Dixa

What it is

Dixa is a Copenhagen-built conversational helpdesk for consumer brands: phone, email, chat, web forms, social, WhatsApp, and SMS land in one queue and are routed to the next free agent instead of sitting in separate inboxes. It is closest in shape to Kustomer or Gladly, with the contact centre (IVR, call recording, callbacks) bundled into every plan rather than bolted on. Its AI agent, Mim, answers customers across the text channels and pulls order data from Shopify and other commerce tools. Dixa was founded in 2015, has raised $158M (a $105M Series C led by General Atlantic in 2021), and says it serves 850+ brands in 42 countries.

Why it shows up in Customer Success stacks

• Voice is in the base price. Every plan, including Growth, carries phone, IVR, call recording, callback, and voicemail. Zendesk sells its contact centre as a separate per-seat add-on, which is the gap Dixa’s own pricing page is built around.
• The AI meter is per conversation, not per resolution. Mim bills a flat rate on every conversation it handles, whether it resolves it or hands it to a human. That is the opposite bet from Intercom Fin or Gorgias, and it rewards brands whose automation rate is already high.
• The 2026 roadmap is commerce-first. Mim reached WhatsApp, Messenger, Instagram, and SMS in Q1 2026. Q3 shipped Mim version control with rollback, deeper Shopify, Recharge, Klaviyo and WooCommerce integrations, and two betas: Mim on Voice (major EU languages plus UK and US English) and sales attribution per agent, human or AI.
• EU data posture is native. A Danish vendor with EUR and GBP billing and EU AI Act compliance work on voice is an easier procurement conversation for EU retailers than a US helpdesk.

Pricing reality

Three plans, priced per agent, with no free trial and every plan sold through a demo.

• Growth — €89/agent/month billed annually, €109 monthly. Omnichannel queue, contact centre, knowledge base, Mim, SLAs, 35+ integrations.
• Ultimate — €139 annual, €169 monthly. Adds skills-based routing, routing on external data, advanced automations, intent detection, and a sandbox.
• Prime — €179 annual, €215 monthly. Adds SSO, custom roles, auto-redaction, multiple organisations, and Advanced Insights.

The older €39 Essential plan is gone, so the entry point roughly doubled. On top of seats: Mim at €0.40 per conversation (Dixa’s FAQ, updated July 2026; guides written before then still quote €0.35), AI Co-Pilot at €39/seat, and unpublished prices for Quality Assurance, AI Auto QA, AI voice transcription, and voice usage.

A worked example: 10 Growth agents on annual billing (€890), Mim on 3,000 conversations (€1,200), and Co-Pilot for everyone (€390) comes to about €2,480/month, or roughly €29,800 a year, before QA and call minutes. The per-conversation bet in numbers: at Intercom’s $0.99 per resolution (about €0.87 at June 2026 rates), Mim is cheaper only if it resolves more than about 46% of what it touches (estimate). Below that, a per-resolution meter would have billed you less.

Best for

Heads of customer service at mid-market European consumer brands — e-commerce, subscription, retail — with roughly 10-100 agents, where phone is still a real channel and Shopify-style order data drives most contacts. It fits best when you want one vendor for the queue, the phones, and the AI agent, and when Mim can carry a high share of order-status and returns traffic.

Do not buy Dixa if you are a B2B SaaS team supporting accounts in Slack (look at Pylon or Intercom), a small shop under about five agents with no phone line (Gorgias or Freshdesk is cheaper), or a US-first team that wants USD billing and a self-serve trial.

Versus the alternatives

• Zendesk — the market-share leader; pick it when you need its app marketplace, admin talent pool, or global scale, and accept paying separately for voice and Copilot.
• Gorgias — the default helpdesk for Shopify brands; pick it under about 10 agents or when ticket-based pricing beats per-seat, and voice is minor.
• Kustomer and Gladly — the closest conversation-centric peers; pick them for US-centred enterprise B2C with larger budgets.
• Siena AI or Ada — the overlay route; pick one when you want a stronger AI agent but not a helpdesk migration. Both already connect to Dixa, so this is also the escape hatch if Mim underperforms.

Watch-outs

• You pay for Mim’s failures. Every conversation Mim touches is billed, including the ones it hands to a human. Guard: before switching Mim on for a channel, sample 200 recent conversations on it, estimate the share Mim could close, and only enable it where that share clears your break-even rate.
• Most add-ons have no list price. QA, Auto QA, transcription, and voice usage are quoted per deal, and Dixa’s own comparison counts “AI and voice usage” outside the seat price. Guard: get every add-on and per-minute rate into the order form, with a cap on the Mim rate for the contract term, since guides from earlier in 2026 quote €0.35 against today’s €0.40.
• No official MCP server, and API auth is admin-issued static tokens. The only Dixa MCP server is a community project last updated in mid-2025. Guard: if agents or internal tools need helpdesk access, scope a dedicated API token with the new per-token permissions and plan to build on the REST and Exports APIs yourself.
• Mim on Voice is still beta. The voice agent is limited to customers who opt in. Guard: buy on the text-channel Mim you can test today, and keep phone traffic with human agents until voice is generally available.

For forecasting and scheduling the agents a Dixa queue routes to, pair it with Assembled; for theme analysis across its conversations, Chattermill.