Can data tell you when to surf?

I surf. Not impressively, but enough that I check the forecast the way other people check the news. And somewhere between the buoy readings and the wind apps it hit me: every surfer is already running a small data platform in their head. They pull numbers from three or four messy sources, clean them up on the fly, and boil it all down to one decision. Go, or make coffee and wait.

That is also, honestly, the entire job I do for companies. So in this post I want to build a surf forecast in the open, because it is the friendliest way I know to show what a data platform actually is. No jargon required. If you have ever planned a beach day, you already understand the hard parts.

One question, three messy sources

The question sounds simple: when do I paddle out this week? To answer it you need three ingredients.

First, wave height. Out at sea there are buoys that measure the swell rolling underneath them. Along Europe's Atlantic coast these are run by national agencies, like Portugal's Instituto Hidrográfico and Spain's Puertos del Estado, and they publish their readings openly all day long. Second, wind, because wind is what turns a promising swell into either clean walls or churned-up chop. Forecast services like Open-Meteo offer a free marine and weather API for exactly this. Third, the tide table, which for many spots decides whether the wave breaks at all, and which often lives in a PDF from the local harbour authority.

Three sources, and none of them agree on anything. The buoy reports in feet, the wind forecast in km/h or knots depending on where you ask. One timestamps in UTC, another in local time. The buoy occasionally goes silent for an hour. The forecast quietly revises itself four times a day.

If you have ever tried to combine two spreadsheets from two departments, this mess will feel familiar. It is not a surf problem. It is what data from more than one place always looks like. So here is what a small data platform does about it, in three moves.

Move one: keep everything, exactly as it arrived

Before fixing anything, you copy every reading into one place, exactly as it came in, and you stamp each copy with when it arrived and where it came from. Feet stay feet. Gaps stay gaps. The weird reading from Tuesday night stays weird.

This feels lazy and it is the most important habit in the whole system. Untouched originals are your receipts. When a number looks wrong three steps later, you can walk back and see precisely what the buoy said, rather than arguing with your own edits. Engineers call this the raw or bronze layer. I call it the shoebox of receipts, and the rule is simple: nothing gets thrown away, and nothing gets "helpfully" corrected on the way in.

Move two: clean it once, in one place

Now, in a second copy, you make everything agree. Feet become metres. Knots become km/h. Every timestamp lands in one timezone. Duplicate readings get dropped, silent hours get marked as missing instead of pretended away. What comes out is one tidy table per spot: time, wave height, wind speed, tide.

The word that matters here is once. Without this step, every person cleans the same numbers again in their own spreadsheet, slightly differently, forever. With it, there is a single trustworthy version everyone reads from. This is the silver layer, and it is where most of the honest work lives.

Here is what a cleaned-up week looks like for Model Point, a surf spot I made up for this post. The numbers are invented but believable: a small early week, a swell arriving Wednesday, peaking into Friday, then easing off.

And the wind for the same week. Calm mornings, an afternoon sea breeze most days, and a genuinely windy Thursday as the weather system that brought the swell moves through.

One small craft note, because it teaches a real lesson: these are two separate charts on purpose. Metres and km/h live on completely different scales, and squeezing them onto one chart with two y-axes lets you stretch either scale until the lines tell whatever story you fancy. Two honest charts, stacked, beat one clever chart every time.

Move three: one number you can act on

Two clean charts are nice, but at 6am with a wetsuit in one hand you do not want charts. You want an answer. So the last move combines wave and wind into one score per morning: waves in the sweet range for the spot, wind light. The two multiply rather than add, because both have to be good. A perfect swell in howling onshore wind is still a no.

Read it and the week snaps into focus. Friday is the day. And notice Thursday: the biggest waves of the week, and a score near zero, because it is blown to pieces. The raw data says Thursday is the most impressive day. The decision layer says Friday is the best one. That gap between impressive and useful is exactly what this final step exists to close. Engineers call it the gold layer: the small, opinionated view that answers the actual question.

The whole platform, in one breath

That is it. Keep everything as it arrived, so you can always check your receipts. Clean it once, in one place, so everyone reads the same numbers. Then boil it down to the few views that answer real questions. In the trade this pattern is called a medallion architecture, bronze to silver to gold, and almost everything else you hear about, the lakehouses and warehouses and pipelines and dbt models, is scaffolding for doing these same three moves at a much bigger scale without losing your mind.

And a business is exactly the same platform: replace waves with customers, wind with costs, and "when do I paddle out" with whatever decision you keep making by gut feel.

If your company's forecast still lives in seventeen spreadsheets and one person's head, book a free 30 minute intake call and we will find your Friday.