Here's a fun problem. You ship a free tool. People actually use it. And the data behind it can change any time 50 different state governments feel like it.
That's the reality of running https://statefeecalculator.com, a free tool that shows car registration fees and taxes for every US state. The build was the fun part. Keeping the numbers right is the actual job. Here's how I do it.
The data starts boring
There's no API for DMV fees. I wish there were.
Every state's numbers started the same way. I went to the official DMV or tax agency site, found the fee schedule, and wrote the numbers down. Some states publish a clean table. Some bury it in a PDF from 2019. Some make you read the actual legislation.
Each state became one record in a structured dataset. Base registration fee. Any extra taxes. EV surcharges. Notes on what's weird about that state. That structure matters, because it makes the next part possible.
Checking beats collecting
Collecting the data once is a weekend project. Trusting it a year later is the hard part.
So I built a routine. Every quarter, I re-check every state against its official source. I diff the new numbers against what's live. Anything that moved gets flagged and updated. Between those reviews, I watch for news about fee changes, because legislatures love passing a fee hike that takes effect mid-year.
Boring? Yes. Necessary? Also yes. A free tool lives or dies on trust. If someone checks my number against their renewal notice and it doesn't match, they're gone. Stale data is worse than no data.
What breaks
Plenty. Here are the classics.
Fee schedules move. A state redesigns its DMV site and the fee page I bookmarked now returns a 404. The numbers didn't change, but my source link did, and I have to hunt the new one down.
Fees change with no fanfare. A state bumps its registration fee by $10 in a budget bill and doesn't announce it anywhere a normal person would look. I only catch it on the next review pass.
Surcharges get renamed. A "transportation fee" becomes an "infrastructure fee" with a different amount. Same money, new label, and my dataset needs the update.
The quirks that surprised me
Two things keep catching people off guard.
First, some states aren't one number. Counties and even cities can stack their own add-ons on top of the state fee. Two people in the same state can pay different totals depending on their county. The state fee is only the start of the bill.
Second, states can't agree on what a fee is based on. Some charge a flat amount no matter what you drive. Others charge based on the car's value, its weight, or its age. A car that costs $40 to register in one state can cost hundreds in the next one over, for the exact same vehicle. Same car, same driver, wildly different bill.
The unglamorous part is the product
Nobody visits a fee calculator because the code is elegant. They visit because the number is right.
So most of my time on this project isn't coding. It's reading DMV sites, diffing numbers, and fixing the stuff that quietly broke. That routine is the product, really. The calculator is just the interface.
If you're building anything that depends on real-world data, build the refresh routine before you need it. Future you will be grateful.
And if you want to see what 50 states of fee data looks like when it's kept current, it's all here: https://statefeecalculator.com
Top comments (0)