Measurement & Attribution
Proving what the brand spend actually did.
Analytics architecture, conversion tracking, cross-channel attribution and independent lift measurement. The job most companies buy last, scope smallest, and then wish they had bought first.
Where this sits in the system
Measurement & Attribution
Measurement touches every other demand job, which is why it cannot be a reporting add-on at the end of an engagement.
The specific failure: measurement is normally bought last and scoped smallest, so the seam between brand media and performance belongs to nobody. That is not a technical oversight. It is the direct consequence of buying the measurement of a channel from the vendor selling that channel.
The problem
The attribution gap, stated plainly.
Platform-reported numbers double count
Four platforms each claim the same conversion, and the total exceeds the revenue in your bank account. Nobody is lying: each platform is reporting what it saw through its own window, and no platform has a reason to tell you that another one saw the same order. Added together they produce a number that cannot be true, which is why the reconciliation is the work rather than a formality before it.
Last click understates everything upstream
Last-click reporting hands the credit to the channel closest to the transaction, which is usually branded search, which is usually the channel that did the least work to create the demand. Run a business on that view long enough and you defund the media that generated the searches, watch branded volume fall a quarter later, and conclude that search is getting more expensive.
What we do
What the work actually includes.
Foundation
- • GA4 architecture, audit and remediation
- • Conversion tracking design, server-side where it is warranted rather than by default
- • Event and data layer specification
- • Offline and CRM conversion import
Attribution
- • Cross-channel reporting that reconciles platform, analytics and revenue data
- • Attribution modeling appropriate to the business rather than whichever model is default
- • Incrementally and holdout testing where scale supports it
- • Media mix analysis for programs large enough to support it
Independent measurement
- • Brand lift studies
- • Web lift studies
- • Sales lift studies
- • Store visitation and foot traffic measurement
- • All of the above run by independent measurement partners rather than by us
Governance
- • Reporting cadence and definitions, agreed once and held
- • One set of numbers the whole business uses
How we run it
Run
Between the reports, not in them.
This is where the operating routine does the most work. Tracking health is monitored continuously, so a broken event is caught in days rather than discovered in a quarterly review. Pacing and performance are read weekly against plan.
Lift studies run on their own schedule and are read in full, including the parts that did not move. The monitoring watches more signals than a team could read by hand. Every conclusion is written by a person who will defend it, and no budget moves because a model suggested it.
Measure
The limits, because you should hear them from us.
What we report: revenue and pipeline against plan, contribution by channel reconciled across platform and analytics data, and lift results from independent partners, with modeled figures labeled as modeled everywhere they appear.
Scale
Every quarter of clean data makes the next decision cheaper.
Measurement is the only capability where most of the value sits in the accumulated history rather than in the current month. Two years of consistent, trustworthy data is what makes a media mix decision possible at all; six months of it supports an opinion.
The practical consequence is rarely priced in: switching agencies restarts the clock, and the cost of that restart is not the setup fee, it is the eighteen months before anyone can answer a hard question with evidence again.
How this works
-
A considered decision. A clearer path to the conversation.
Home & Building
Connected Capabilities
Media Planning & Buying
Media you cannot measure gets cut first, whatever it was worth.
Paid Search
Brand and non-brand cannot be separated without the tracking to do it.
Conversion & Marketing Technology
A conversion recorded incorrectly cannot be optimized toward.
FAQ
Questions we get asked
What is cross-channel marketing attribution?
Cross-channel attribution assigns credit for a conversion across every channel that contributed rather than only the last one clicked. In practice it combines platform data, analytics, first-party revenue data and, where budgets allow, controlled tests that establish what would have happened anyway. The last part is what separates attribution from bookkeeping: without a test you are dividing credit, not proving contribution.
How should a company measure upper-funnel advertising?
With independent lift studies comparing exposed and unexposed audiences, supported by downstream signals such as branded search volume, direct traffic and store visits. Last-click reporting will always understate upper-funnel media, because the demand it creates converts somewhere else. The baseline has to be captured before the media runs. Retrofitted afterward, the effect is a story about a coincidence.
Why do platform conversion numbers disagree with analytics?
Because they measure different things. Platforms use their own attribution windows, include view-through conversions, and each one claims the same order. Analytics typically uses last click and sees only what reached the site. Neither is lying and neither is complete. The reconciliation between them is the work, and any agency that hands you one number without saying which system produced it is hiding the question.
What is incrementality testing and when is it worth running?
Incrementality testing withholds advertising from a comparable group to establish what would have happened without it. It is worth running when a channel’s budget is large enough that being wrong is expensive, and when volume is high enough for the result to be readable. Below that, a test consumes budget to produce a number with error bars wider than the decision it was meant to inform.
How long does it take to see a brand media effect in search data?
Often days for a change in search behavior, and considerably longer for revenue in categories with a considered purchase. Two conditions decide whether you will see it at all: the window has to be set before the media runs, and the branded search baseline has to be measured first. Without both, the effect is unprovable after the fact even if it happened.
Send us the accounts.
You will work with the people who set the strategy.