Showmaker

01Overview

Showmaker is an events operation, a business where a great deal happens quickly and most of it generates data that nobody has time to read. My contribution was not the events; it was the layer underneath them. I built the systems that turned scattered operational activity into intelligence the team could actually use.

This case study is about that layer: the AI, the analytics, and the automation that together moved the operation from reporting on the past to acting on the present.

02Business Challenge

Operations heavy businesses drown in their own activity data. Numbers exist everywhere, across tools, sheets, and inboxes, but assembling them into a picture takes so long that by the time the picture is ready, the moment to act on it has passed.

• Reporting assembled by hand, late, and inconsistently from one week to the next.
• Decisions made on instinct because the evidence arrived after the decision was needed.
• Skilled people spending hours on collation instead of judgement.

03Discovery

I started by tracing where information actually lived and how it moved, every source, every manual hop, every place a human retyped a number from one system into another. Those manual hops are where both the delay and the errors accumulate, and they became the map for what to automate first.

04Research

The research question was narrow and practical: which decisions does this team make repeatedly, and what would each need to see to make it well and quickly? Reporting built to answer that question is used; reporting built to show everything is ignored.

• Inventory of recurring decisions and their owners.
• The minimum evidence each decision actually required.
• The cadence at which that evidence needed to arrive to still be useful.

05Strategy

The strategy was to build reporting that maintains itself. Rather than producing a better manual report, the aim was to remove the manual step entirely: pipe the sources together, let automation keep them current, and use AI to turn raw activity into readable intelligence. The team's job would shift from assembling information to acting on it.

06Execution

I built the pipeline that connected the operation's data sources into a single, current view, then layered reporting and AI assisted summarisation on top so the output was not just accurate but legible at a glance.

• Consolidated fragmented sources into one dependable dataset.
• Automated the collection and refresh that had been manual and weekly.
• Built reporting framed around the team's real decisions.

07Product Decisions

The central product decision was restraint: build the smallest system that made the important decisions faster, and resist the temptation to surface every possible metric. A dashboard that answers three questions well beats one that raises thirty.

08AI Integration

AI did the interpretation work that used to require a person: reading raw operational data and turning it into plain language summaries and signals. That's where the leverage was, not in replacing judgement, but in delivering the team a readable starting point instantly, so their judgement could begin from a summary rather than a spreadsheet.

09Automation

Automation carried the whole thing from “built once” to “runs itself.” Collection, refresh, and delivery happen without anyone touching them, which is the difference between a report that exists and one that stays true. The operational efficiency gained here is measured in hours returned to the team every week.

10Analytics & Reporting

The analytics layer translated activity into understanding: what was working, what was slipping, and where attention would pay off most. Built as decision-ready reporting rather than raw dashboards. The emphasis was always on the “so what,” not just the “what.”

11Business Results

The system changed how the operation runs day to day. The direction of travel:

12Lessons Learned

• The value of reporting is set by its latency: accurate but late is only marginally better than nothing.
• Start from the decision, not the data; it's the only way to know what to leave out.
• Automation's real product is trust: a report that maintains itself is one people actually rely on.

13Next Steps

Natural extensions would push from descriptive toward predictive, using the same consolidated data to anticipate rather than only report, and to widen the AI layer so more of the interpretation, and eventually the recommendation, happens automatically.