Acquisition checks

Trace the leak from ad spend to landing experience.

Give FunnelLoom campaign spend, the landing path, a page check, and last-known-good rates. It flags down pages and conversion drops, keeps every source row, and holds or confirms each flagged path. Pause suggestions stay behind a person.

Inspect a campaign path See the flagged warp

Sign in and you go straight back to the FunnelLoom conversation. Work happens in chat. This page never asks for your card details or email address.

PEGS ON THE BEAMDiagnose in this order. A later drop never overrides a down page.

01Page first5xx is landing_page_down
02Then CTRBelow 70% of last-known-good is ad_engagement_drop
03Then formBelow 70% of last-known-good is form_conversion_drop

01 / BRING THE SNAPSHOT

Intake is the spend file, the path, and last-known-good.

Performance marketers who need unusual spend connected to the exact landing path a visitor experiences, with a held or confirmed verdict on every flagged row.

01
Campaign spend, clicks, impressions, conversions, and form submits for the window
02
Landing path plus a page-check status
03
Last-known-good CTR and form rate for the same path

02 / THREE COMMANDS ON THE LOOM

Diagnose, decide, then pin the flagged table.

Spend can keep burning after a landing page dies, or a conversion drop is treated as a creative problem when the form is the break.

01
anomaly_table.py diagnose computes CTR and form rate, then checks page status before engagement before form conversion.
02
decide recommends pause only when the landing is down and spend crossed the stop-loss, with approval_required true.
03
table copies source rows and last-known-good values; page-down paths are confirmed, metric-only drops are held.

WHAT COMES OFF THE WARP

Healthy paths stay off the flagged table.

Output 01

CTR and form rate to one decimal, plus landing_page_down, ad_engagement_drop, form_conversion_drop, or no_anomaly

Output 02

A campaign-by-landing table with source rows, last-known-good values, and a held or confirmed verdict for every flagged path

Output 03

A pause suggestion that always requires approval and is never executed here

CONTROLLED EVAL ROW

CMP-4 /pricing is confirmed. CMP-9 /blog/guide is not on the table.

Verified table output for window 2026-07-14: checked 2 paths, flagged 1. Source row page_status 503 and spend 320 stay as supplied. Last-known-good CTR 5.0 / form rate 4.1 stay as supplied. This is not a production ads outcome.

Campaign: CMP-4
Landing path: /pricing
Diagnosis: landing_page_down
Verdict: confirmed
Decision: recommend_pause
Approval: approval_required true

THE PAUSE PEG STAYS WITH YOU

FunnelLoom flags the warp. A person still decides the pause.

For performance marketers who need unusual spend connected to the exact landing path a visitor experiences.

It does not

Live writes to ad accounts, bids, or budgets
Calling Ads or GA4 hosts instead of using the supplied snapshot

Before acting

Ad-platform reporting delays can make recent anomalies incomplete.
Budget, bid, and campaign changes require an authorized marketer.

BEFORE YOU BRING A SNAPSHOT

Questions

Does it pause the campaign by itself?

No. recommend_pause always sets approval_required true. The command never writes a budget or calls an ads host.

What is a confirmed versus held verdict?

Confirmed means the page check returned 500 or higher. Held means CTR or form rate dropped while the page still loaded, so a person still has to confirm.

Are source numbers rewritten?

No. table copies source_row and last_known_good unchanged. Only computed CTR, form rate, diagnosis, verdict, and the budget suggestion are added.

STARTER / MONTHLY

Starter

Billing appears inside the signed-in workspace. Card payment runs through Stripe Checkout.