“We're already planning further tracking data-quality improvements, given how impactful better Meta data has been for the business.”
Jeff Lai leads Growth at Proper Cloth, a brand that has spent years turning custom menswear into an online, scalable, repeatable experience. He runs the full growth motion across acquisition, retention, and experimentation as a one-person team.
At Proper Cloth’s level of Meta and Google spend, every percent of conversion data quality compounds, and every hour of delay between a bidding change and the data to read it caps how fast the business can move.
Proper Cloth partnered with Converge to fix tracking data quality going back to Meta, layer real-time attribution on top of their warehouse model, and bring a natural-language analyst into the loop with the Converge MCP.
From engineering tickets to a measurement layer that just works
Distrust in the conversion tracking was eating Jeff's calendar
Proper Cloth spends heavily on Meta. At that level of spend, the quality of the conversion data going back to the platform is paramount. It’s what determines who Meta targets, what it optimizes for, and where the next dollar lands.
But Jeff kept getting pulled into the same frustrating loop. Something would look off in Ads Manager. Confidence in the numbers would drop. Then he’d be back in Google Tag Manager, or asking engineering to help debug the setup. Instead of spending his time on growth, he was chasing tracking issues.
Result: Jeff was spending time on tracking ghosts instead of the high-leverage work like creative testing and experimentation, the things that actually move the business. It stalled the whole growth motion.
An audit and a measurement layer with best practices baked in
Converge started with a tracking audit that quickly showed where the setup was breaking down. Then it replaced the fragile pieces with a measurement layer that had the right practices built in from the start.
With Converge live, Proper Cloth sent cleaner data back to Meta. Event matching improved through server-side re-identification, more conversions made it back into the platform, and new customers were identified more reliably. In the weeks that followed, Proper Cloth doubled Meta spend with flat CPA in both Meta’s reporting and Proper Cloth’s internal attribution model.
Impact:
- Meta spend doubled while CPA stayed flat, in both Meta’s reporting and Proper Cloth’s internal model
- Conversion coverage moved from ~78% to 100%, with cleaner event matching across the board
- 13% more new customers identified for Meta’s acquisition models
From a 24-hour iteration loop to same-day bidding decisions
The internal model refreshed once a day
Proper Cloth’s internal attribution model is the source of truth for ad-buying decisions. The Data Team built it on top of their own warehouse, and it refreshes once every 24 hours.
That cadence works for monthly reporting. It doesn’t work for active bidding. Ecommerce is a real-time business, and when Jeff ships an audience tweak or a target change in the morning, the feedback he needs to evaluate it doesn’t show up until the next day.
Result: For a one-person growth team, every test costs a day before Jeff can read it. The whole iteration loop was capped at the model’s refresh rate.
A real-time attribution view on top of the warehouse model
Converge sits on top of Proper Cloth’s internal model as a real-time attribution layer. Same campaigns, same touchpoint logic, but the numbers refresh continuously through the day.
Jeff ships a bidding change in the morning, reads pacing by mid-afternoon, and calls the next move before the end of the day. The warehouse model audits overnight. The team stopped waiting on it to decide and started waiting on it to confirm. That compressed loop let Proper Cloth scale non-brand Google spend 2.4x with new-customer ROAS holding steady, by optimizing bidding strategy multiple times per day.
Impact:
- Non-brand Google spend scaled 2.4x with new-customer ROAS holding
- Weekly new-customer order volume nearly doubled
- Iteration loop compressed from 24-hour to same-day
From small data questions piling up to a data analyst on hand
Growth's small data questions sat at the bottom of the company's reporting queue
Once Jeff was iterating same-day on non-brand Google, smaller operational questions came up constantly. Which attribution model actually fits a top-of-funnel push? Is mid-day CPA a reliable proxy for end-of-day CPA on this campaign, confident enough to pull channel spend back intraday? How does this campaign halo into remarketing?
Proper Cloth has a skilled data analyst, but that analyst’s queue is business-wide reporting. Growth’s smaller questions either waited their turn, got skipped and decisions got made on gut, or Jeff burned a half-day in spreadsheets pulling them himself.
Result: Either the analysis fell behind the decisions, or Jeff spent the time the new iteration loop was supposed to give back.
The Converge MCP as a data analyst on hand
The Converge MCP exposes Proper Cloth’s full attribution and conversion data through Claude. Jeff asks a question in natural language; the MCP runs the queries against Converge’s data layer and hands back the answer in minutes.
With a data analyst effectively on hand, those questions were suddenly within reach. The MCP led Jeff to the best-fit attribution model for measuring his top-of-funnel Google campaigns in Converge, landing on inverse-J-shaped paid as the daily working view. And it let him validate whether mid-day CPA was a reliable proxy for end-of-day CPA on a given campaign. When it was, he could pull total channel spend back intraday instead of waiting for the day to close.
Impact:
- Inverse-J-shaped paid identified as the best-fit attribution model for top-of-funnel Google
- Mid-day CPA validated as a proxy for end-of-day CPA, enabling intraday channel spend pullbacks
- Intelligence at his fingertips, giving Jeff even more leverage as a one-person growth team