TL;DR: Support and CS teams build a ring of tools around the help desk: triage add-ons, AI agents billed per resolution, QA tools, a customer-success platform for health scores, a survey tool and an escalation integration. A full CS platform often costs tens of thousands of dollars a year, and AI agents add per-resolution fees on top. Intercom Fin, for example, is $0.99 per outcome. Keep the help desk and the tools that own a real data model. Move triage context, health scores, CSAT follow-ups, renewal-risk digests and engineering escalations into Ballet playbooks.
The help desk handles tickets well. But most of the questions a support agent or CSM asks are answered in other systems:
Each question tends to get its own tool or integration, and agents still switch between tabs to answer them.
Prices are published list prices at the time of writing (November 2026) unless marked as buyer-reported.
| Category | Typical tools and price | Move into a Ballet playbook | Keep the specialist if you need |
|---|---|---|---|
| Ticket triage and context | Triage add-ons; Zendesk Support plans from $19/agent/month (pricing) | Tag and route by plan tier (looked up in Stripe), language, topic or SLA; add the account owner and renewal date to the ticket | |
| AI resolution | Intercom Fin $0.99 per outcome, on top of your plan (pricing); Zendesk AI agents use resolution-based tiers (docs) | Draft replies from your Brainfish knowledge base for agents to review, with billing and account context attached | A fully autonomous, customer-facing chat agent inside the help desk's own widget |
| Customer health scores | Gainsight, ChurnZero, Vitally, Planhat and Totango are quote-only; third-party estimates put typical contracts at $35,000–$50,000+/year (example estimate) | A health score from ticket volume, payment status, plan tier and survey results, written to the CRM; a weekly renewal-risk digest | A full CS data model with account timelines, adoption telemetry and stakeholder maps across a large CSM team |
| CSAT and NPS | Nicereply from $59/month billed annually (pricing); AskNicely quote-only | A post-resolution CSAT email, response capture, detractor alerts and recovery tasks | Survey infrastructure at high volume, sampling and benchmarks |
| QA and coaching | Zendesk QA is quote-only; MaestroQA buyer-reported from about $35/user/month (G2) | Sample conversations, flag low CSAT or policy breaches and build a weekly coaching queue | Scoring every conversation, calibrated scorecards, speech analytics |
| Engineering escalation | Native help desk integrations, for example Zendesk for Jira | Create a Linear or Jira issue with ticket, customer, plan and reproduction context; check for duplicates; update the ticket when it's fixed |
Rule of thumb: keep the systems where your team works (the help desk) and any platform that owns a data model you rely on. Move the lookups, scores, alerts and follow-ups that join those systems into playbooks.
These are illustrative scenarios built from Ballet's Zendesk, Intercom, Freshdesk, Stripe, CRM and Brainfish integrations, plus Slack and Linear through MCP. They are not case studies of named customers.
Before: every ticket lands in one queue. Agents open Stripe to check the plan and the CRM to find the owner before deciding what to do.
With Ballet: a playbook runs on each new ticket:
Ask Ballet prompt:
When a new Zendesk ticket arrives, look up the requester's email in Stripe. Tag the ticket with their plan, and if they're on Enterprise or renew within 60 days, move it to the Priority group and notify the account owner in Slack. Add a suggested reply from our Brainfish docs as an internal note.
Before: a CS platform bought mainly for its health score, while CSMs still live in the CRM.
With Ballet: a nightly scheduled playbook gathers each account's:
It calculates a score with your weights, writes it to a CRM field, and every Monday sends each CSM the accounts that dropped into red.
Before: support pastes a ticket link into Slack, an engineer asks which customer and plan it is, and nobody tells the customer when it's fixed.
With Ballet: tagging a ticket bug starts a playbook that checks Linear for a matching open issue, creates one with the customer, plan, revenue and steps to reproduce if none exists, links it to the ticket, and adds an update to the ticket when the issue closes.
A hypothetical 10-agent support team with 3 CSMs:
| Tool | Price basis | Annual cost | After consolidation |
|---|---|---|---|
| CS platform (health scores) | Third-party estimate | ~$35,000+ | Health score and renewal digest move into playbooks |
| Nicereply | $59/month | $708 | A CSAT playbook using your existing email |
| QA tool (10 users) | ~$35/user/month, buyer-reported | ~$4,200 | A weekly sampled QA queue. Keep a QA tool if you need full coverage |
In this example, about $40,000 a year could be replaced. Teams with large CSM organisations and complex adoption data often still need a CS platform, so treat this as the upper end for smaller teams. Set the savings against the cost of your Ballet plan.
See the parent article, Tool consolidation with Ballet, for the full method.