82 Games, 82 Manual Reports: How I Automated an NHL Scouting Workflow

I spent 4 years playing college golf. Every week was the same cycle: practice, travel, compete, review, repeat. Somewhere around junior year, I realized the best players weren't the ones with the most natural talent. They were the ones who eliminated the most wasted motion -- in their swing and in their preparation.

That lesson followed me from the golf course to an NHL video room.

The Setup

After college, I landed a data analytics role with the Florida Panthers. My job involved turning raw game data and video footage into actionable scouting intelligence for the coaching staff. Before every game, they needed a scouting report on the opponent: tendencies, play patterns, key matchups.

An NHL season is 82 regular-season games. For each one, the process was the same:

1. Pull video clips from the opponent's recent games
2. Tag each clip by play type, zone, and outcome
3. Export the tagged data into a structured format
4. Aggregate the data to identify patterns and tendencies
5. Build a formatted report the coaches could review before morning skate

Each report took the better part of a day. By game 40 in a season, I was running on fumes.

The Breaking Point

It was a road trip. Late at night in the video room. I'd done this exact process dozens of times that season. Same steps, same format, same output. The only thing that changed was the opponent.

The quality of my work at game 60 was nowhere near the quality at game 5. Not because I got worse at analysis. Because the manual work was draining all the energy I needed for the thinking work.

It hit me the same way it hit me on the golf course: the repetitive stuff isn't just boring. It's the thing that drains the energy you need for the work that actually matters.

The Automation

I wrote a Python script that night. It was rough, barely functional, and held together with duct tape. But it automated the most painful parts of the workflow:

• Data pulling -- automated queries to extract play-by-play data and clip metadata
• Tagging and categorization -- rules-based classification of play types, zones, and outcomes
• Aggregation -- statistical summaries, tendency breakdowns, and pattern detection
• Report formatting -- structured output that matched the format the coaching staff expected

By the end of the season, the pipeline ran in under 2 minutes. What used to take half a day was reduced to a process I could trigger while walking to get coffee.

What This Has to Do With Your Business

You probably don't need an NHL scouting report generator. But I'd bet you have your own version of the same problem.

Every business has processes that look like this:

• Pull data from one system, reformat it, enter it into another system
• Build the same report every week by copying numbers from 3 different dashboards
• Send follow-up emails by checking a spreadsheet and remembering who you haven't contacted
• Onboard new clients by manually creating folders, sending welcome emails, and setting up project boards

Each one of these is a pipeline. Trigger, process, output. The same pattern I automated for scouting reports.

The Pattern Behind Every Automation

Whether it's a scouting report, a lead follow-up sequence, or a weekly digest email, every automation I've built follows the same structure:

1. Trigger

Something starts the process. A new form submission. A daily schedule. A status change in your CRM. An overdue invoice hitting day 3.

2. Process

The system does the repetitive work. Pull data, transform it, classify it, route it, format it. This is where the time savings happen. A human doing 45 minutes of data entry becomes a pipeline doing it in 30 seconds.

3. Output

The result goes where it needs to go. A formatted email. A Slack notification. A PDF report. A CRM record. A task in your project management tool. The right information, in the right format, at the right time.

From Sports Analytics to Business Automation

When I left the Panthers, I took two things with me: the technical skills to build data pipelines, and the hard-won understanding that eliminating manual work isn't laziness -- it's how you protect the quality of your real work.

My golf analytics company (ShotLab) runs on the same automation philosophy. Lead capture, content publishing, client reporting -- all automated. When I started AP Workflow, I applied the same approach: build the automation first, then use the time it saves to do the work that actually grows the business.

The first workflow I built for AP Workflow was a lead capture pipeline. Someone fills out the interest form on my website, and within 30 seconds: a CRM record is created, a personalized follow-up email goes out, and I get a Slack notification with a lead score. No spreadsheets. No manual data entry. No leads falling through the cracks.

That's the same pattern as the scouting reports. Different context, same principle: automate the pipeline, protect the thinking time.

What 10-20 Hours Per Week Looks Like

Most businesses I audit have 10-20 hours per week of automatable work. That's not a guess -- it comes from mapping actual processes during an automation audit.

10-20 hours per week is 1-2 full workdays. Every week. Being spent on tasks that a system could handle in seconds.

At $30-$60/hour fully loaded (the typical range for the people doing this work), that's $1,200-$4,800 per month in labor spent on work that doesn't need a human brain.

The automation doesn't cost $1,200-$4,800 per month to run. It costs $5-$20/month for server hosting. The rest is a one-time build investment that pays for itself within weeks.