Product updates
Gorgias adds version markers to Skills performance trends
Vendor-reportedChecked: [release][guide]
Editorial attribution: Gosia Madej
Prepared with AI assistance. Editorial attribution does not imply personal product testing or manual review of every story.
Published: · Event date:
Gorgias’s October 26 announcement lets teams compare a skill’s handovers and CSAT around published updates. The useful distinction versus competitors is monitoring changes versus testing their effects: Decagon documents controlled experiments, while Fin adds recommendations from unresolved conversations.
Editorial assessmentChecked: [release][guide][decagon][fin-recommendations]
What changes for users
For a refund skill, first inspect handovers and rated tickets around the version marker, then compare similar request types and survey coverage. If the decision is whether to roll a risky change out more widely, ask for a controlled test and its sample size, exposure definition and regression limits. A clean trend line alone is weaker evidence for that decision.
Editorial assessmentChecked: [guide][decagon][fin-csat]
Open AI Agent > Skills, select a skill, then Performance > Trend. Gorgias says the chart shows handovers or CSAT over time, marks each published skill version and reveals its publication date on hover. The announcement describes a default 28-day view and a custom date range.
The supporting guide lists Owner, Admin and Lead roles, all plans with AI Agent, and connected Shopify stores. Teams can open the tickets behind a skill’s metrics, including handovers and conversations with satisfaction ratings. This is a per-skill investigation workflow, rather than just an aggregate AI Agent score.
Vendor-reportedChecked: [guide]
The version markers make it easier to spot a change worth investigating. They do not establish that the edit caused a CSAT or handover movement: the mix of requests, ticket volume and who answers surveys can also change. Read the underlying conversations before attributing an improvement to a version.
Editorial assessmentChecked: [guide]
For isolating a change’s effect, Decagon’s documented Experiment Hub goes further than a before-and-after chart. Its October 22 announcement describes experiment blocks with a shared control and up to 20 variants, locked test content while other updates continue, and results limited to conversations reaching the tested step. It also describes confidence intervals and significance judgments, plus Simulations before live traffic. These are vendor-described capabilities; the announcement does not establish edition eligibility or account-level enablement.
Editorial assessmentChecked: [decagon]
Fin/Intercom offers a different operational advantage: Recommendations links unresolved conversations to proposed content, customer-data and action fixes. Its guide requires the Pro add-on; audience segmentation currently applies only to content recommendations. For a team deciding what to fix next, that documented prioritization workflow can be more useful than a trend alone. It is not equivalent to a controlled experiment or proof that a recommended fix works.
Editorial assessmentChecked: [fin-recommendations]
Fin’s CSAT guide also documents opening ratings and remarks to inspect conversations, plus rating filters in custom reports. Fin CSAT is available on plans including Fin AI Agent, but the granular Conversation rating dataset’s CSAT attributes require Advanced or Expert. Fin defines its CSAT score as the share of positive ratings; Gorgias describes an average satisfaction rating. Those descriptions do not justify treating the two numbers as interchangeable.
Editorial assessmentChecked: [fin-csat][guide]
Does a competitor do it better? For documented controlled change evaluation, Decagon has the stronger case; for suggested fixes and content-audience prioritization, Fin has the more explicit workflow in these sources. Gorgias’s addition is useful for checking a particular skill after an edit. This is a comparison of documentation, not hands-on usability, accuracy or value testing, and it establishes no overall winner. A capability absent from these pages is not proven absent from a product.
Editorial assessmentChecked: [guide][decagon][fin-recommendations][fin-csat]
Checked against sources
AI-assisted source check
How this was checked
AI-assisted source checking used the full retained October 26 Gorgias announcement, its currently accessible Skills guide, Decagon’s full October 22 Experiment Hub announcement and two current Intercom guides. All are vendor publications; no product account, experiment, simulation or customer outcome was tested.
Editorial assessmentChecked: [release][guide][decagon][fin-recommendations][fin-csat]
The Gorgias announcement’s publication timestamp converts to October 26, 2026, 11:20 UTC. Its wording says the feature can now be used, but does not establish the exact first rollout time or percentage of accounts enabled. The supporting guide’s relative update label was not used as a launch date. Later retrieval of the announcement and its image failed; the earlier full read was retained, without bypassing access restrictions.
The comparison distinguishes observational trends, controlled live experiments, pre-live simulations and suggested fixes. Vendor documentation does not independently verify statistical safeguards or measured customer gains. No genuine product screenshot was captured; the lead image is an original conceptual illustration with no measured data.
Editorial assessmentChecked: [guide][decagon][fin-recommendations]
Source evidence
Original source excerpt. This shows what the source says; it does not establish product performance.
Markers on the chart show each time you published a new version.
Sources
- [release] Gorgias: See how each skill performs over time, and whether your changes workedVendor-reportedPublished: 26 October 2026 · Checked: 26 October 2026
- [guide] Gorgias: See how your skills and knowledge are performingVendor-reportedSource publication date not stated · Checked: 26 October 2026
- [decagon] Decagon: Introducing Experiment HubVendor-reportedPublished: 22 October 2026 · Checked: 26 October 2026
- [fin-recommendations] Intercom: Optimize Fin instantly with the help of AIVendor-reportedSource publication date not stated · Checked: 26 October 2026
- [fin-csat] Intercom: Analyze and report on Fin AI Agent CSATVendor-reportedPublished: 21 July 2026 · Checked: 26 October 2026