Blog

Essays on pLTV, value-based bidding and the signals ad platforms learn from. Written by the people who build Churney.

Monthly data bars on an isometric platform, with a green magnifying glass over a run of empty months

The Six Months Nobody Missed

A free, read-only review that tells you in five minutes whether your warehouse can support pLTV, and finds the problems that usually surface six weeks into a pilot.

A lever with a pLTV price tag lighting up some of the people on an isometric platform in green

The Label Is the Lever

Meta and Google decide what to pay for each customer using the value you send back. Here's how to tell if that number is costing you money, and what to send instead.

Vibe-Coding a Cohort pLTV Model

Your day-7 ROAS is lying to you. Here is how to build the cohort pLTV model that fixes it, and how to know whether you can trust it.

The Reward Layer for AI

Session-level success is not business success. Here is the loop that closes the gap.

An ad platform bids the same on two identical users. Past a day-two wall, one becomes high value and one churns, and a pLTV signal carries that value back to the bidder.

Ad platforms cannot see which path leads to real value. Here's why and what to do about it.

Not every ROAS spike is customer value

Gymshark’s omnichannel chapter (stores, loyalty, drops, and US scale) and why promo-week ROAS is a poor proxy for long-term customer value. Plus what a value-signal layer looks like for data warehouse-mature brands.

Gymshark* fixed measurement. What is Meta still learning on?

How Gymshark moved past Ads Manager to MMM and MTA, and why the next constraint for scaled DTC sporting-goods brands is what ad platforms learn on, not whether you can measure at all.

Signal Engineering Is Right. Now Comes the Hard Part.

Signal engineering is the right idea. Building a value signal that ad platforms can learn from, and proving it works, is the hard part.

Signal Optimization: Why pLTV is a Control System Problem (Not Just a Prediction Problem)

Why a pLTV signal for Meta and Google is a control system problem, not a prediction problem, and what it takes to run one in production.

Growth Is a Living Algorithm

Scale what works or test something new? Growth teams solve multi-armed bandit problems every day. Here is how pLTV signals make that explicit.

The modern UA Manager’s struggle, and why I switched sides.

After 10 years of working in digital marketing and user acquisition, I’ve seen the job evolve significantly. This is what's next.

Conversion Signals Loss

What an ad network needs to exit the learning phase, why conversion signals get lost, and how pLTV events keep campaigns learning.

pLTV for User Acquisition: Two Models Are Required

Budget allocation and ad-platform optimization need different pLTV models. Here is why one model cannot do both jobs.

pLTV Isn’t for Everyone, Is It for You?

pLTV bidding works best when customer value varies widely and shows up late. A quick way to tell whether it fits your business.

pLTV: What are we trying to prove?

How a pLTV pilot is set up and measured: a split campaign, a long-term ROAS goal, a scaling test and three ways to read the results.

DIY pLTV for customer acquisition: A Cautionary Tale and a Smarter Path Forward

Why in-house pLTV builds for ad optimization often fail, what they cost, and why to benchmark a DIY model against a specialist first.

Predictive LTV for User Acquisition: Go Causal or Go Home

Achieving sustainable success in pLTV-based user acquisition (UA) optimization is suboptimal without causal models.

See whether your data predicts long-term value.

One 30-minute call. We look at your setup and tell you plainly whether pLTV is worth testing.

Book a call Or run the free pLTV Audit first →

SOC 2 Type II

GDPR

Read-only access. Identifiers hashed before anything leaves your warehouse.