Last updated: November 20, 2026. Facts checked against Meta's Business Help Center and AdAdvisor's Nova and pricing pages in November 2026.
TL;DR
DTC ecommerce clients usually leave a Facebook ads agency when the numbers they see (store revenue, profit) stop matching the numbers in the report (Facebook ROAS). Before the client says anything, an agency can check four things in the account: reported ROAS against break-even ROAS, store revenue against Facebook-reported revenue, cost per purchase drift, and creative fatigue. Nova, AdAdvisor's approval-first AI media buyer, can surface these signals across client accounts, and in Copilot mode it changes nothing until you approve.
Quick answer
The Two-Ledger Check is AdAdvisor's way to spot Facebook ads agency client churn early: the client reads one ledger (orders and money in the store), the agency reports from another (Facebook-attributed purchases and ROAS), and when the two drift apart without an explanation, the agency has a problem to investigate before the client raises it.
Profit line.
Is reported ROAS above the client's break-even ROAS?
Revenue line.
Does Facebook-reported purchase value move with the revenue the store recorded?
Cost line.
Is cost per purchase drifting up against the client's allowable cost?
Creative line.
Are frequency and fatigue statuses rising on the ads that carry the account?
This guide gives no churn percentage, only signals an agency can check on Monday morning. Facebook interface details were checked in Meta's Business Help Center on November 20, 2026.
Why do DTC clients leave a Facebook ads agency?
A common, preventable source of friction in DTC agency relationships is a report that stops matching the store. The founder watches orders, payouts and margin; the report shows ROAS from Ads Manager. When the two disagree for a month and nobody explains why, the client has reason to doubt the work, even if the campaigns are reasonably well run.
Generic advice about "better communication" misses this, because the communication problem usually has a data problem underneath. A client who is never told which number answers which question will usually pick the one that hurts more.

Drowning in Meta Ads?
Put your campaign on autopilot with Nova.
Read moreWhich signs of churn can you see in the ad account?
An ad account cannot tell you whether a client plans to leave, but it can show the problems that tend to erode a client's confidence if left unexplained. Some signals live in Ads Manager and Events Manager; others need the store's numbers and your own reporting log. The table is our editorial position from AdAdvisor's account-review work (8+ years in media buying, $60M+ in managed ad spend, and an ex-Meta engineer on the team), not a validated churn predictor.
| Signal | Where you see it | What to do |
|---|---|---|
| Reported ROAS sits near or below the client's break-even ROAS | Ads Manager ROAS against a break-even figure agreed with the client | Recalculate break-even from contribution margin and show both numbers side by side |
| Facebook-reported purchase value rises while store revenue is flat | Ads Manager purchase conversion value against the store's orders report, same dates | Explain the gap (attribution setting, view-through and engage-through credit, other channels, refunds) before the client finds it |
| Cost per purchase drifts up for two or three weeks | Cost per purchase column, compared with trailing 7, 30 and 90 days | Compare it with the client's allowable cost per purchase and say what you are changing |
| Frequency rises while results fall, or a "Creative limited" or "Creative fatigue" status appears | Frequency column; Delivery column status on the ad set or ad | Brief new, materially different creative before results deteriorate |
| Reports go out late or skip a week | Your own delivery log | Fix it first: late reporting is an avoidable communication problem |
| Purchase events stop matching orders | Purchase events in Events Manager against the store's orders; Diagnostics tab for setup issues and duplicate events | Check tracking before reporting, so you are not defending numbers that may be wrong |
Each signal points to one of four problems, sometimes several at once. A performance problem is cost per purchase rising against the allowable cost. A measurement problem is tracking that no longer matches orders. A profitability problem is reported ROAS that does not clear break-even. A reporting problem is an agency that cannot say what changed or what happens next. Our recommended order is measurement, then economics, then the next decision: changing budgets before a data gap is explained tends to make the conversation harder.
On the creative row: Meta notes that if performance drops as frequency (impressions divided by reach) rises, the audience may be experiencing ad fatigue. Ads Manager shows Creative limited when cost per result is above past ads but less than twice as much, and Creative fatigue at twice as much or more. The labels only appear for eligible ad sets, so a missing label is not proof the creative is fresh. See detecting creative fatigue.
| What you compare | ROAS-only reporting | Two-Ledger Check |
|---|---|---|
| Main evidence | Facebook-attributed results | Facebook results plus the store's orders and margin |
| Profitability | Assumed from ROAS | Checked against an agreed break-even |
| Revenue gap | Often left unexplained | Investigated and explained |
| Limit | Attribution is not profit | Still does not prove incremental impact |
Why can a good ROAS still lose the client?
A positive ROAS does not establish a profitable campaign; the comparison that matters is reported ROAS against the brand's break-even. If a store keeps 40 cents of contribution margin per dollar of revenue, break-even ROAS is 2.5 (illustrative). A reported 2.8x looks healthy and leaves very little for refunds, testing and the agency's own fee.
The second reason is attribution. Under standard attribution, Facebook can credit a purchase within 1 or 7 days after a link click, within 1 day after an impression (view-through), or within 1 day after a non-link engagement such as a 5-second video play (engage-through). The store records every order, including repeat buyers and other channels. Facebook-attributed revenue and total store revenue measure different things, so a gap between them is a question to investigate, not proof of bad tracking.
Meta's incremental attribution, which reports only conversions its model predicts the ad caused, is one more number to explain. Our guide to break-even ROAS from contribution margin covers the calculation, and Facebook attribution covers why ad platform numbers and analytics numbers rarely match.
What should a monthly report show?
A useful DTC report shows whether reported performance clears the client's break-even, how store revenue and margin moved, and where attribution limits what you can conclude.
In most accounts that means three layers. Observed: spend, Facebook-reported purchases and ROAS next to store orders, revenue and average order value. Interpreted: what you think explains the movement, and what remains uncertain, including any change in attribution setting. Decided: what changes next, who owns it, and which number you will check next month. If the client has to work out alone whether the month was good, the report has missed its main job.
What should you agree in the first 30 days?
The first 30 days should set a reporting baseline before results are ever in dispute. Agree on the client's break-even ROAS, their allowable cost per purchase, the attribution setting you will report on, who supplies store revenue, margin and refund data, and when reports arrive.
The client should own the ad account, the business portfolio, the pixel and the catalog. Meta's role guide suggests the Advertiser role for an agency managing campaigns: it can create and edit ads and see reports, but cannot change the payment method or add or remove other people's access. The client's side of that handover is covered in what to do after firing a Facebook ads agency.
What do you do when results drop?
When results drop, diagnose before you defend, and tell the client before they ask. First check whether the drop shows in Facebook reporting, in store revenue, or in both. A drop that only shows in Ads Manager often points to measurement or attribution, so check the Events Manager Diagnostics tab for setup issues and duplicate events, then compare purchase events with store orders yourself. Then work through audience, creative, offer and budget, using the five layers in Facebook ads not working.
Then send the client a short note: what changed, what you found, what you will try next (for example over one or two weeks) and the metric that will show whether it worked, before they find the bad week in the store dashboard.
Which churn can't you prevent?
Some churn is a fit problem, not a performance problem, and no report fixes it.
| Cause | Why it is hard to prevent | What the agency can still do |
|---|---|---|
| Poor fit | The brand needed a channel or service the agency does not do well | Say so early and refer out |
| Budget too small | Low purchase volume makes results more volatile and creative tests harder to read | Set slower expectations, or decline the account |
| Low product margin | Break-even ROAS is above what the market is likely to deliver | Show the math before taking the retainer |
| Business change | Funding, a product change or a decision to bring ads in-house | Leave the account clean and documented |
An unexplained reporting gap is different: the agency can address it even when the business conditions behind it are outside its control.
Where this guide stops
We have no proprietary data on how many DTC clients leave Facebook ads agencies, or why, and we did not find a current, methodology-transparent public benchmark specific to Facebook ads agencies serving DTC brands, so this guide gives no churn percentages. The Two-Ledger Check has not been validated as a statistical predictor of churn; it makes problems visible, and it does not promise you will keep a client. Financial examples here are illustrative, platform statuses and attribution windows come from Meta's documentation, and the Nova price comes from AdAdvisor's pricing page.
Where Nova fits
Running the Two-Ledger Check every morning across many client accounts is the core problem in AI media buying for agencies. Nova, AdAdvisor's AI media buyer, reads each connected ad account, the pixel and the client's Shopify store, then judges every campaign against that client's break-even, so a campaign that looks fine on ROAS and loses money on margin gets flagged instead of scaled. Nova is $199 a month per business for up to five businesses, with Enterprise plans beyond that, and in Copilot mode every change waits for approval with its reasoning attached.
FAQ
Summary
Facebook ads agency client churn in DTC ecommerce usually starts as a gap between two ledgers: the store's revenue and profit, and the agency's Facebook-reported ROAS. An ad account cannot predict a cancellation email, but agencies that check break-even ROAS, store revenue, cost per purchase drift, fatigue and tracking every week can usually find and explain the gap before the client raises it. If your clients sell on Shopify, try Nova on a client store for 7 days and see what it flags.
Sources
- Meta Business Help Center: Frequency (checked November 20, 2026)
- Meta Business Help Center: About creative fatigue recommendations in Meta Ads Manager (checked November 20, 2026)
- Meta Business Help Center: About attribution models and attribution settings (checked November 20, 2026)
- Meta Business Help Center: About Facebook's attribution system (checked November 20, 2026)
- Meta Business Help Center: About diagnostics in Meta Events Manager (checked November 20, 2026)
- Meta Business Help Center: Ad account permission roles in Meta Ads Manager (checked November 20, 2026)
- Meta Business Help Center: Add partners to your business portfolio (checked November 20, 2026)
Related reading

Strategy & Planning
You Fired Your Facebook Ads Agency: What to Do in the First 30 Days
Confirm who owns each Meta asset, copy everything you cannot rebuild, replace and then remove agency access, and keep stable campaigns running while the next operator takes over.
Read more
AI & Automation
AI Media Buying for Agencies: Run More Client Accounts
How agencies use AI media buying to run multiple client Meta accounts: the Per-Client Operator Model, per-client approvals and guardrails, cross-account QA, and client reporting.
Read more
Strategy & Planning
How to Calculate Break-Even ROAS From Contribution Margin (Returns, Fees and Shipping Included)
Most break-even ROAS math stops at gross margin. Here is the by-hand worksheet that adds returns, shipping and payment fees, converts the result to a break-even CPA, and shows why a 65% gross margin can still need a 3.1x ROAS.
Read more
Analytics & Reporting
Meta Ads Attribution: Windows, Why Meta and GA4 Never Agree, and Which Number to Trust
Facebook attribution assigns credit by click and view windows, not last touch. Here's why Meta and GA4 always disagree, and which number to use for each decision.
Read more
Performance Optimization
Detect Meta Ads Creative Fatigue With AI (2026)
Detect Meta ad creative fatigue before CPA breaks: watch leading signals against each creative's own baseline, diagnose concept, audience or auction, and refresh with a materially different concept.
Read more



