SolutionsGrowth Marketing

Attribution is not one problem, which is why no tool has ever solved it.

Every tool promises you the real number, and there is not one. Different growth questions need different evidence, so we build the measurement that matches each question to the method that can actually answer it, and we are honest about how much confidence each answer carries.

The playbook

Each piece settles a decision your team already faces every week.

Together they make one measurement system your people and your AI both read from. You do not need all of it, and almost nobody starts with more than two.

Incrementality & media mix

First-party journeys cross-checked with MMM and geo tests, so you know which spend adds growth and which just takes credit for it. The same numbers hold up in a board deck or in diligence.

Unified spend source of truth

Automated pipelines deliver one trusted spend number, finalized every Monday by 11am ET. Budget meetings argue about the plan, not the data.

Creative asset analytics

See how creative choices shift quality through the funnel: emotional versus problem-led, branded versus prospecting. Fatigued angles get killed before they burn budget.

Forecast & budgeting

Scenario models that tie CAC targets to LTV, so you know when to press spend and when to hold.

Funnel & experience optimization

Landing pages and offers designed by funnel stage, so top-of-funnel clicks become engaged prospects instead of bounce statistics.

Pricing & promo modeling

Simulate volume, margin, and timing tradeoffs before you commit, so promotions stop being expensive guesses.

Predictive LTV & customer scoring

Know which new customers will still be worth it in month twelve, and buy those instead of averages.

Automated reporting pipelines

Dashboards built for decisions, not decoration. Refreshed automatically from governed definitions, so the number in the meeting is the number in the warehouse.

The part the pitch decks skip

Every method we use has a failure mode, and we will tell you what it is.

MMM

Needs real variance in spend to learn from, and will fit a confident-looking curve to noise if you let it. Refresh cadence and honest priors matter more than model sophistication.

Geo and lift tests

The closest thing to causal truth here, and routinely oversold. A 3% point estimate with a confidence interval spanning zero to ten is not a 3% lift, and the decision usually needs lift divided by spend.

First-party journeys

See far more of the path than GA4, and they are still correlational. They show what happened before a conversion, never what would have happened without it.

Multi-touch attribution

Structurally blind to offline spend and to demand it never touched. Fine as a directional read on digital sequencing, dangerous as a budget authority.

If you do not know where a method breaks, you cannot tell a result from an artefact of the method. And a large share of what gets called an attribution problem is a decomposition problem, which no attribution tool will fix.

The reads, side by side

What each method is good for, and where it misleads.

Every brand we work with runs several of these at once, and they rarely agree. The job is knowing which one to trust for which decision.

Method

Good for

Where it misleads, and our rule

Platform ROASMeta, Google, TikTok

Creative and audience decisions inside one channel.

Every platform claims the same order. Never used to compare channels.

Last clickGA4, Shopify

Bottom-of-funnel capture, branded search.

Punishes anything upstream. A 1.5x last-click channel was one brand's best top of funnel. Read with the survey before cutting it.

Post-purchase survey

Channels pixels can't see: TV, podcast, word of mouth.

Recall skews to memorable channels. Read as a share, against the pixel.

First-party qualityyour own pixel

A leading indicator across every channel, on your yardstick.

Says who looked, not who bought. Paired with survey and MMM before a cut.

Geo tests

True incrementality for one channel, when properly powered.

A "3% lift" with a range of 0 to 10% can't set a budget. We report the range and lift per dollar, and skip underpowered tests.

MMM

Budget mix, long-cycle channels, seasonality.

Only as good as its last reconciliation. Checked against actuals every ~2 months, not twice a year by a vendor.

Governed marketing data

Everyone knows CAC by Monday morning, and you can drill from it to the ad.

Weeklyone CAC definition finance signed off, finalized Mondays before the meeting.

Drill-downCPMs, CTRs, and conversion rates decomposed to the campaign and the creative themes

Self-serveask it in Slack and get an answer that knows whether it is governed.

The definitions live in code in your warehouse, versioned and tested, so the number in a Slack answer is the number in the board pack.

How the governed AI layer works →

Why did CAC move this week?Governed layer · live

Head of GrowthWhat was CAC last week, and is there room to push Meta harder this month?

CAC $58 → $63

~80% of the move is Meta CPMs, up ~12% on the same CTR and conversion rate. Spend mix and conversion are flat.

Governed · finance definition, tested nightly

Creative-level: two fatigued prospecting angles drive most of the CPM rise. That cut is directional; it has been flagged for review before it drives the brief.

Not governed · sent for review

Illustrative example.