14 days to 24 hours.
Activation up 47 points.

Implementation managers as the bottleneck.

A 200-person B2B SaaS was losing roughly a third of new customers between contract sign and first value. Implementation took 14 days because every account needed manual config, custom rules, and three back-and-forth Slack threads with the implementation team. Implementation managers were the rate limit. Hiring three more was the next-quarter plan, until the CTO asked whether software could do this work instead.

Ship the configurator, not the wizard.

We mapped 90 days of completed implementations and found that 78% of customer setups were variations on six templates. The other 22% were genuinely custom. The plan: an AI configurator that walks customers through setup in a chat surface, applies the right template, and routes the truly custom 22% to a human. Self-serve for the bulk, white-glove for the edge cases.

The wizard wasn't the answer.

• Linear setup wizards always failed for B2B because every customer asks 'but what about my case' three steps in. A chat-shaped configurator could answer that without breaking the flow.
• Implementation managers loved it because they finally had time for the genuinely hard cases. Headcount did not shrink, scope did.
• The template clusters drift quarterly as new customer types come in. The eval pipeline catches drift before it shows up in churn.