I read every word on the Hyros Shopify page last week. One line stopped me.

I kept reading. Further down the page:

Pinnacle. Patent-pending. AI. Three heavy words stacked on a fourth — tracking — that's existed longer than most of the tools you use.

Here's what the tracking actually does. A JavaScript snippet sits on the page. It listens for form submissions, reads UTM parameters off the URL, captures click IDs from Facebook and Google, drops a cookie, fingerprints the device, and sends the result to a server. Every attribution tool does it. Your own engineer could write it in an afternoon.

The gap between "patent pending Print Tracking with AI" and "form listener that reads URL parameters" is the entire business.

As the founder of an ad agency managing paid media for D2C brands — and a YouTube research tool used by 100,000+ marketers — I've spent close to a decade watching every attribution tool in this category get installed and then quietly disappoint everyone. I'm writing this because the language keeps getting thicker, and I'm tired of it.

Strip the language away and attribution is two pieces.

One: the tracking. A JavaScript snippet sits on your pages and in your checkout. It reads where the visitor came from (UTM tags, click IDs), captures identifying information (email, cookie, fingerprint), and sends it to a server. When the visitor buys, the server joins the visit to the order. Every tool does this. The differences are real but small: which identifiers the script captures, how it handles iOS 14.5, how it deduplicates across devices.

Two: the weighting. Once you have touchpoints joined to orders, you assign credit. First-touch gives 100% to the first ad the customer saw. Last-touch gives it to the last. Linear splits evenly across all touchpoints. Time-decay shifts weight toward recency. Position-based gives more to first and last than to the middle. Five models. All arithmetic. Pick one. A formula runs. A number shows up.

That's the whole thing. Collect signals. Apply weights.

Now look at the homepages.

Hyros: "The best data on earth." "Proven to increase AD ROI by at least 15%." "Teams scaling spend 43%+ in 6 months with HYROS." Over on the Shopify page: "HYROS tracks 20–50% more Shopify sales."

Triple Whale: "AI that actually works for ecommerce — this isn't ChatGPT with your data bolted on." You only write that sentence when customers keep saying exactly that.

Northbeam: "Northbeam's Clicks + Deterministic Views is the world's first deterministic view-through attribution model."

Read them carefully and try to pin down what you're actually being promised.

"Best data on earth" — measured how?

"20–50% more Shopify sales" — more than what, counted by whom?

"First deterministic view-through" — first by which definition, deterministic compared to what?

None of those sentences are making a claim you could check. They're there to sound impressive, not to be measured against.

Underneath every one of them is the same script doing the same thing. Form listener. UTM parser. Click IDs. A formula. The costume changes. The code doesn't.

The complexity isn't there to make the tool better. It's there to make the price feel earned. A sales team can't defend $1,500 a month against a two-paragraph explanation. So the explanation has to be thick enough that the price stops raising questions. The mystery IS the product.

You don't have to take my word for it.

Triple Whale's own help center admits the chatbot "may invent metric definitions, feature names, or API endpoints that sound plausible." That's the vendor warning you, in writing, that their AI hallucinates.

A Hyros customer on Trustpilot: "I have been using Hyros for 2 years now and it only worked for us for the first 6 months. And then it started over reporting the leads data. It over reports leads by 20-30%." Not a bug anyone fixed — a pattern the customer lived with for a year and a half.

Or this review of Triple Whale, from a paying customer: "Daily revenue totals are wrong, entire order blocks are missing, and every week we have to open new support tickets just to get our numbers halfway close to what our channel actually reports." That's not a Moby bug or a pixel misfire. That's the core product.

I've watched this pattern for eight years. Pick any Monday morning — panicked Slack message from a client before 10am about ROAS, three tabs open (Meta, Shopify, their attribution tool), all three telling a different story. Every operator I know has sat through some version of this, done the math on a calculator, shrugged, and gone with gut. The attribution tool was supposed to end that ritual. It never did. It just added a fourth tab with a fourth number.

When iOS 14 rolled out, reported numbers broke across every account our agency was running — not because anything about those businesses changed overnight, but because Meta could suddenly see less of what was happening. If a privacy change at Apple can move the headline number across an entire book of business, the number was never measuring the business. It was measuring what the platform still had permission to track.

The customers have figured it out. The vendors' own docs confirm it. The only people still in denial are the sales reps.

Attribution gives direction, not certainty. Anyone selling you more is bluffing — the dashboard just makes the bluff look official.

The value isn't in precision. Precision is mostly theater.

The value is in transparency. Can you see the model? Can you read the rules? When a number looks off, can you figure out why without filing a support ticket and waiting three weeks for an answer?

If the answer to any of those is no, you don't have a tool. You have a very expensive window into someone else's dataset.

I got tired of recommending tools that didn't live up to their homepages. So we built one that does.

I got tired of recommending tools that didn't live up to their homepages. So we built one that does.

Bratrax Lite goes live May 12. Pre-built attribution and analytics for Shopify D2C brands. All five attribution models live in a config file you can actually open, see exactly how each number is calculated, fully transparent.

We don't believe in walled gardens. So instead of a proprietary chatbot, we ship an endpoint — ask Claude or ChatGPT anything about your ad spend, using your own AI subscription and your own API key. Your data, your tools, your call.

Flat $79/month for the first 100 founding members, locked for life. No GMV tiers that penalize you for growing. No contract. 30-day money-back guarantee.

I know this reads like a pitch. It isn't. We're trying to ship a tool that works at a fair price — because that's what this category deserves. The vendors charging $1,500 a month built bloated companies and are billing you to keep them running. We didn't. So you don't.

Before the official launch, we're opening 10 free beta spots — a full month of unrestricted access, on us.

If you run a D2C brand on Shopify, you spend on Facebook and/or Google, and you're frustrated with what you're using now, hit reply to this email by April 24, and I will personally get back to your with details.

First 10 qualifying replies get in. After that, the waitlist opens for founding member pricing.

- Brat