Activation

“We're sitting on data we've never used.”

Years of customer history doing nothing. Acquisition is funded; retention and reactivation are still spreadsheets and hunches. The fix is to make that history operational: turn it into audiences and signals that sync themselves to the places you can act on them.

The symptoms

You probably recognise a few of these.

You have years of purchase or booking history and none of it drives a campaign.
Winback is a quarterly idea someone means to get to, not a flow that runs.
High-value and at-risk customers look identical to your tools.
Retention lives in a spreadsheet a person updates by hand.
The warehouse is full and the ad platforms are starving.

booking dataBigQueryactivation

lapsed_90d synchigh_value sync

GCP · reverse-ETL · ads + lifecycle

What I build

Pipelines that make historical data operational. The warehouse stops being a graveyard.
Audiences synced to ads and lifecycle platforms automatically. Built once, kept fresh.
Winback and lapse flows that trigger themselves. No one has to remember.
Churn and high-value signals surfaced while there is still time to act. Early, not in the post-mortem.

Capabilities

The specific, findable things I do inside this. Open one for detail.

Warehouse-native audience building

Audiences modelled where the data already lives, in BigQuery, Snowflake, and dbt, instead of rebuilt by hand in each ad tool. Because the logic sits in the warehouse, a segment definition is version-controlled and consistent everywhere it is used. You define high-value or at-risk once and reuse it across every channel.

BigQuery
Snowflake
dbt

Audience sync to ads and lifecycle

Segments pushed to Meta, Google, Klaviyo, and Customer.io automatically and kept fresh, not exported once and left to go stale. As customers enter or leave a segment in the warehouse, the ad and lifecycle platforms update on their own. Your targeting reflects reality instead of a snapshot from last quarter.

Meta
Google
Klaviyo
Customer.io

Winback and lapse flows that trigger themselves

Reactivation that runs on its own the moment a customer lapses, so no one has to remember to launch it. The signal comes from the warehouse and triggers the flow in your email or ads tool without manual intervention. Winback stops being a quarterly good intention and becomes something that just happens.

Klaviyo
Customer.io

Churn and high-value signal detection

At-risk and high-value customers surfaced while there is still time to act, not flagged in a post-mortem. I model the behavioural signals that precede churn or predict lifetime value and pipe them where your team will see them. You get to intervene before a good customer is already gone.

BigQuery
dbt

Reverse-ETL activation pipelines

Reverse-ETL pipelines that turn warehouse models into live activation, proven on booking platforms like Mariana Tek and MindBody. Historical data on GCP and BigQuery becomes audiences that sync themselves to the tools you spend in. The warehouse stops being a graveyard and starts driving revenue.

Census
Hightouch

How this kind of engagement runs

Model the history into usable audiences, wire the syncs to your ad and lifecycle tools, then set the flows to trigger on their own. Project or retainer. You work directly with me.

Tools I reach for

BigQuery
Snowflake
dbt
GCP
Meta
Google
Klaviyo
Customer.io

Related proof

Bookings / Wellness

Booking data turned into retention revenue

The problem

An appointment-based business sat on years of booking history it had never activated. Retention and winback were spreadsheets and good intentions, while the ad platforms ran on cold audiences.

What I built

I built a reverse-ETL pipeline on GCP and BigQuery, modelled lapsed-visitor and high-value segments (proven on booking platforms like Mariana Tek and MindBody), and synced them to paid and email with winback flows that trigger themselves.

The result

Lapsed customers became a reactivatable audience the business could actually market to.

12.4klapsed customers reactivatable

Explore other services

Measurement“We don't trust our numbers.”View service →Integration“Our tools don't talk to each other.”View service →Automation“Everything depends on someone doing it by hand.”View service →Building“We need something that doesn't exist yet.”View service →

Let's look at your stack.

A 30-minute call. You describe the mess; I'll tell you honestly whether I can help.

Book an intro call