phone to be a real channel that deflects calls instead of just queuing them, that's a different product category. See ecommerce helpdesk phone integration for the breakdown.

How long should writing a help desk requirements doc take?

A focused afternoon if you copy a 15-20 item template like this one. Two weeks if you start from scratch and survey your team. The doc gets better when ops, marketing, and a senior agent all weigh in on weights.

What's the minimum CSAT a Shopify brand should target?

Ecommerce industry average CSAT is roughly 82% per SurveySparrow's 2026 data. Strong DTC brands hit 88-92%. To get there, the phone channel has to be resolved, not just logged. Aggregate CSAT hides phone-specific drag.

Should AI features be a hard requirement in 2026?

Yes, but be specific about which AI. Deflection AI (resolves contacts without a human) is a hard requirement. Summarization AI (writes a TL;DR of a ticket) is useful but optional. Score vendors on the % of contacts they resolve autonomously, not on how often they say "AI" in their demo.

Where does Ringly.io fit in this requirements doc?

Ringly is AI phone support for Shopify brands. It fills Req 8 (voice integration that does more than route calls), Req 14 (voice deflection), Req 15 (sub-10s escalation), and Req 16 (voice coverage). It's not a help desk replacement. Calls that need a human still escalate to your Shopify helpdesk app of choice. It's the voice layer most stacks are missing.

Wrap

If your shortlist dies six months in, it almost always dies for the same reason. The voice channel was an afterthought in the requirements doc. You picked a helpdesk that's great at email and chat, then bolted a phone tool on, then realized the phone tool doesn't deflect calls, just routes them.

Score the four "most teams miss" reqs at weight 5 from day one. The rest of the doc usually sorts itself out.

Try Ringly.io free for 14 days and get the AI answering your calls in under an hour.

Article by Ruben Boonzaaijer. Co-founder of Ringly.io. We build AI phone support for Shopify brands so they can scale support without hiring a phone team.