One owner, seven days, no new reporting platform. That is the limit before you ask your Meta agency to scale holiday campaigns, and it is enough to produce a decision record—not a promise that Ads Manager and your store will match.

I am writing this for store owners already paying for Meta management. From Nov 12, 2026, Meta turns on “Measure conversions on customer devices” by default for datasets already sending events, with no code change needed and an opt-out per dataset in Events Manager. The setting can recover purchases previously missing from reports. That means Ads Manager conversions can rise without any change in shop orders, 15 days before Black Friday. If you have not agreed on revenue and a reconciliation method, a reporting change can look like permission to spend more.

1. Set the constraint: one owner, 30 minutes a day

Cost: 30 minutes a day for seven days, your agency’s current weekly report, and read-only access to Shopify or WooCommerce orders. You are the internal owner. The agency supplies the exports, explains the gaps, and accepts the resulting scale rules. What you buy is one written record: what counts as revenue, how you compare Meta with the store, which differences need investigation, and who can approve more spend.

What gets cut is deliberate: no new dashboard, pixel rebuild, full Conversions API audit, or holdout test inside this week. I used to tell clients to fix tracking first. I was wrong for a holiday decision on this clock. A full audit and an incrementality test need 2–4 weeks plus observation time to detect lift. You can plan that work later. Right now, you need a baseline you can use without pretending two different measurement systems should produce the same number.

Write the constraint at the top of the page you will use all week. It stops a 30-minute review from becoming a pitch for another reporting project.

Shop owner comparing a Meta ad report with store orders on a laptop

2. Day 1: confirm whether the Nov 12 change touches your dataset

Cost: 20 minutes. Buy: a screenshot, not a guess. Email your agency one question: Does the Nov 12 “Measure conversions on customer devices” switch apply to our pixel dataset, and what is it set to? Ask for the dataset ID, its current on/off state, the date that state took effect, and screenshots of the Settings screen and platform notice. The setting lives per dataset in Events Manager > Data sources > pixel > Settings.

Do not let the agency answer from memory. Do not opt out merely because reported purchases moved. Your job today is to establish what changed, or may change, before anyone calls the next print a win.

The advice around Meta’s notice is to record a pre-change baseline, track Meta-attributed purchases against backend orders, keep Conversions API active, and run a holdout before crediting lift. You cannot complete all of that in seven days. You can preserve any available pre-Nov 12 numbers and put the unanswered questions on the record: whether device-matched purchases carry value and currency for ROAS and value bidding, or only counts, and whether they feed delivery and audiences or reporting only.

Free tool: the ROAS Calculator works out your return on ad spend, your break-even ROAS and the budget a revenue goal takes.

Label those items unknown. An agency that guesses today will have a convenient explanation for any number tomorrow. Day 1 ends with a dated dataset status, not a decision to scale or opt out.

3. Day 2: agree on the revenue number before debating ROAS

Cost: 15 minutes. Buy: one definition both sides can use. Put this sentence in the shared record and ask the agency to confirm it: “For scale decisions we use [gross or settled] store revenue, [including or excluding] refunds and cancellations, in [currency], on [store order date in time zone].” Choose the terms. Leaving the brackets open is not agreement.

Why bother? Meta-reported purchases, revenue, and ROAS are gross: a later refund does not remove the purchase from Ads Manager. The linked illustration makes the arithmetic plain. $100k credited revenue on $25k spend shows 4.0x; after $15k in refunds, settled revenue is $85k and settled ROAS is 3.4x. If returns matter to your business, gross ROAS can flatter a weekly report. Your store definition should say which number you will actually judge.

Now record the ad account’s attribution setting. Meta’s default counts purchases up to 7 days after a click or 1 day after a view, while Shopify marketing reports default to last non-direct click and count orders on placement date. These are different rules, not competing versions of one till receipt. Changing the attribution setting changes reported ROAS without changing sales: the linked worked example on $5,000 spend at a $90 average order value runs from 1.71x on 1-day click to 3.06x on 7-day click plus 1-day view.

Keep Meta comparisons on the same attribution setting. Keep the store number identified as store revenue, not “Meta revenue.” Day 2 ends with two named numbers and one agreed rule for using them.

4. Days 3–4: export four weeks and keep the boring ones

Cost: 60 minutes across two days. Buy: a baseline that has survived a second look. Ask the agency for four complete weekly exports of spend, attributed purchases, attributed purchase value, and the attribution setting used. Pull the matching weeks from Shopify or WooCommerce: order count, gross revenue, refunds and cancellations, and the report’s time zone. Use the oldest complete weeks available before the change where possible, and mark any week that crosses a measurement change.

Avoid treating yesterday as a settled result. Meta can keep adding purchases to recent dates for up to seven days after a click; a 7-day report ending yesterday includes days still collecting conversions. If a week is still immature, mark it provisional rather than quietly folding it into the baseline. A neat spreadsheet does not make an unfinished week mature.

For each week, calculate Meta-attributed revenue divided by store gross revenue. Write the ratio beside the two underlying numbers so nobody mistakes it for a conversion rate or an incrementality result. It tells you how the reports relate; it does not tell you how many sales Meta caused.

Expect differences. Meta and Google can both claim the same sale, so their combined attributed totals can exceed store orders. That overlap does not explain Meta alone reporting more purchases than the entire store. Treat a Meta-only excess, value-less purchases, or counts doubling overnight without a corresponding change in orders as a fault to investigate, not a reassuringly normal gap. For the mechanism behind reported ROAS, read Meta ROAS Can Rise Without More Sales. Here’s How. alongside your numbers.

Finally, flag weeks with a promo, stockout, price change, or tracking edit. Set those aside rather than letting an unusual week define “normal.” Aim to keep two clean, boring weeks. If you do not have two, say so on the page. Days 3–4 end with four weeks exported, clean weeks identified, and exceptions visible.

5. Day 5: make the agency explain the gap

Cost: one 30-minute screen-share. Buy: an exception log with causes and owners. Draw two columns. The left is explainable differences: view-through purchases, purchases arriving days after a click, refunds that remain in Ads Manager but leave your settled store total, and sales claimed by both Meta and Google. The right is faults to investigate: purchases with no value, Meta alone reporting more purchases than the whole store, or counts doubling overnight without an order increase. Use Before You Hire a Meta ROAS Agency, Make Your Purchase Numbers Agree as the fuller check behind the exercise.

Ledger separating explainable Meta reporting gaps from tracking faults

Ask for numbers in the left column, not adjectives. In Ads Manager, the agency can open Columns and use Compare attribution settings to show the same campaign under several windows. Default minus 7-day click shows the view-through share; 7-day click minus 1-day click shows the delayed-click share. Those comparisons can help explain why an attributed Meta total differs from a store report. They do not prove those sales were caused by the ads.

Log the shares, then put measurement edits beside campaign changes with dates: attribution-setting changes, a dataset opt-in or opt-out, Conversions API fixes, and refund batches. If a difference remains unexplained, assign it to a person and give it a next-check date. Day 5 ends when every material gap has either an explanation or an owner. “Tracking discrepancy” is not an owner.

6. Day 6: write the scale rules while the budget is still yours

Cost: 30 minutes and one approved page. Buy: a rule the agency cannot rewrite after a good Ads Manager print. Do not scale solely because reported conversions rise after Nov 12. If store orders do not rise too, at the same attribution setting, the performance case remains unproven. Check the exception log before calling the movement a win.

Write the conditions in order:

  1. Economics: the agreed store-based revenue measure supports your breakeven ROAS or profitable CPA threshold for 7 to 14 days, not one good print. Keep that store check distinct from Meta’s attributed ROAS; total store revenue is not proof Meta generated every order.
  2. Stability: the ad sets you plan to scale have roughly 50 or more conversions per week, day-over-day spend is steady, Pixel and Conversions API are firing, and frequency is under control.
  3. Evidence for the increase: store orders rise in the same week, the attribution setting is unchanged, and the exception log accounts for any measurement change.

Then cap the move: increase budgets 10 to 20% every 48 to 72 hours, and pause increases if the agreed economics fall below breakeven. The practitioner scaling checklist also points to sustained ROAS, CPA at or below target, enough weekly volume, stable delivery, and working tracking; it recommends keeping 60 to 70% or more of spend on prospecting rather than retargeting. Put those checks on the page before a holiday result tempts everyone to skip them.

If the agency wants a larger jump, ask for evidence of incrementality, not a wider attribution window. A proper holdout withholds ads from a control and compares business outcomes because observed sales and caused sales are different numbers. That test is work for after this seven-day decision, not a result to pretend you have now. Day 6 ends with an approved ceiling and a reason to stop.

7. Day 7: replace the status call with a decision

Cost: one 30-minute meeting run from one page. Buy: an instruction, an owner, and a date. Open with the clean-week baseline from Days 3–4, the dataset status from Day 1, and the exception log from Day 5. Then choose one outcome:

  1. Approve: store orders have risen, the attribution setting is unchanged, and the agreed store economics remain above breakeven. Fund only the increase allowed by the Day 6 rules.
  2. Hold: the data is immature or confounded by a promo or measurement change. Keep spend flat and set the next review date.
  3. Investigate: a right-column fault remains, such as Meta purchases above all store purchases or purchases without value. Name the agency owner and the date for an answer.

No fourth option. I have sat through too many November calls that ended with “let’s watch it.” Watch it is hold, with a date attached.

Put the reconciliation in the report you already pay for. Require three lines at the top of each weekly agency report through December: store gross revenue and order count; Meta-attributed revenue and purchase count at the stated attribution setting; and the ratio, plus any measurement change and its date. Include the agreed refund treatment so the store line does not quietly change meaning halfway through the season. If the agency cannot produce those lines, you do not have a dashboard problem. You have a management problem. Day 7 ends with one decision, one owner, and one next date.

When the seven-day constraint lifts, keep the record and drop the clock. If your agency did this cleanly, keep it and judge it against the store economics you agreed on. If it guessed at the dataset setting or pushed for more spend on a higher Ads Manager print without more orders, replace that way of working. For paid and organic search, groas takes the alternative seriously: always-on execution, a human strategist accountable for direction, logged actions, and a flat monthly fee rather than a cut of spend. Do not confuse that search model with a claim that it runs your Meta account. The test for your current Meta agency stays simpler: you did the reconciliation once in seven days. From here, it should be part of the job you already pay for.