18 hours of manual reporting eliminated. Leadership finally flying with instruments.

The situation

Summit Analytics Group is a Calgary-based professional services firm with 28 consultants across four practice areas. They advise mid-market companies on operations, finance, and technology — and their own internal operations were a contradiction of everything they told clients.

Every Monday, Summit's operations manager spent the entire morning pulling data from four systems — their project management tool, time-tracking software, accounting platform, and CRM — copying numbers into a master spreadsheet, and distributing a PDF report to the leadership team. By the time the report landed in inboxes, it was based on data from the previous Friday. By Wednesday, it was already stale.

Leadership was making staffing and billing decisions on week-old data. They had no visibility into which consultants were under-utilized, which projects were running over budget, or which clients were at risk of churning — until the next Monday report told them about it retroactively.

The diagnostic

Before building anything, we mapped the full reporting workflow to understand exactly where time was being spent and what decisions the reports were meant to support.

The 18 weekly hours broke down as:

• 6 hours pulling and reconciling data from four systems (many with no native integration)
• 4 hours cleaning the data — fixing duplicate client names, correcting miscategorised time entries, reconciling invoices
• 3 hours formatting the Excel report and creating the PDF version
• 3 hours answering follow-up questions from leadership about the report data
• 2 hours ad-hoc requests that couldn't wait until the next Monday

The underlying data was actually in good shape — the problem was entirely in the manual assembly process. The right fix was not a better spreadsheet. It was eliminating the spreadsheet entirely.

What we built

The data layer: one source of truth

We connected Summit's four systems — Teamwork (project management), Harvest (time tracking), Xero (accounting), and HubSpot (CRM) — via a lightweight ETL pipeline into a single structured data store. Data refreshes every 15 minutes. Every metric the leadership team needs is derived from this single source, with one agreed definition, with no manual intervention.

We also built a data quality layer that automatically flags anomalies — time entries that don't match a project, invoices without corresponding bookings, clients with open projects but no recent time logged. The operations manager now reviews a 10-minute anomaly digest instead of spending four hours on data reconciliation.

The dashboard: designed for decisions, not data

We built three dashboard views — one for each audience:

Automated weekly digest

We replaced the Monday morning report with an automated email digest — generated every Monday at 7am, sent to the leadership team before they sit down. It pulls the previous week's key movements: revenue vs. target, utilization change, new pipeline, flagged risks. The ops manager no longer writes it. It writes itself.

Before and after

The tech stack

Results at 3 months

• Manual reporting hours: zero (down from 18 hours/week)
• Data freshness: real-time (15-minute refresh) vs. 7-day lag
• Consultant utilization visibility: same-day instead of week-old
• At-risk clients surfaced proactively: 3 in first month, all retained
• Leadership decision speed: 3× faster on staffing and billing decisions
• Ops manager time on reporting: 10 min/week (anomaly review) vs. 18 hrs