Monday, 28 September / signal check

Before spending more,
trust the orders you count.

Repair a duplicated purchase signal. Keep visitors without advertising consent out of activation. Then test what retargeting adds.

Purchase signals received

1,400

Unique orders

1,400?

Consent missing

420

Two checks before activation

Browser + server count the same purchase200 pairs share an order ID. Reconcile them once.

420 visitors have no advertising permissionThey remain excluded. Repairing measurement cannot grant consent.

Excluded

Inspect sample events

Order	Source	Action
ML-1042	Browser	Keep
ML-1042	Server	Duplicate pending
ML-1043	Server	Keep

Activation blocked

A cleaner starting line

Audience tests need one count per order. Resolve the duplicate signal before freezing a test.

Moss & Loom / returning shoppers

Separate return visits
from extra orders.

01 Clean signals02 Freeze an audience03 Compare a holdout

Audience rule

Cart activity withinExclude purchasers from the last 90 days

Only consented, matched shoppers qualify. Exclusions apply across Meta and Google in this simulation.

Demo evidence set

These fixed examples demonstrate uncertainty. They do not predict this store's real results.

Eligibility preview

Cart visitors12,000

Without advertising consent−420

Recent purchasers−1,580

Eligible10,000

14-day cart audience · excludes 90-day purchasers

20% remain in a holdout with no retargeting exposure. Assignment stays fixed for this test.

Frozen test / simulated evidence

More orders. A bounded claim.

Retargeting

Holdout

How the estimate works

Difference in order rates × exposed people estimates extra orders. A normal approximation gives a 95% interval for the rate difference. This demo assumes random assignment, independent people, complete orders, and no contamination.

Order margin is $30 after refunds and variable costs, before ads. Subtract $3,000 test spend. The estimate applies only to the tested people and period.

Extra orders, estimate

95% interval, orders

Northline / client review

A decision with its evidence.

No brief yet

Finish an audience test before preparing a client decision.

Pricing concept

Charge for clearer decisions.

Store / $299 per month

One store. Signal diagnostics, audience exclusions, holdout evidence, and client briefs.

Agency / $1,499 per month

Up to ten stores. Shared agency review and read-only client access.

Proposed pricing from the build plan. No real subscription or billing in this demo.

Choose a concept plan

Before spending more,trust the orders you count.