Accelerator № 01 — Marketing Profit
Your ads bid on revenue. Make them bid on profit.
Meta and Google grade their own homework, then bid on the grade they gave themselves. We capture first-party conversion signal, model true profit per deal against your books, and feed those profit values back into the platforms in real time — so the algorithms optimize for margin instead of revenue.
Live in six weeks · From $4,000/mo · Your warehouse, your models
The Problem
Every platform says it's working. Your P&L disagrees.
Self-Graded Homework
Meta and Google attribute revenue using their own pixels and their own models, then hand you the report card. Of course it looks good.
Two Versions of the Truth
Marketing's dashboard says one thing. Finance's books say another. When the numbers don't reconcile, decisions fall back to gut feel.
Revenue Isn't Profit
A campaign can drive sales and still lose money. Without margin modeled into every deal, ROAS is a vanity metric.
We first built this for a business spending over $2 million a year on ads — with no reliable answer to which ones made money. Now it's productized.
What You Get
One warehouse. One version of the truth.
The process is productized. The models aren't — they're built on your funnel, your margins, your definition of profit.
First-Party Signal Capture
Conversion tracking served from your own domain, not a third-party script. It captures every click ID the platforms emit — gclid, fbclid, gbraid, wbraid, msclkid — so attribution survives the cookie loss, ad blocking, and iOS privacy controls that silently destroy 20–30% of conversion signal. We host and maintain it; the events land in your warehouse.
Every Source, One Warehouse
Managed pipelines move your ad platforms, CRM, and financial systems into BigQuery — deployed in your cloud account, under your name.
Profit Modeled Per Deal
Margin modeled into every sale, so you know what each deal actually contributed and which ad sourced it. Built in dbt, in your repo, auditable line by line.
Profit Values Pushed Back to the Platforms
The part that changes your numbers. True profit values stream back to Google, Bing, and Meta in real time, so Smart Bidding and Advantage+ optimize toward margin instead of raw revenue. Most platforms have never seen your real economics. Now they bid on them.
Decision-Ready Reporting
Dashboards built for the Monday meeting: what to cut, what to scale, what changed. No analyst required.
How It Works
Six weeks from kickoff to truth.
Connect
Weeks 1–2
Fivetran pipelines from your ad platforms and revenue systems into BigQuery. No rip-and-replace — your existing tools keep running.
Model
Weeks 3–4
We build the models joining spend to revenue to margin: your funnel, your attribution windows, your cost structure.
Report
Weeks 5–6
Dashboards go live. Your first profitability review: which ads make money, which don't, and what to do Monday.
Optimize
Ongoing
Monthly profitability reviews, pipeline monitoring, and model updates as platforms, campaigns, and channels change.
Ownership
What you own. What we license.
Yours, permanently
Your BigQuery project on your GCP billing account. The raw and modeled data. The dbt repo containing every transformation. Your dashboards and Fivetran configuration. If you leave, all of it keeps working on your historical data.
Ours, licensed for the term
The first-party collection platform and the real-time conversion export service. We build, host, monitor, and maintain them so you get enterprise-grade tracking infrastructure without building or staffing it.
No black boxes in the modeling layer. Every number on every dashboard traces to SQL you can read.
Why It Works
The platforms bid on the count, not the customer.
A lead that closes at 3% and a lead that closes at 18% both report to the platform as “1.” Smart Bidding and Advantage+ optimize toward whichever source produces more of them, so the cheaper, worse lead takes the budget — and keeps taking it, because nothing in the loop ever tells the algorithm what a lead was actually worth.
Cost per lead is the number every platform optimizes toward by default, and it is the number most likely to point at the wrong channel. Send profit per deal back instead and the same bidding engine works in your favor. At real budgets, that correction is a large number.
Shift 2% of a $3M annual budget from losing ads to winning ones. That alone covers setup, three times over.
Shift 10%. The program pays for itself several times over — every year it runs.
Arithmetic on a $3M media budget, not a forecast. What your own numbers do is what we work out together on the fit call.
Pricing
Scoped upfront. Priced like a product.
Priced against what you spend on media, on a published curve. Move the slider; the number you see is the number we quote.
Your investment
plus $20,000 one-time setup
- First year, all in
- $95,000
- Your media spend, same year
- $3,000,000
- Our share of it
- 3.2%
Above the curve
The machine is the same one described above. The contract isn't: at this level there's a security review, a negotiated SLA, and usually more than one business unit feeding the warehouse.
We'd rather quote that properly than have a slider guess at it. The published curve still sets the starting point.
Talk to us about scope →Not yet
Delivery costs about the same whether you spend $60,000 a month or $600,000, so under $100,000 the fee eats the upside. We'd rather say that here than on a call.
What we'd do at your scale: get first-party conversion tracking in place, and keep margin data in one system your team trusts. That's most of the value, and you don't need us for it. Come back when media spend clears $100k.
One-Time Setup
- First-party collection deployed on your domain, hosted by us
- Managed pipelines from every ad platform and revenue system into BigQuery, in your cloud account
- Profit-per-deal models built in dbt, in your repo
- Real-time conversion export live to Google, Bing, and Meta
- Dashboards live, with your first profitability review
Final scope quoted before we start. Fixed fee — no hourly billing.
Ongoing
- Collection and conversion-export platform hosted, monitored, and maintained
- Pipeline monitoring and fixes when platforms change their data
- Model updates as campaigns and funnels evolve
- Monthly profitability review with your team
- New channels and sources onboarded as you grow
Month to month. Your warehouse, data, and models are yours permanently. Collection and conversion feeds run on our platform under license — 90-day wind-down if you leave.
Infrastructure — Fivetran and BigQuery — bills to you directly, typically $500–$2,000 a month at this scale. No markup, no middleman. Your data and your warehouse are yours, full stop.
Built for businesses spending $100,000+ a month on paid media. Below that, the math doesn't justify it — and we'll tell you so.
Common Questions
Asked on nearly every fit call.
Which platforms do you support?
Anything with a Fivetran connector — 500+ sources. Meta, Google, TikTok, LinkedIn, Amazon, and Microsoft on the ad side; Shopify, Stripe, Salesforce, HubSpot, QuickBooks, and NetSuite on the revenue side. Running something without a connector? We build custom ones — scoped and priced separately.
We already have a warehouse.
If it's BigQuery, the timeline shortens. If it's something else, the accelerator still deploys on BigQuery and syncs results wherever your team works — one standard stack is how we deliver in six weeks at a fixed price. Your existing warehouse keeps doing its job.
Is this another attribution dashboard?
No. Attribution SaaS rents you a black box. We build transparent models in infrastructure you own — auditable down to the SQL, extensible to any question your data can answer.
Who owns what?
You own your data, your BigQuery warehouse, your dbt models, and your dashboards — permanently, in your own cloud account. The first-party collection platform and real-time conversion export run on infrastructure we build, host, and maintain, licensed to you for the term of the engagement. Raw events land continuously in your warehouse, so your data history is yours regardless. If you leave, everything you own keeps working, we run collection for 90 days while you migrate, and we'll document the path to a self-hosted alternative.