Compare Club maps its ROI-first AI strategy with Scribe Optimize

• Compare Club, a 580-person insurance comparison business in Sydney, requires every AI project to prove a 4x return — but with no workflow data, its 3-person automation team built those cases from interviews and memory, two weeks per case, on baselines that often turned out wrong.
• After UiPath's task mining was deprecated, business analyst Luke Penfold piloted Scribe Optimize on the Concierge team, work he'd already documented himself, so he could validate the output against what he knew.
• Optimize found inefficiencies the team hadn't seen: agents double-entering call dispositions in two systems, costing 76 hours over a 4-week pilot, and amendment emails being automated against a one-minute estimate that actually took five.
• Case-building fell from 2+ weeks to about 2 days (86% less analyst time), letting Luke carry 9 business cases as Compare Club's only analyst and prove post-rollout savings against a real baseline.

Compare Club's AI team has a mandate to automate repetitive work, but every project must demonstrate a 4x return before it receives approval. Without reliable workflow data, the team was building those cases from memory.

‍‍

‍

While many companies implement AI for AI's sake, Compare Club prioritizes where in the business AI will drive the most ROI. Every project its AI and automation team proposes must prove a 4x return before it moves forward, which keeps the team's limited capacity pointed at the biggest impact. But that created a catch: the team had to prove a project's impact before they could investigate the work itself. With nothing else to go on, they leaned on interviews.

Each business case started with one week of interviews and another week of reviewing screen recordings, all to produce an estimate built on memory. Then the real cost showed up. The team would invest in automating the work only to discover the baseline was wrong: 30 minutes was actually 2 hours, half the steps were never mentioned, and the ROI they'd promised couldn't be proven. Closing out a single project could mean 3 weeks of debating where the numbers came from.

The team had tried solving this with software before. The UiPath task mining capability they relied on captured the work reasonably well, but getting usable data out of it fell flat. When that capability was deprecated, Luke Penfold, the business analyst responsible for building every one of these cases, went back to interviews: the necessary evil that at least got the job done.

That put Compare Club's entire AI strategy on shaky ground: every prioritization call, business case, and measured result traced back to how well someone remembered doing their job.

Compare Club didn't go looking for a process intelligence tool. It found one inside a product the team was already using.
‍

When they stopped using UiPath, Luke turned to Scribe Capture so SMEs could document their own processes instead of walking him through them in interviews. Inside Capture, Luke discovered an AI feature called improve workflows, which suggested better ways to do the work. This got him asking what else Scribe could do. The answer was Scribe's second product, Optimize: automatic capture of  workflows, giving you a complete, data-driven view of how work gets done, no interviews required.

Luke deliberately chose to start their proof of concept with the Concierge team, as he'd captured their work before, so he knew exactly what the data should show. Would Optimize find the repetitive tasks he knew about, and follow the work across Salesforce, email, templates, and Google Chat? That would tell him whether he could take its numbers straight into a business case.

Optimize surfaced far more duplicative work than anyone realized, along with inefficiencies hiding inside what the team considered standard practice.

Luke found the biggest inefficiency fast: call dispositioning. Turns out Agents were entering the same information in two separate systems, a step nobody had questioned, and it cost the team 76 hours over a 4-week pilot. Optimize didn't just size the leak, it pointed straight to the fix: sync the two systems. It turned an invisible time drain into exactly the kind of finding his business cases are built on.

The same pattern showed up in amendment emails. Luke's team had started automating the workflow based on a one-minute estimate. Optimize showed the real number was five minutes. That gap let them rebuild the business case on real data, and prove results against reality instead of a guess.

Luke didn't stop at measurement. He connected Optimize's workflow data to Claude through Scribe's MCP. Now it pulls data straight into his Jira boards and Confluence workbooks, drafts SOPs, and builds out a business case before he walks into a meeting. And what makes it useful is what Optimize feeds it: the specialized intelligence of how Compare Club actually works, something no general-purpose AI model knows on its own.

For a team that runs on proving value first, that was the shift: Luke could finally measure instead of ask.

The analyst work behind a business case fell from at least 2 weeks to about 2 days.
‍

That capacity is the whole story for a team this size. Compare Club's AI and automation function is 3 people, and Luke is its only business analyst, currently carrying 9 business cases on his roadmap. His advice for any company trying to run an AI program without it is blunt: "You'll need a lot more business analysts."

The deeper change is what happens after a project ships. Because the original case is built on captured data rather than estimates, Compare Club can use the same baseline to measure the result after rollout. That gives the team a way to avoid the weeks of debate that unreliable starting numbers could create. Optimize doesn't just help Compare Club make the case for AI; it gives the team the evidence to show whether the result matched the case.

Every additional workflow expands Compare Club's specialized intelligence, giving Luke and the agent he built a broader base of company-specific process data to draw from with each new project. 

Compare Club used to build its AI roadmap from memory. With Optimize, it builds from evidence.