Step-change capabilities that break your forecast.
Data Activation covers the advanced capabilities that require the full foundation to be in place first: an AI analyst trained on your warehouse, a custom attribution model that recovers misattributed revenue, and advanced ML models that change how you think about acquisition, retention, and targeting.
Why did Meta CPA spike on Tuesday?
Meta CPAs rose 34% vs. the prior 7-day avg. Checkout conversion dropped simultaneously — looks like a site-side issue, not a media efficiency problem.
Checkout page load time spiked to 4.2s on Tuesday morning — correlates exactly with the CPA increase.
AI Activation
Step-change in speed via a growth analyst embedded in Slack. Analyst-grade answers to any question about performance, in seconds.
Three capabilities, scoped to your readiness.
Each capability is scoped and delivered as a separate engagement. You do not need all three. We will recommend which to prioritise based on your data maturity and where the highest-leverage opportunities are.
01
AI Activation
Create a step change in speed with an AI growth analyst embedded in Slack. Ask any question about your channel performance, retention, unit economics, and get analyst-grade responses within seconds. Pro-active recommendations and alerts to improve performance and stay on top of what is happening.
- Natural-language querying of your warehouse
- Trained on your brand context and channel data
- Proactive performance alerts
- Available to any team member in Slack
02
Custom Attribution Model
Build a re-attribution model to help you understand the true value of upper funnel activity. We utilise post-purchase survey responses to re-distribute direct and brand conversions back to the marketing activity that influenced them. We also build an identity graph to enable tracking of users across devices and browsers.
- UTM and last-click as the base layer
- Post-purchase survey re-attribution
- Cross-device identity graph
- Transparent, fully documented model
03
Advanced Strategy
Advanced data builds that create step-change insight: predictive LTV using machine learning to enable instant payback prediction at acquisition; churn prediction triggers integrated with your CRM; ML-driven customer segmentation; and platform data activation passing enhanced conversion signals back to Meta and Google.
- Predictive LTV model (ML)
- Churn prediction triggers for CRM activation
- ML-driven customer segmentation
- Enhanced conversion data back to Meta and Google
Recover the revenue that platform attribution misses.
Platform attribution is biased toward the last click and toward the platform doing the reporting. Our custom attribution model combines GA4 UTM tracking with post-purchase survey responses and cross-device identity to build a picture that is closer to the truth.
Every rule in the model is visible and documented. No black box. No proprietary weighting you cannot inspect.
Typically unlocks 20–30% more attributed revenue for mid-funnel channels that platform attribution undercounts.
Attribution waterfall — orders this month
Total attributed: 501 orders · +31% vs UTM only
Custom Attribution
Re-attribution model combining UTMs, post-purchase surveys, and cross-device identity to recover 31% more attributed revenue.
Predictive LTV
36-month LTV at acquisition
Churn Triggers
ML signals for at-risk customers
ML Segmentation
Behavioural clusters for targeting
Platform Activation
LTV segments pushed to Meta & Google
Advanced Data Modelling
Predictive LTV, churn triggers, ML customer segmentation, and platform data activation — passing enhanced conversion signals back to Meta and Google.
Use machine learning where it actually creates value.
Predictive LTV tells you at acquisition which customers are worth bidding more for and which are likely to churn. Churn triggers integrate with Klaviyo to act on those signals in real time. ML segmentation gives you more precise audiences to push to Meta and Google.
These models require 12+ months of clean, modelled data to be accurate. They are not the first thing you build, but when the foundation is solid, they create a compounding edge.
Prerequisites
- Data foundation live (Unified Data Layer)
- Minimum 12 months of clean, modelled data
- Forecasting and experimentation frameworks recommended first
- AI Activation specifically requires 6+ months of warehouse history
What leads here.
Data Activation requires these capabilities to be in place first.
Unified Data Layer
The clean, tested data that the AI analyst and ML models are trained on.
MomentumExperimentation
The experimentation programme that attribution data and LTV models feed directly into.
ExpectationsForecasting
Predictive LTV plugs directly into the forecast model to improve the accuracy of acquisition economics.
Ready for the advanced build?
Book a call and we will assess your data maturity and scope the right activation capabilities for where you are now.
Book a Call