Savepoint Analytics

Today, I took the plunge hoping the waters are deep enough. Now I have everything. I purchased and received my business license for Savepoint Analytics, purchased the domain names savepointanalytics.com, and savepointanalytics.ai (safeguarding for later use). Savepoint is a solo-consultancy aimed at providing key services and products to game companies. It will be the gateway the the game-analytics platform that I’m building from scratch.

It’s an interesting time in my life. For one, I’m a father for a 20 month old boy whose the world to me. That alone made me realize how precarious the game industry is at this point in time. It also made me realize that what you work on, and who you work for matters. Relationships matter. Not every job is a blessing. All roles require time and energy to be put in. Some forms of work and workplaces are rewarding in their own right because the sense of accomplishment and knowing you helped build something impactful (or just incredibly fun) for players is important. Right now for me, however, it means taking the time to kick a soccerball around with my son and enjoy this time I have with him.

I see the monetary promise already being a SaaS provider in the post Agentic AI world.

I’d like to grow it, and I’m equally aware there are only so many new studios each year. I’m also aware that there are a lot of companies already doing this and its becoming a crowded field with no shortage of good competition. Recognizing the potential and the harsh reality, I realize that it will be hard to use this consultancy to make a living. What it does provides however is a way to stay hands-on and up to date with the craft I spent a long time honing. The products I am trying to build and deploy will be tools. The platform will be my multitool swiss army knife that I can take with me for future use. So it isn’t an exit plan; it’s a foothold in a field I love.

I’m not leaving Vancouver Island for sometime now. I want my son to grow up here, or somewhere where he can truly enjoy the outdoors. Therefore, that limits my employment options. Hopefully though, a consultancy like Savepoint will be able to allow me to stay on the island even if I am in between working with studios and publishers. If and when I join another game company, Savepoint will probably take a back seat (I’m glad to keep that boundary clean and in writing).

Why a Github code-based data platform?

Quick to deploy and grow

I got my start in the gaming industry in Victoria, BC. Back in 2014, the city had a healthy and thriving ecosystem of indie game companies focusing on mobile free-to-play (F2P) titles. And the story was all the same:

• The game was built with care and attention to quality.
• The analytics was secondary to the primary goal: making a fun game!
• Then came pouring money in marketing the game.
• Initial KPIs looked highly promising compared to industry standards.
• Then monetization would slump and there was no good way to understand why.
• Analysts would be hired after the fact with whatever funds they could gather to figure out the why in order to reverse the trend.

This is how I got my start. I was a very junior analyst. Suffice to say, I helped but my help didn’t reverse the fortunes of the first game I ever worked.

The way I put it on the Savepoint site is that it’s the analytics foundation most studios stand up too late — built from first principles, and ready before your first player. It’s shaped by a decade of senior work in live-service games and built entirely from scratch as Savepoint’s own tooling, so the day your game launches the numbers are already trustworthy: not reconstructed six months and one crisis later.

A rule I believe in, right up until I don’t

I am a believer in the 80/20 rule. Do the first 80% of the work and you often have 95% of the answer; the remaining 20% tends to buy diminishing returns. When you are in the thick of it, that is a sound way to spend a finite week. Most of the time, it is exactly right.

But carry that logic onto everything and it stops being a shortcut and starts being a barrier — and in a few places it does something worse than waste effort. It hands you a confident, clean-looking answer that is quietly wrong.

Where “good enough” quietly becomes wrong

There is a specific class of game-data problems where the last 20% is not polish — it is the answer. Cut it and the number you ship is not slightly rough; it is misleading:

• Revenue attribution and economy ledger-keeping — where a dropped event or a loose join silently moves money between sources and sinks.
• Adversarial and bad-actor detection — cheaters, payment abuse, and fraud live precisely in the tail you rounded off.
• Experiment assignment — a small leak in who-saw-what contaminates every result downstream.
• Identity resolution — the merges you skip become double-counted players and inflated LTV.
• UA attribution — the ambiguous cases are exactly the ones that decide whether a channel is profitable.
• Regulatory and financial reporting — loot-box probability disclosure, minor playtime limits, COPPA age-gating, GDPR deletion cascades, reported earnings. “Approximately right” is not a defensible posture here.
• Match completeness — incomplete match records quietly corrupt skill ratings, leaderboards, matchmaking quality, and competitive balance.

Some of this is about how events are recorded. Some is about how the data model uses them. Both have to be right, and in these domains “mostly” is not a passing grade.

Why studios end up here anyway

None of this is a knock on studios. The sensible ones build real infrastructure — S3, Glue, and Athena, or Snowflake; pipelines orchestrated with Prefect, Dagster, or Airflow; modeling in SQL and Python with dbt. That stack is entirely capable of meeting the bar and staying accurate.

The trouble is that the 80/20 instinct rides along inside it. Under deadline pressure, the hard 20% of each of these problems is the part that gets deferred — and it is a genuinely large amount of careful work to build and maintain. The pieces that matter most are the pieces most likely to be left for later. Later rarely comes.

Building for the 20% that counts

That gap is the entire reason Savepoint exists. I am building the telemetry contracts, the ledgers, the identity and attribution logic, and the metric definitions from scratch — not to be exhaustive everywhere, but to be correct where correctness is the only acceptable answer. The 80/20 rule still governs the rest; it should. The point is knowing which problems are the exception, and refusing to round them off.

“We’re building something here… and all the pieces matter.” That is the right posture for exactly these corners of a game’s data — and the wrong posture almost everywhere else. The skill is telling the two apart.

The specific corners Savepoint builds to get exactly right — and the ones it treats as rigorous-but-never-certain — are laid out on the Savepoint site in the metrics we refuse to get wrong. This post is the “why” behind that list.

Where’s it at

At the time of this writing, it’s not finished. Each module carries an honest maturity label, and several are still prototypes. The labels are there so you can tell the difference — which is, come to think of it, the whole point of everything below.

Why I’m the one building it

I do my best work when I’m learning something new, inside a team building something I actually care about. Games are where I’ve spent my career and where I’d most like to stay. The industry is turbulent at the moment, and I’m clear-eyed that the right role might sit adjacent to it rather than squarely within — but either way I’m drawn to hard, novel problems and to teams that would rather get the answer right than merely get it quickly.

Titles matter to me less than the work and the people. I once moved from a Data Scientist title to a Product Analyst one on purpose, because the Product Analyst seat was where the interesting science actually lived and the team delivered exactly what they’d promised. I’d make that trade again. I optimize for impact and craft, not a line on an org chart.

Ten years in has taught me the difference between the environments where I thrive and the ones where I flounder. I thrive where analytics is tied to real decisions, where the last stubborn 20% of the work is respected when accuracy genuinely matters, and where analysts are trusted to say plainly what the data does and does not support. Savepoint is, in a sense, me building that environment on purpose — getting the load-bearing corners right up front, so that when the game ships, the hard 20% is already done and nobody has to discover, mid-crisis, that it was skipped.

That is also why I apply selectively. When I reach out to a studio, it’s because I value the game and the team, not because a listing happened to be open. And it’s why the Savepoint modules wear their maturity labels honestly rather than claiming a blanket “100%.” Some things I can promise are provably complete and reconciled. Others are rigorous and honest about their uncertainty. Pretending there’s no difference would be its own kind of rounding error — and this whole thing is a long argument against those.