What One AI-Resolved Support Ticket Actually Costs in Ecommerce — Gorgias, Intercom Fin, Tidio Lyro, Zendesk, Ada & Kustomer Unit Economics (2026)

**Gorgias Automate** is the AI layer bolted onto Gorgias's Shopify-native helpdesk. It does three things: deflects FAQs with an AI agent trained on your help center and product catalog, auto-tags and routes tickets, and writes draft replies your human agents can edit. The pricing model — 30% of your underlying Gorgias plan plus roughly $0.40 per AI auto-response over your included allowance per https://www.gorgias.com/pricing — means the per-ticket cost is highly sensitive to plan size. A merchant on the $60/mo Basic plan adds ~$18 for Automate; a merchant on the $900/mo Advanced plan adds ~$270. The 'resolution' here is generous — any ticket where the AI sent a response counts, even if a human later replied.

**Intercom Fin AI Agent** is the cleanest pricing on this list. Flat $0.99 per resolved conversation across web, email, WhatsApp, SMS, and in-app messenger, per https://www.intercom.com/fin. 'Resolution' is strictly defined: the customer confirms the issue is solved or doesn't reply within 24 hours after Fin's last message. That definition matters — it means Intercom doesn't bill you for messages where the customer immediately asked for a human. You still pay for an Intercom seat on top ($39–$139/agent depending on tier), but the AI cost is predictable and easy to model.

**Tidio Lyro** is the SMB workhorse. Lyro is metered separately from Tidio seats — the Plus plan at $749/mo includes 50 Lyro conversations, and additional packs run $39 per 50 conversations per https://www.tidio.com/pricing/. That math works out to ~$0.78 per Lyro conversation. Lyro 'conversations' are counted on a 7-day window basis, so a customer who comes back three times in a week is one billable conversation. For a sub-5k-tickets-per-month Shopify store, Tidio's combination of cheap seats and per-conversation AI is hard to beat.

**Zendesk Advanced AI** is the enterprise add-on. It costs $50/agent/month layered on top of Zendesk Suite Pro at $115/agent/month per https://www.zendesk.com/pricing/. The effective per-resolution cost depends entirely on agent productivity — at 500 AI-resolved tickets per agent per month, you're at $0.30 per resolution; at 200 resolutions, you're at $0.80. The AI features include answer bots, agent copilot, generative replies, and intent prediction — broader than just deflection. The hidden cost is the Suite Pro requirement: you cannot buy Advanced AI without it.

**Ada** is enterprise-only automation that lives outside your helpdesk and routes deflected tickets back when needed. Pricing is custom and starts in the high five figures annually per https://www.ada.cx/ — typical contracts I've seen quote in the $0.50–$1.00 per resolution range at 500k+ ticket scale, with volume discounts kicking in past 1M resolutions. **Kustomer KIQ**, Meta's CX product, runs a similar usage-based AI charge on top of $89–$169/agent seats per https://www.kustomer.com/pricing/. Both are overkill for sub-$10M GMV stores.

Vendors price AI support in four distinct ways, and conflating them is the single biggest mistake I see in procurement. Model one is **per-resolution flat fee** — Intercom Fin at $0.99 per resolved conversation per https://www.intercom.com/fin is the cleanest example. You know exactly what each AI ticket costs, you can model break-even, you can forecast next quarter's bill. The downside is you're paying a premium for that predictability — at scale, custom enterprise contracts on Ada or Kustomer beat $0.99 by 30–50%.

Model two is **plan percentage plus per-message overage** — **Gorgias Automate** at 30% of plan plus ~$0.40 per auto-response over allowance per https://www.gorgias.com/pricing. This is friendly to small merchants on the $10–$60 plans (you pay $3–$18/mo for AI) and brutal to anyone on the $900 Advanced plan ($270/mo baseline before overages). The per-message overage compounds — if you have a viral product moment and ticket volume 5x's, your AI bill 5x's too, with no volume discount.

Model three is **per-conversation package pricing** — **Tidio Lyro** with $39 per 50 Lyro conversations per https://www.tidio.com/pricing/. This is essentially a prepaid bucket model. The economics only work if you actually use the bucket — buying a 50-conversation pack and using 12 is a $7.80 effective cost per resolution, not the advertised $0.78. SMB merchants routinely overbuy and underuse here, inflating their real per-ticket cost 3–5x.

Model four is **seat-based AI add-on** — **Zendesk Advanced AI** at $50/agent/month on top of Suite Pro $115/agent per https://www.zendesk.com/pricing/. This is the worst model for predictability because the cost per resolution depends on agent productivity, ticket mix, and AI adoption rate. A high-volume team with strong AI adoption gets to $0.30/resolution; a low-volume team sees $0.80+. Sales reps love this model because it hides AI cost inside seat cost.

The fifth, undisclosed model worth flagging: **enterprise custom**. **Ada** and **Kustomer KIQ** quote on annual usage commits with tiered breakpoints per https://www.ada.cx/ and https://www.kustomer.com/pricing/. Effective per-resolution costs land $0.50–$1.00 depending on volume, contract length, and whether you're a logo win for them. Always benchmark these quotes against Intercom Fin's $0.99 floor — that's the public price they have to beat to win the deal.

HelpScout's 2025 customer service benchmark report puts a fully-loaded human-agent ticket at $5–$12 (https://www.helpscout.com/blog/customer-service-stats/), with US-based teams clustering at $8–$10 and offshore at $3–$5. 'Fully loaded' includes salary, benefits, tooling, training, management overhead, and typical 60–70% agent utilization. That number is the baseline every AI vendor benchmarks against — when **Intercom Fin** says $0.99 per resolution, they're implicitly arguing 5–10x cost reduction versus your human team.

The math gets interesting when you separate ticket types. Order-status, refund-status, and shipping-tracking tickets — roughly 40–60% of ecommerce volume per Gorgias's 2025 benchmark report (https://www.gorgias.com/blog) — are nearly free to resolve with AI. The AI just queries Shopify, fires back the answer, and closes the ticket. Effective cost: the per-resolution fee, full stop. **Gorgias Automate** at $0.30–$0.55 here is excellent unit economics because the AI doesn't need to do anything clever.

Returns, product questions, and 'where is my order, it's been three weeks' tickets are different. These often require AI to understand context, sometimes escalate to a human, and resolution is squishy. Vendors that bill per AI message (Gorgias overage at $0.40, Tidio Lyro at $0.78 per 7-day conversation window) charge for the attempt; vendors that bill per resolved conversation (**Intercom Fin** at $0.99) eat the cost of failed attempts. For ticket mixes with high escalation rates (>30%), Fin's model is cheaper despite the higher headline number.

The deflection-rate variable is what makes vendor ROI calculators dishonest. Vendors quote 50–70% deflection rates, but real-world numbers I see in ecommerce land at 25–45% for well-tuned setups and 10–20% for setups merchants throw up in a weekend. At 25% deflection on a 10k-ticket/month store, AI cost is $750–$2,500/month depending on vendor; human agent cost saved is $12,500–$30,000. That's still 5–15x ROI — but it's not the 20–40x vendors promise.

The real procurement question isn't 'cheapest per ticket.' It's 'what blended cost — AI fee plus human-agent fallback plus any platform seat cost — gets me to my target cost-per-contact?' At sub-5k tickets/month, **Tidio Lyro** plus 1–2 human agents typically beats everyone. At 5k–50k tickets/month, **Gorgias Automate** or **Intercom Fin** win on different ticket mixes. Above 50k, **Zendesk Advanced AI** or custom **Ada** contracts pull ahead on volume discounts.

**Gorgias Automate** has the deepest Shopify integration on this list — it's been the Shopify-native helpdesk since 2018 and the AI agent can natively read order data, modify orders, issue refunds, and reship without writing custom workflows. That's worth real money: every other vendor on this list requires you to build refund/reship automations either via Shopify Flow, custom webhooks, or the vendor's no-code builder. For a Shopify DTC brand, this single architectural advantage is often decisive — and it's why Gorgias gets away with the 30%-of-plan AI premium documented at https://www.gorgias.com/pricing.

**Intercom Fin** integrates with Shopify via the official Intercom Shopify app but the depth is shallower than Gorgias — Fin can read order data, but actions like refunds require Intercom Workflows configuration. The upside is Fin works equally well across web messenger, email, WhatsApp, and SMS, where Gorgias is strongest in email/chat. If your support volume is multi-channel and you don't want to maintain two AI configurations, Fin's architecture wins. Pricing for the Intercom seat plus Fin add-on is documented at https://www.intercom.com/fin.

**Tidio Lyro** sits inside the Tidio platform — a Shopify app store install gets you live chat, email, and Lyro AI in one widget. Architecturally simpler than Intercom or Gorgias, with less customization headroom. Lyro is trained on a knowledge base you upload and product feeds it pulls from Shopify. For sub-5k-ticket stores, that's plenty. For complex DTC brands with subscription, RMA, and multi-warehouse logistics, Lyro hits its ceiling quickly. Pricing details at https://www.tidio.com/pricing/.

**Zendesk Advanced AI** sits on top of Zendesk Suite — which means you're committing to Zendesk's entire data model (tickets, organizations, users, custom fields) before you ever turn on AI. That's a non-trivial migration cost if you're not already on Zendesk. The AI itself is genuinely sophisticated — intent prediction, generative replies, agent copilot, answer bot — but the architectural commitment is large. Pricing breakdown at https://www.zendesk.com/pricing/.

**Ada** and **Kustomer KIQ** are platform-replacement plays, not add-ons. Ada wants to be your automation orchestration layer across whatever helpdesk you use. Kustomer wants to replace your helpdesk entirely with their CRM-style model. Both require 3–6 month implementation timelines and dedicated CX engineering resources. The unit economics only pencil out above 500k tickets/year — below that, you're paying enterprise prices for SMB volume.

**Scenario A: Shopify DTC brand, 2k–10k tickets/month, primarily email + chat, US team.** Winner: **Gorgias Automate** on the $60–$360 plans plus Automate at 30% premium. Effective cost per AI resolution lands $0.30–$0.55 per https://www.gorgias.com/pricing, and the deep Shopify actions (refund, reship, exchange) mean the AI actually closes tickets instead of just deflecting them to a human. Total AI spend on a 5k-ticket/month store: ~$200–$400/month. Beat that with anything else on this list.

**Scenario B: Multi-channel DTC, 10k–50k tickets/month, WhatsApp + SMS + email + chat, growing internationally.** Winner: **Intercom Fin** at $0.99 per resolution per https://www.intercom.com/fin. The per-resolution flat fee is predictable as you scale globally, Fin works identically across channels, and you avoid building parallel AI configs for WhatsApp Business and SMS. Total AI spend at 30% deflection on 30k tickets: ~$8,900/month. Pricier than Gorgias on equivalent volume but operationally simpler.

**Scenario C: SMB store, sub-2k tickets/month, single founder + 1 part-time agent.** Winner: **Tidio Lyro** on the Growth or Plus plan plus a Lyro conversation pack per https://www.tidio.com/pricing/. Total monthly bill: $59–$749 plan + $39–$200 in Lyro packs. The economics only break if you let unused conversation packs roll over unused. For this scenario, even Gorgias is overkill — Tidio's all-in-one widget plus Lyro is the right tool.

**Scenario D: Enterprise multi-brand, 100k+ tickets/month, omnichannel + multi-region.** Winner is typically **Zendesk Advanced AI** if you're already on Zendesk Suite, or **Ada** if you're greenfield and want best-in-class automation. Zendesk math: $115 Suite Pro + $50 Advanced AI = $165/agent/month, plus ~$0.30/AI resolution at volume per https://www.zendesk.com/pricing/. Ada quotes typically come in 20–40% below that at equivalent volume, but require 3–6 months to deploy.

**Scenario E: Meta-ecosystem brand heavy on WhatsApp + Instagram.** **Kustomer KIQ** is the obvious play given Meta ownership and the deep WhatsApp Business + Instagram DM integration per https://www.kustomer.com/pricing/. Effective per-resolution costs in the $0.50–$1.00 range. The catch is enterprise minimums — Kustomer is not a fit below $50M GMV. For mid-market Meta-heavy brands, **Intercom Fin** with WhatsApp Business integration is usually a better economic fit at $0.99/resolution flat.

Before you sign anything, ask each vendor exactly how they define a 'resolution' or 'conversation' for billing. **Intercom Fin** publishes its definition (customer confirms or doesn't reply within 24 hours) at https://www.intercom.com/fin. **Tidio Lyro** uses a 7-day window per https://www.tidio.com/pricing/. **Gorgias Automate** bills on auto-response sent, not resolution per https://www.gorgias.com/pricing — meaning failed AI attempts still cost you. Make this definition contractual; don't accept marketing-page language.

Data residency matters for EU merchants under GDPR and increasingly for US merchants under state privacy laws. **Intercom** offers EU and AU data residency on Advanced and Expert tiers. **Zendesk** offers US, EU, AU, and JP regions on Suite Pro and above per https://www.zendesk.com/pricing/. **Gorgias** offers EU residency. **Tidio** runs EU customers on Frankfurt infrastructure. **Ada** negotiates residency contractually. **Kustomer** runs US and EU regions. If you have any GDPR exposure, get residency in writing and make sure your AI prompts and customer messages stay in-region — some vendors process AI calls in the US even if data is stored in the EU.

SSO/SAML is gated to higher tiers across the board. **Gorgias** Pro and above, **Intercom** Advanced ($99/seat) and above, **Tidio** Plus ($749/mo) and above, **Zendesk** Suite Growth and above. If SSO is non-negotiable for your security team, your floor pricing just went up significantly. Budget for it explicitly; don't let a vendor sell you the Starter tier and then sandbag SSO as an upgrade conversation in month three.

Annual contract terms vary in punitive ways. **Gorgias** and **Tidio** allow monthly billing at no surcharge. **Intercom** requires annual on Advanced and above. **Zendesk** Suite Pro requires annual contracts with limited mid-contract reduction rights. **Ada** and **Kustomer** typically require multi-year commits with annual usage minimums — even if you don't use the AI volume, you owe the minimum. Negotiate a usage-based ramp clause if you're growing fast, or a downgrade option if you're not sure of volume.

Finally, AI model quality and prompt customization matter more than vendors admit. **Intercom Fin** uses GPT-4-class models with extensive Intercom-tuned guardrails per https://www.intercom.com/fin. **Gorgias** uses a mix of in-house and OpenAI models. **Zendesk** uses OpenAI under the hood. **Tidio Lyro** has historically been more opaque about its model stack. If your brand voice matters or you need fine-grained prompt control, build your system prompts externally — exactly what tools like our AI Prompt Generator are designed for — and treat the vendor's prompt UI as a thin wrapper, not your prompt strategy.

Per-resolution cost is the headline number, but it's not what you pay. Total cost of ownership on AI support is seat cost + AI fee + adjacent Shopify apps (returns, shipping, reviews, subscriptions) — and merchants routinely underestimate the bundle by 30–50%. On a 10k-ticket/month Shopify DTC brand running **Gorgias** Advanced ($900) + Automate (30% = $270) + estimated $200 in overages, the AI line item is ~$1,370/month — but the full helpdesk bundle including order-status app, returns app, and shipping integration usually runs $2,000–$2,500/month per the cost breakdown in our AI Shopify app cost calculator.

**Intercom Fin** plus Intercom seats at scale gets expensive fast. A team of 8 agents on Intercom Advanced ($99/seat) is $792/month in seat cost, plus Fin at $0.99 × 3,000 AI resolutions = $2,970/month, for $3,762 total. That's the all-in cost. Compare against Gorgias on equivalent volume: seat cost is bundled into plan price, so a $900 Advanced plan + Automate $270 + overages $200 = $1,370. Gorgias is roughly half the price at this volume — but Intercom buys you multi-channel parity that matters if WhatsApp and SMS are 30%+ of your inbound.

**Tidio Lyro** is the cheapest TCO under 2k tickets/month, full stop. Growth plan at $59 + Lyro conversation packs at $39–$200/month gets you under $300/month all-in. Above 5k tickets/month it loses its edge — Plus plan jumps to $749, and Lyro conversation costs start adding up. The Tidio sweet spot is small, and merchants outgrow it. Plan your migration path early — moving from Tidio to Gorgias or Intercom at 10k tickets/month is a 3–4 week project.

**Zendesk Advanced AI** TCO is where finance teams get blindsided. Suite Pro is $115/agent, Advanced AI is $50/agent — so an 8-agent team is $1,320/month before you've resolved a single ticket. Add typical Zendesk add-ons (Sunshine Conversations $500+, advanced reporting, custom apps), and a real Zendesk deployment is $2,500–$4,000/month for an 8-agent team. The AI per-resolution cost is competitive at $0.30 at volume, but the platform overhead is the largest on this list.

If you want to model the full Shopify app stack — helpdesk plus AI plus returns plus shipping plus reviews — across vendors, our AI Shopify app cost calculator lets you plug in ticket volume and see the bundled monthly bill. The point of doing this math up front: the wrong helpdesk choice locks you into a multi-year switching cost. Get the unit economics right before you're 18 months in and discovering your AI bill is 3x what you budgeted.

For most ecommerce merchants under $50M GMV, building your own AI agent on OpenAI or Anthropic APIs is a worse business decision than buying. The per-call API cost is dramatically lower — a typical GPT-4o resolution costs $0.02–$0.05 in raw API spend — but the engineering cost to build the helpdesk integration, knowledge ingestion, escalation logic, audit trail, and compliance tooling routinely runs $150k–$400k in year one. At under 50k tickets/year, the buy-vs-build math heavily favors **Intercom Fin** or **Gorgias Automate**.

Above 500k tickets/year, custom builds start to pencil out — that's why **Ada** exists as the middle ground, and why enterprise CX teams increasingly use Ada or in-house automation platforms with raw OpenAI/Anthropic APIs underneath. At that volume, $0.99/resolution × 500k = $495k/year in pure AI fees, and you can hire a small team and pay API costs for less. The trigger isn't ticket volume alone — it's also whether your AI handling has truly differentiated logic (subscription billing, complex returns, multi-warehouse) that off-the-shelf vendors can't accommodate.

The hybrid model worth considering: buy the helpdesk (**Gorgias** or **Zendesk**), skip the vendor AI add-on, and pipe tickets through your own OpenAI/Anthropic prompt stack via webhooks. This is increasingly common — vendor AI prompt engineering is generic, your prompts can be brand-specific. The savings: you avoid the 30% Gorgias Automate premium or $50/agent Zendesk Advanced AI fee. The cost: engineering time to build and maintain prompt versioning, evals, and the integration layer.

If you're going this hybrid route, the prompt engineering becomes your moat. This is where production-grade prompt tooling matters — our AI Prompt Generator builds system prompts that work across **OpenAI**, **Anthropic Claude**, **Google Gemini**, and **Meta Llama** APIs with consistent behavior, eval coverage, and version control. The point isn't that prompts are hard — it's that maintaining 30+ production prompts across model versions without regressing is genuinely engineering-grade work.

The build path also lets you control AI cost variance, which matters at scale. **Intercom Fin** at $0.99 per resolution is predictable but high. Raw API cost on Claude Haiku or GPT-4o-mini for a typical ticket-resolution prompt is $0.005–$0.02. The gap — $0.97 per resolution — is what you're paying Intercom to run the platform, do the integrations, handle the eval, and stand behind the SLA. That's a fair trade at low volume. At 500k tickets/year, it's a $500k question worth answering deliberately.