E-commerce
June 26, 2026
An e-commerce chatbot can save time, respond faster, and help increase sales. But before signing up for a subscription, one question matters more than the demo: how much does it actually return compared to what it costs?
The ROI of a chatbot is not calculated like AOV, CAC, or LTV. Here, the stakes are more operational: avoided tickets, saved agent time, assisted conversion, saved carts, handoff quality, and maintenance costs.
This article #19 provides you with a concrete method to build an honest business case, featuring simple formulas, conservative scenarios, and playbooks to validate the numbers post-launch.
Summary
Why calculate the ROI before launching?
ROI forces us to separate the desire to automate from economic reality. A chatbot can be excellent in a shop saturated with WISMO, and not very profitable in a low-traffic shop with few repetitive questions.
What the calculation decides
Budget: subscription, setup, maintenance, supervision
Scope: support, sales, or both
Priority: intents to automate first
Validation: KPIs to track at 30, 60, and 90 days
The angle is intentionally different from AOV, CAC, or LTV articles: here we measure a support and conversion investment, not the overall profitability of the shop.
Which costs should be included?
A credible ROI starts with the total cost, not just the sticker price.
Subscription: monthly/annual fee, AI option, channels
Usage: cost per conversation, resolution, or ticket processed
Setup: Shopify connection, content, intents, testing
Maintenance: updating policies, products, promotions
Supervision: quality review, escalations, corrections
Gorgias points out that an AI business case must connect costs, time saved, customer experience, and revenue, rather than just counting automated interactions (Gorgias, AI Agent ROI).
Also, add the costs of translation, compliance, and training if the bot covers multiple markets or several teams.
How can support savings be quantified?
Start with your actual ticket mix.
Export 30 to 90 days of tickets
Classify WISMO, returns, invoices, sizing, stock, disputes
Identify repetitive and well-documented inquiries
Calculate the average cost of a human ticket
Apply a conservative AI resolution rate
Formula: monthly support savings = automatable tickets × AI resolution rate × human ticket cost.
Example: 800 tickets/month, 50% automatable, human cost €7, conservative AI resolution 45%. Savings = 800 × 50% × 45% × €7, which equals €1,260/month.
What automation rate should be used?
Avoid maximalist promises. Take a low, median, and high scenario.
Low: 25% of automatable requests resolved
Median: 40 to 50% after source enrichment
High: 60% or more on simple and highly documented intents
Gorgias indicates in its 2026 data that the median brand resolves 45% of tickets touched by AI end-to-end, while the top quartile reaches 65% (Gorgias research). Use these benchmarks as guardrails, not guarantee.
How do you calculate conversion gains?
The chatbot can also help to sell: answers about size, comparisons, delivery, returns, recommendations, abandoned carts.
Prudent formula
monthly conversion gain = additional attributable orders × average gross margin.
Realistic attribution
Do not count all orders after chat as being fully won by the bot. Compare assisted and non-assisted sessions, or use an attribution coefficient: 20%, 30%, 50% depending on the level of proof.
Intercom explains that AI agents are evolving from a pure measurement of resolution towards outcomes: resolving, qualifying, performing an action, or preparing a useful handoff (Intercom, outcomes). This logic also helps to measure assisted conversion.
Which formula should be used in the spreadsheet?
Keep the formula readable for the entire team.
ROI (%) = (annual gains - annual costs) / annual costs × 100
annual gains = support gains + conversion gains + proven operational gains
annual costs = subscription + usage + setup + maintenance + oversight
Payback
payback in months = launch costs / net monthly gains. If the payback exceeds 6 to 9 months on a support tool, check the scope or start smaller.
Essential columns
Add a hypothesis column, a source column, and an observed reality column. This way, after 30 days, you will know what was correct, over-optimistic, or poorly measured.
Small Shopify example: support first
DTC store: €45,000 in revenue/month, 450 tickets/month, €72 average order value, 40% gross margin.
Costs: €179/month, 8 hours setup, 2 hours maintenance
Support: 45% of tickets can be automated, 45% resolved, €7 per ticket
Conversion: 15 additional orders attributed at 30%
Support savings: 450 × 45% × 45% × €7 = €638/month. Conversion gain: 15 × €72 × 40% × 30% = €130/month. Monthly gains: €768. If the average total monthly cost is around €300, the ROI quickly becomes positive.
Decision
For this profile, the test must remain simple: automate tracking, simple returns, and frequent product questions. Adding too many scenarios from the start increases setup time and slows down the payback.
Growing DTC example: support + sales
DTC brand: €180,000 in monthly revenue, 1,600 tickets/month, average order value (AOV) €88, 42% margin, high product traffic.
Costs: €499/month, 25h setup, 6h/month maintenance
Support: 55% automatable, 50% resolved, €9 per ticket
Sales: 90 additional assisted orders with 40% attribution
Support savings: 1,600 × 55% × 50% × €9 = €3,960/month. Conversion gain: 90 × €88 × 42% × 40% = €1,331/month. Even with €1,000 of fully loaded monthly cost, the business case is solid if CSAT and handoff remain clean.
Decision
In this profile, the ROI often justifies a broader scope: post-purchase, product inquiries, shopping assistant, and VIP escalation. However, each expansion must be measured separately.
Which limits should be integrated?
Quality: a deflected but poorly resolved ticket is expensive
Repetition: a customer who returns twice cancels out part of the gain
Handoff: a slow escalation destroys the experience
Data: catalog, stock, and policies must remain up to date
Causality: an open chat does not mean a purchase was caused by the chat
A good ROI therefore includes safeguards: CSAT, recontact rate, escalation rate, negative reviews, refunds, and response errors. A financial figure without quality is a trap.
Spreadsheet error
Do not value the same gain twice. If a conversation avoids a ticket and generates an order, distinguish the support share and the sales share with a clear attribution rule.
How to validate at 30, 60, and 90 days?
At 30 days
Check volume, most frequent intents, incorrect answers, escalation rate, and first avoided tickets.
At 60 days
Compare cost per automated conversation, CSAT, assisted conversion, and agent time saved.
At 90 days
Decide: expand, reduce, change placement, or enrich sources. The ROI must be reviewed with real data, not with launch assumptions.
To be linked with e-commerce chatbot KPIs and human handoff.
Stop threshold
Define before launch the threshold that stops or scales down the pilot: for example, too low CSAT, too high escalation, or cost per resolution higher than human cost.
How does Qstomy help prove ROI?
Qstomy helps connect conversations, Shopify data, and business impact: fewer repetitive tickets, more helpful pre-purchase answers, and clearer chatbot management.
DTC Beauty Scenario
Shopify store, 1,200 tickets/month. Qstomy starts on WISMO, simple returns, sensitive skin questions, and routine recommendations. Pilot hypothesis: 480 automatable tickets, 45% resolved, cost per ticket €8, representing €1,728 in monthly support savings. On the sales side, 70 assisted orders with a €64 basket size, 45% margin, 35% attribution, representing €706 in assisted margin.
Estimated gains: €2,434/month before costs. If the total cost of the pilot is €650/month, the ROI becomes viable from the very first weeks, provided CSAT and escalations are tracked.
See ROI calculator, AI customer support, AI sales assistant, Shopify integration and request a demo.
Which playbooks should be applied this week?
Playbook 1: ticket mix
Export 90 days of tickets. Classify by reason, complexity, channel, estimated cost, and automation potential.
Playbook 2: cost per ticket
Calculate the real human cost: agent time, tools, management, seasonal overload.
Playbook 3: low-case scenario
First build the ROI with a 25% resolution rate on automatable requests. If the project still holds up, it is robust.
Playbook 4: conversion proof
Track assisted sessions and compare margin, cart addition, and conversion with a comparable unassisted group.
Useful internal linking
Tickets: reduce support tickets
Channel: WhatsApp or onsite chat
Data: train with Shopify
Enzo
June 26, 2026
