TrueROAS

What should I check in my attribution tool every week?

The 15-minute routine

This works in any attribution tool. Do it on the same day each week, for the last full Monday to Sunday.

Weekly attribution check
CheckMinutesNormalAct when
1. Orders the tool sees vs orders in your store2Close to every orderIt drops from one week to the next
2. Platform's sales vs matched sales, per channel3Similar to last weekIt jumps or falls by a third or more
3. ROAS on matched sales, per campaign4Above your break-evenBelow break-even for 2 weeks in a row
4. Share of new customers, per channel2StableA prospecting campaign mostly sells to returning buyers
5. Event Match Quality in Meta Events Manager2StableThe score falls after a site or app change
6. Write down one decision2One lineEvery week

1. Does the tool still see all your orders?

Compare the order count in your attribution tool with your store for the same week. This catches the silent breaks: a theme update that removed a script, a new checkout app, a payment method that skips the thank-you page. Browser-only tools miss a lot even when nothing breaks. GA4 recorded a median 76% of real orders across 16 stores we track in October 2026. What matters each week is that the share does not drop.

2. Has the gap between platform and orders changed?

Every platform reports its own sales, and they rarely match your orders. In our sample, Meta reported a median 1.12x the sales we matched to real orders, and Google Ads 0.81x. The size of your store's gap is less useful than its change. A sudden jump in what Meta claims often means a purchase is being sent twice. A sudden fall usually means tracking broke. Why the gap exists at all is covered in why Meta shows more sales than Shopify.

3. Which campaigns are below break-even on real orders?

Sort campaigns by spend and look at ROAS on matched sales, not the platform's ROAS. A quarter of the stores in our Meta sample (26%) had ROAS under 1 on real orders over the 4 weeks. Below 1 means the ads brought in less than they cost, before product costs. Your own line is your break-even ROAS, which depends on your margin; see how to calculate break-even ROAS. Give a campaign 2 weeks below the line before you cut it, unless it is far below.

4. Are the ads finding new customers?

Check the share of matched sales from first-time buyers per channel. In our sample the median was 91% for Meta and 75% for Google Ads. A prospecting campaign whose buyers are mostly returning customers is taking credit for people who would likely have bought anyway.

5. Is the data you send Meta still good?

Open Events Manager and look at Event Match Quality for your purchase event. Meta describes it as a score out of 10 that shows how well the customer information you send matches events to Meta accounts, and it is calculated in real time (Meta for Developers). A drop after a site change usually means email or phone stopped being sent.

6. Write down one decision

One line: what you will change this week and why. Cut a campaign, move 20% of a budget, fix a tag, or change nothing on purpose. After a month, the log shows whether your decisions worked. The rules for moving budget in steps are in the budget reallocation playbook.

Why the last full week, not yesterday

Recent numbers are still filling in. Shopify notes that ad spend from marketing apps can take up to 24 hours to sync (Shopify Help Center). In our experience, ad platforms also keep adding late-reported sales to past days for up to about 3 days. A week that ended at least a day ago gives you numbers that will not move much after you have acted on them. If a check fails and you cannot see why, start with the list in common ad tracking setup mistakes.

Usually $300/hour • Free for a limited time

Find Out What's Holding Your Ads Back

30 minutes, 1-on-1 with the founder of TrueROAS.

✓Live review of your Meta & Google ad performance

✓Find the tracking gaps eating your profit

✓Leave with a clear plan to scale


Marius T.
"TrueROAS finally showed me the conversions Meta was missing. We scaled spend 2.3x in 30 days without guessing."

Marius T. • Senior Media Buyer