The share of Zora AI pilots that convert to paid production subscriptions fell from 35% to 20% in two quarters. How would you find the root cause?

Introduction

Zora AI is Deloitte's agentic AI platform with agents for finance, procurement, supply chain, customer and human capital work. Like most enterprise AI products, it is usually sold through a pilot: a client tests one or two agents on a limited process for 8–12 weeks, then decides whether to sign a production subscription. The pilot-to-production conversion rate is therefore one of the most important numbers for the Zora AI business. In this case it has fallen from 35% to 20% over two quarters. That is a 43% relative drop, and because pilots take months, the cause may be several months old. I'll clarify, rule out external factors, understand the journey, break down the metric, form and test hypotheses, and finish with a decision framework and resolution plan.

Step 1

Clarifying Questions (3 minutes)

• How many pilots are we talking about, and how is conversion defined?

• Why it matters: Small samples can swing widely.
• Hypothetical answer: About 60 pilots finished per quarter; conversion means a signed production subscription within 90 days of pilot end.
• Impact: A 15-point drop is about 9 fewer conversions a quarter; meaningful but needs careful segmentation.

• Has the mix of pilots changed?

• Why it matters: New agent types or industries may convert differently.
• Hypothetical answer: The number of pilots doubled after a push to "put Zora AI in every engagement", with more pilots in supply chain and human capital agents.
• Impact: Mix shift is a strong candidate.

• Did pricing, packaging or the commercial process change?

• Why it matters: Commercial changes affect conversion directly.
• Hypothetical answer: Pilots became free in most cases to accelerate adoption; previously clients paid a reduced pilot fee.
• Impact: Free pilots may attract less committed clients.

• Who runs pilots?

• Why it matters: Delivery capability affects outcomes.
• Hypothetical answer: Engagement teams, often supported by forward deployed engineers and USI delivery teams.
• Impact: Capacity or skill gaps could reduce pilot quality.