Michael Hill picks Impact Analytics for AI merchandise planning
Michael Hill, the international fine jewellery retailer operating more than 280 stores across Australia, New Zealand and Canada, has selected Impact Analytics to overhaul its merchandise planning capabilities. The deployment will cover demand forecasting, inventory replenishment, store clustering and assortment planning, consolidating those functions onto a single AI-native SaaS platform.
The contract follows a competitive evaluation process in which Michael Hill said it prioritised retail-specific AI maturity over general-purpose planning tools. Richard Price, Head of Supply Chain Systems at Michael Hill, cited confidence in the forecasting engine as a decisive factor, particularly for high-value, low-volume product categories such as fine jewellery. "Accurate forecasting underpins every planning decision we make, and the results we saw across our assortment reinforced that we had selected the right partner," he said.
The deal
Matt Keays, Chief Technology Officer at Michael Hill, framed the change as a structural shift in how planning teams operate, moving from spreadsheet-driven, line-by-line analysis to an exceptions-management model. The company did not disclose contract value, implementation timeline or the name of any incumbent vendor being replaced.
Impact Analytics, founded over a decade ago, positions its platform as purpose-built for retail rather than adapted from manufacturing or FMCG planning systems. The vendor says its platform is powered by more than one million machine learning models and covers end-to-end planning, pricing and promotions. The company has previously highlighted recognition from Inc. 5000 and the RIS Leaderboard, though independently verified benchmark comparisons were not included in the announcement.
Market context
The retail merchandise planning software market is a well-established category with a number of significant vendors, including Blue Yonder, o9 Solutions, Relex Solutions and Anaplan, alongside newer AI-native challengers. Competition has intensified as legacy players have added machine learning layers to existing platforms, making it harder for buyers to distinguish genuine AI-native architecture from retrofitted capability. Jewellery retail presents specific forecasting challenges: relatively small unit volumes per SKU, high product value, long replenishment lead times and significant localisation requirements across geographically dispersed store networks.
The move away from spreadsheet-based planning is a recurring theme across mid-market specialty retailers, where planning teams frequently manage assortments using tools that were never designed for the purpose. Exception-based workflows, where planners review only those items the system flags as anomalous, are increasingly offered as a productivity metric alongside forecast accuracy.
Regulatory and standards read-across
Because Impact Analytics processes transaction and inventory data across three jurisdictions, the deployment will need to satisfy data-handling requirements under Australia's Privacy Act, New Zealand's Privacy Act 2020 and Canada's PIPEDA, with the latter currently undergoing reform under Bill C-27. None of those obligations were addressed in the release, and Michael Hill's data-residency arrangements with the vendor are not publicly disclosed.
The broader enterprise AI procurement context is also shifting: the EU AI Act's requirements for high-risk AI systems do not apply here, but both UK and Australian regulators have signalled interest in transparency obligations for algorithmic decision-making tools used in commercial settings. For now, retail planning AI sits outside mandatory disclosure regimes, but enterprise buyers are increasingly asking vendors for explainability documentation as part of their own governance frameworks.