Case study · the only engagement we can show

Eight tools into one system — and the problem it didn't solve.

Quagga Designs got working infrastructure and closed anyway. Both halves matter, so both are here.

ClientQuagga Designs — DTC furniture, US & Canada

FounderCarl Heinrichs

StackWebflow + Foxy.io + Ottawa Logistics (3PL)

EngagementOne year

OutcomeInfrastructure delivered · brand closed

This is the only engagement AGIT can show, and it is not a success story. Presenting it as one would fall apart on the first question. What follows is what was built, what it measurably changed, and what it failed to change. Every figure is Carl's own, published under his name.

The situation

Quagga was spending on paid traffic without a system underneath it.

Attribution unclear — no reliable answer to which channel drove revenue
Eight-plus tools that didn't talk to each other
Inventory updated by hand
Order-to-fulfillment split across three separate systems, none automated
Founder's day dominated by operational questions

“We were spending on paid traffic without a solid system underneath it. Attribution was unclear, operations were manual, and our tool stack was expensive and disconnected.”

Carl Heinrichs

What was built

Custom admin panel	All business data in one interface instead of tab-hopping between eight tools
Order-to-fulfillment automation	Foxy.io → inventory → 3PL, running without anyone re-keying it
Real-time dashboards	Margin and inventory visible as they change, not reconstructed weekly
Attribution layer	Which channel actually drove a sale, instead of three systems each claiming it
Stack consolidation	Eight-plus subscriptions replaced by one owned system

Carl kept the code, the data and the accounts. Ownership was the point, not a closing argument.

Quagga admin panel, orders view: order ID, date, customer, country, fulfillment status and total in one table. Customer data masked. — The admin panel, orders view — one interface replacing eight tools. Shipped and Sent-to-3PL statuses come from the automated pipeline, not from anyone updating a sheet. Customer names and totals masked.

Quagga overview dashboard: revenue, orders, AOV, ad spend, ROAS and CAC tiles, revenue and sales dynamics charts, and performance by channel. Figures masked. — The overview dashboard — revenue, orders, AOV, ad spend, ROAS, CAC and channel attribution on one screen, updating as the business ran. Figures masked at the client's request.

System monitor: n8n, Klaviyo, FoxyCart and database status with latencies, and nine active n8n workflows covering 3PL inventory sync, order status, place order, abandoned cart and ad spend sync. — The automation layer itself — nine workflows moving orders between the store, inventory and the 3PL, with integration health visible in the same place. This is what “automated” means in the table above.

What it measurably changed

Result	Figure	Source
Annual SaaS cost eliminated	$4,646	Carl's accounting
Manual work automated	15+ h/week	Carl's estimate
Fulfillment pipeline	fully automated	Carl's recommendation
Business visibility	real-time	Carl's recommendation

“The results were concrete: $4,646 in annual SaaS savings, a fully automated fulfillment pipeline, and a dashboard that gives me real visibility into the business for the first time.”

Carl Heinrichs · LinkedIn recommendation · 17 June 2026

What it did not change

The brand closed. Three forces were working against Quagga at once:

Tariffs closed their US market — the majority of addressable demand
Labour at $28–32 an hour in Canada, against competitors paying a fraction
Eleven consecutive months of negative margin before the automation existed

Operations were one problem out of three. The system removed manual work and software cost. It could not restore a market closed by trade policy, and it could not change what an hour of labour costs in Ontario.

On payroll. No one was let go and no payroll line went down. The team was positioned for a shift to part-time that never happened, because the business ran out of runway first. What the system removed was manual work and software cost — that part is measurable, and it's above.

What this proves, and what it doesn't

Proves

The mechanical layer of a DTC operation is codeable, and consolidating it works
Ownership transfers cleanly — code, data, accounts
A founder can get real visibility without buying another subscription

Does not prove

That automation changes a company's trajectory
That payroll cost comes down
That the model works at scale — this is one engagement

What changed in how AGIT works because of this

Dependency is a bad strategy

Carl had a panel he couldn't maintain himself. When he stopped paying, the dependency protected nothing — he closed anyway. Every build since is owned by the client from day one and maintainable without us.

Automation is not a rescue

A brand with a closed market and a cost structure it can't survive doesn't need a dashboard. Qualification now exists partly to identify that case and decline it, rather than take the money and deliver a system into a business that won't be there in a year.

A guarantee has to be measured in something the client verifies

Quagga had no baseline. Nobody wrote down what the processes cost before the work started, so “it's better now” was the only available conclusion. Every engagement now records a signed baseline in week one and measures the same processes at day ninety.

Scope has to be bounded

One year on one client with no written blueprint is how a project becomes open-ended. Six weeks now means a signed blueprint at the end of week two, and anything outside it is a change request.

Verify any of this Carl's LinkedIn recommendation →
Google review →

Carl is reachable and the brand's closure is public. If any figure here doesn't hold up, it should be easy to find out — that's the point of putting it in writing.