{"slug": "finance-one-uses-callcoach-to-review-every-customer-call", "title": "Finance One Uses CallCoach to Review Every Customer Call", "summary": "Finance One, an Australian non-bank lender, has deployed Icana.AI's CallCoach to analyze every customer conversation across its contact centers, replacing sample-based quality checks for about 2,000 calls per day. The first year covered 197,514 calls and 21,088 hours of audio, with internal process measures improving by 42% to 48% between May and October 2025, though these are company-reported results without independent accuracy evaluation. The system flags calls for manager review rather than making final decisions, reflecting a governed approach to AI automation in regulated environments.", "body_md": "# Finance One Uses CallCoach to Review Every Customer Call\n\nFinance One says Icana.AI's CallCoach now reviews every customer conversation, replacing sample-based quality checks across an operation handling about 2,000 calls a day. A vendor case study reports that the first year covered 197,514 calls and 21,088 hours of audio, with several internal process measures improving between May and October 2025. These are company-reported operational results, not an independent model-accuracy study.\n\nAustralian non-bank lender Finance One has deployed Icana.AI's CallCoach to analyse customer calls for quality, compliance and signs that a borrower may be experiencing hardship. The deployment was detailed in an Icana case study and in separate July 27 reports from iTnews and IT Brief Australia.\n\n### From samples to full-call review\n\nFinance One CIO Chris Doyle told iTnews that the group's contact centres handle about 2,000 calls a day and 10,000 a week. Calls lasting at least one minute are passed through an internal workflow to CallCoach, which returns analysis to dashboards and reports. Team leaders are then directed to a smaller set of calls that may need follow-up instead of manually sampling recordings.\n\nIcana's case study says CallCoach analysed 197,514 calls and 21,088 hours of audio during Finance One's first year with the system. IT Brief later reported cumulative totals above 300,000 calls and 33,000 hours. Those figures cover different time horizons and should not be treated as contradictory.\n\n### Reported outcomes and limits\n\nThe case study reports a 42% improvement in adherence to Finance One's hardship workflows, a 48% improvement in compliance with its sensitive-data processes, a 42% improvement in query-process adherence and a 14% improvement in tone-of-voice consistency between May and October 2025. Finance One and Icana supplied these measurements; LDS did not retrieve an independent evaluation of the scoring model, its false-positive rate or its ability to detect hardship.\n\nThat distinction matters. Reviewing every call expands coverage, but it does not mean every automated flag is correct. The official case study says the teams performed extensive calibration to address non-deterministic AI behavior and build confidence in the scoring. Managers still review surfaced calls and decide what action is appropriate.\n\n### A governed approach to broader automation\n\niTnews also reports that Finance One has completed an initial AI strategy and framework, is experimenting with LLMs and agents, and has introduced a platform for centrally managing LLM access. Doyle said low-risk, low-volume processes may be prototyped by business users, while core or regulated workflows require more involvement from the technology team.\n\nFor data and AI teams, the useful pattern is the separation of coverage from authority: automation can review a larger population and prioritize attention, but people remain responsible for consequential decisions. A production rollout also needs calibration, access controls, retention rules, drift monitoring and a clear escalation path for errors, especially when conversations involve hardship or sensitive financial data.\n\n## Key Points\n\n- 1Icana's case study says Finance One's first year with CallCoach covered 197,514 calls and 21,088 hours of audio.\n- 2Finance One reported double-digit improvements in several internal process measures, but no independent accuracy evaluation was retrieved.\n- 3The lender uses automated analysis to triage calls for managers rather than treating model flags as final decisions.\n\n## Scoring Rationale\n\nThe deployment provides a concrete regulated-industry example of full-population conversation analysis, operational metrics and human review. The results are relevant to applied AI teams, but they are self-reported and do not include an independent model-accuracy or customer-outcome evaluation.\n\n## Sources\n\nPrimary source and supporting public references used for this report.\n\nPractice interview problems based on real data\n\n1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.\n\n[Try 250 free problems](/problems)", "url": "https://wpnews.pro/news/finance-one-uses-callcoach-to-review-every-customer-call", "canonical_source": "https://letsdatascience.com/news/finance-one-uses-callcoach-to-review-every-customer-call-d8e272d3", "published_at": "2026-07-26 20:16:00+00:00", "updated_at": "2026-07-29 00:27:10.255576+00:00", "lang": "en", "topics": ["artificial-intelligence", "ai-products", "ai-tools", "natural-language-processing", "ai-ethics"], "entities": ["Finance One", "Icana.AI", "CallCoach", "Chris Doyle", "iTnews", "IT Brief Australia"], "alternates": {"html": "https://wpnews.pro/news/finance-one-uses-callcoach-to-review-every-customer-call", "markdown": "https://wpnews.pro/news/finance-one-uses-callcoach-to-review-every-customer-call.md", "text": "https://wpnews.pro/news/finance-one-uses-callcoach-to-review-every-customer-call.txt", "jsonld": "https://wpnews.pro/news/finance-one-uses-callcoach-to-review-every-customer-call.jsonld"}}