Normalise Meta and Google Ads Data for Fair Comparison

To normalise Meta and Google Ads data, you put both platforms on the same click-based attribution window, remove view-through credit, split prospecting from branded and retargeting campaigns, and then check each channel against orders in your own store. Meta and Google count conversions with different default windows and credit rules, so raw ROAS from the two dashboards cannot be compared directly. This guide gives you a six-step method, a decision table and a worksheet you can fill in with your own numbers.

Key takeaways

• Meta's default is 7-day click plus 1-day view; Google Ads defaults to a 30-day click-through window. Match them on click-only before comparing anything.
• Google Ads data-driven attribution is available to every conversion action. The old advice that you need 50 conversions a month no longer matches Google's own documentation.
• Compare prospecting to prospecting. Branded search and retargeting inflate ROAS because the buyer already knows you.
• Use your own store orders as the reference total, and expect a gap. Report that gap as a number rather than hiding it.
• Log the date you change any attribution setting and never compare periods across that line without a note.

Why can't you compare Meta and Google Ads numbers directly?

Each platform reports only the conversions it can see and credits itself by its own rules. Meta counts a purchase if someone clicked an ad in the last seven days or viewed one in the last day, under its default. Google Ads counts a conversion by its conversion window and attribution model, and by default that is a 30-day click-through window and, for most conversion actions, data-driven credit.

Neither platform sees the other. A customer who sees a Meta ad on Monday, searches your brand on Google on Wednesday and buys is claimed by both. Add them up and you get more revenue than your store recorded. That double counting is the core problem, and normalising does not remove it entirely; it makes the two reports measure the same kind of event so the remaining gap is smaller and explainable.

Meta's attribution has also moved recently. Reporting from Dataslayer states that the 7-day view and 28-day view options were removed from the Ads Insights API in January 2026, leaving 1-day view as the only view window. If you have older exports built on longer view windows, they are not comparable with current data. Jetfuel Agency dates the removal to January 12, 2026 and notes the default remains 7-day click plus 1-day view.

What is each platform's default attribution?

Start by writing down what each source counts. The table below uses the vendors' own documentation and reporting from the sources linked in this post; check your own account because settings can be changed per ad set or per conversion action.

The fair way to read this table is that no row is the truth. The store row is the only one that records actual orders, but it cannot say which ad caused them. Your job is to make the first two rows as similar as possible and use the third as the ceiling.

How do you normalise Meta and Google Ads data? (six steps)

Step 1: Write down your current settings and freeze a baseline

Action: Before changing anything, export the last full month from Meta and Google Ads with the settings you have today, and record the attribution setting, conversion window and model for each conversion action in a simple log. Result: you hold a labelled legacy baseline. Why: once you change a setting, reported revenue will shift, and without the baseline nobody can tell whether performance changed or only the measurement did.

Include the date, who changed what, and the conversion action names. If you work with an agency or several buyers, this log is the single document that stops arguments three months later.

Step 2: Put Meta on a click-only view

Action: In Ads Manager, add comparison columns for attribution rather than editing the ad set's optimisation setting. Jon Loomer Digital describes compare attribution settings columns, including 7-day click, and a 28-day click column you can add. Use 7-day click as your comparison basis. Result: a Meta revenue figure that excludes view-through. Why: view-through credit has no equivalent in a standard Google Search click report, so keeping it in makes Meta look better by construction.

Keep this as a reporting view. Changing the ad set's attribution setting changes how the delivery system optimises, which is a different decision from how you read results. Our guide on what resets the Meta learning phase explains why editing live ad sets casually can cost you stability.

Step 3: Check Google Ads conversion windows and the attribution model

Action: Open each conversion action your account bids on and note its click-through window, view-through window and attribution model. Google's documentation says that if you do not customise the click-through window the default is 30 days. For a like-for-like comparison with Meta's 7-day click, either set the Google click-through window to 7 days for the conversion actions you compare, or pull a 7-day-click view from your store using UTM data. Result: both platforms now describe purchases that followed a click within a similar period. Why: a 30-day window credits Google for purchases that Meta's seven-day window cannot, which tilts the comparison towards Google for products with long consideration.

Changing a conversion window applies from that day forward, not retroactively, according to the same Google Ads Help page. That is another reason to log the date. The same page says assisted conversions are shown over 30, 60 and 90 days, so you can still see the longer influence without letting it inflate your headline number.

Step 4: Align the credit model, and correct the 50-conversion myth

Action: In Google Ads, open Attribution under the Goals menu and check which conversion actions still use last click. The attribution reports page from Google Ads Help describes a Switch to DDA tab for moving eligible actions from rules-based models to data-driven. Result: Google credit spread across touchpoints rather than handed to the final click. Why: Meta and Google credit conversions by different rules, so moving Google off a single last-click rule narrows the gap between the two reports.

Older guides said data-driven attribution needs 50 or more conversions a month. Google's current page on data-driven attribution says all conversion actions are eligible regardless of conversion or interaction volume, while noting that performance improves with more data. So low-volume accounts can use it, but should treat its credit split as less stable. If the model looks erratic, Google says you can choose another model.

One caution: do not expect Google and Meta to use an identical model. They credit conversions by different rules, so the aim is comparable windows and no view-through, not mathematical identity.

Step 5: Split campaigns by funnel role before you compare

Action: Tag every campaign as prospecting, retargeting or branded. In Google Ads, branded search goes in its own campaign or label. In Meta, broad, interest and lookalike audiences are prospecting; site visitors, cart abandoners and customer lists are retargeting. Result: three groups you can read separately. Why: branded search and retargeting reach people who already know you, so their ROAS reflects demand created elsewhere, often by the other platform.

Then compare only prospecting Meta against non-branded Google Search and Shopping. This is the nearest thing to a fair test of which platform is better at reaching new buyers. Retargeting and branded results are still worth tracking, but as a conversion layer on top of demand, not as proof that the channel generates it. If your retargeting pools include past buyers, see our walkthrough on excluding existing customers from Meta retargeting, and on overlap, how to fix audience overlap.

Step 6: Reconcile to your own order data and report the gap

Action: Using consistent UTM parameters, pull orders and revenue from your store for the same dates and sum them by source. Compare that total with the sum of your normalised Meta and Google numbers. Result: a reconciliation gap you can name. Why: platforms overlap, so the sum of their claims will normally exceed what the store recorded, and the store gives you the ceiling for any single channel.

UTMs are what make this step work. Without a clean naming scheme, paid social and paid search traffic ends up in direct or unassigned. Our UTM structure guide sets out a convention you can reuse for ads as well as creators. For the model choice at this stage, read last-touch versus multi-touch attribution.

What does the reconciliation worksheet look like?

You do not need a data warehouse. Fill in this table once per month for the same date range. Use your own figures; the logic matters more than any example.

Worked method: if A plus B plus C is greater than D, the platforms collectively claim more than the store recorded, and F tells you the share of claimed revenue that is double counted or unattributed. Apply F as a haircut when you compare channels, or better, track F month to month. A rising F often means retargeting is expanding faster than prospecting. Divide each channel's adjusted revenue by its spend, or enter both into our Google and Meta ROAS calculator, which also shows blended ROAS against break-even.

How should you choose which comparison to trust?

Different questions need different views. This decision table maps the question to the data that answers it.

The last row matters most. Normalisation tells you how to read reports; it does not prove causation. If a budget decision is large, run a geo or audience holdout and compare. We explain the logic in how to calculate campaign incrementality, and the same approach works for paid ads.

How do you handle data from before you changed settings?

Treat the change date as a hard boundary. Label everything before it as legacy attribution, keep the baseline export from Step 1, and compare quarter on quarter only between periods using the same method. Where you must show a trend across the boundary, put the last legacy month and first new-method month side by side and annotate the chart.

Also be careful with Meta data pulled through the API. The Hyros write-up on the January 2026 change says it affected reporting rather than delivery, meaning a fall in reported conversions around that date may not reflect real sales. If your dashboard shows a step down in mid-January 2026, check whether your tool was pulling the removed view windows before you blame creative or budget.

A like-for-like comparison needs three things held constant: the window, the credit rule and the audience type. Change one without noting it and the trend line is no longer evidence.

Which normalisation checklist should you follow each month?

1. Export last month's data with unchanged settings and save it as the baseline.
2. Record Meta's attribution setting and the comparison columns you use (7-day click as the basis).
3. Record each Google conversion action's click-through window, view-through window and model.
4. Confirm no conversion action has been double counted, for example the same purchase firing as two conversion actions.
5. Tag all campaigns as prospecting, retargeting or branded.
6. Pull store orders by UTM for the same dates and complete the worksheet.
7. Record the overlap rate and any setting changes with their dates.
8. Only then compare ROAS between channels, prospecting to prospecting.

What are the most common mistakes?

• Comparing account-level ROAS. It blends branded, retargeting and prospecting in different proportions on each platform.
• Believing the platform total. Adding Meta and Google revenue and treating it as one channel's output ignores overlap.
• Editing live settings without a log. The change shifts reported revenue and nobody can reconstruct why.
• Using a 30-day Google window against a 7-day Meta window. This rewards Google for slower buyers and is not a performance difference.
• Skipping UTM hygiene. Untagged clicks fall out of the store report and make a channel look weaker than it is.
• Treating normalised data as causal. Cleaner reporting is not the same as an incrementality result.

Many of these show up as a sudden cost swing. If your numbers moved overnight, check why Meta cost per result can increase suddenly and why dashboard ROAS can look better than the bank account before you rebuild your reporting.

How does Pulse help with Meta and Google Ads comparison?

Pulse by Nurdd is a web dashboard that connects to Meta Ads and Google Ads through read-only OAuth, explains campaign performance in plain language, diagnoses what changed and recommends specific actions without making you build reports. It belongs in this answer because the hardest part of cross-platform comparison is reading two differently structured accounts side by side and knowing why a number moved.

Book a walkthrough at https://www.nurdd.club/demo. Pulse is in private access and has no free trial. Details are on the Pulse page.

Pulse reads your ad accounts; your store data is still the check on the total. For more on reading each platform without an analyst, see Meta Ads versus Google Ads performance, how to read a Meta breakdown report and the Google search term report. If you are weighing software against outside help, agency versus software covers the trade-offs.

How should you split budget once the data is comparable?

Use normalised prospecting results to set the starting split, then adjust with marginal returns. If a channel's prospecting ROAS is higher but it is already near its audience ceiling, spending more will not return the average. Test increases in small steps and watch cost per result rather than jumping the budget.

Plan the split with the marketing budget planner, which divides a budget across creators, Meta and Google. For a wider measurement system that includes creators and AI search, the D2C marketing measurement framework shows how the pieces connect, and D2C KPIs that beat ROAS shows what to track once ROAS is comparable.

How we researched this

We reviewed the platforms' own help documentation for Google Ads conversion windows, attribution models and data-driven attribution, and industry write-ups on Meta's attribution settings and its January 2026 changes. Everything was checked on the publication date of this post. We did not run hands-on tests of ad accounts, and the worksheet uses your inputs rather than example data. Platform settings change often, so confirm defaults in your own account before acting.

Frequently asked questions

What is the fairest way to compare Meta and Google Ads ROAS?

Compare prospecting Meta campaigns with non-branded Google campaigns, both on a click-only basis with similar windows. Exclude branded search and retargeting from that comparison. Then check the combined result against your store's order data.

What is Meta's default attribution setting?

According to Jon Loomer Digital, the default is 7-day click and 1-day view. Reporting from Dataslayer says longer view windows were removed from the Insights API in January 2026.

What is Google Ads' default conversion window?

Google Ads Help says the click-through conversion window defaults to 30 days when you do not customise it when creating a conversion. You can change it per conversion action, and the change applies from that day forward.

Does data-driven attribution need 50 conversions a month?

Not according to Google's current documentation, which says all conversion actions are eligible regardless of volume. It adds that performance improves with more data, so low-volume accounts should treat the credit split with caution.

Why do Meta and Google together report more revenue than my store?

Both platforms can claim the same purchase when a customer touched both before buying, and each counts by its own rules. The store records one order. The difference is the overlap, and tracking it monthly is more useful than trying to eliminate it.

Should I remove view-through conversions from Meta reporting?

For comparison with Google, yes: use 7-day click as the basis. You can still look at 1-day view as a separate column to understand how much of Meta's reported revenue depends on impressions without clicks.

Can I compare periods before and after changing attribution settings?

Only with a clear note. Keep a baseline export from before the change and compare like periods with like methodology. Otherwise a drop or rise may simply be the measurement changing.

Conclusion

Meta and Google dashboards answer different questions, so normalising them means matching windows, removing view-through, separating funnel roles and reconciling to your own orders. Do that and you get a comparison you can defend, with the remaining gap stated honestly. Start with the baseline export and the worksheet this week.

To see Pulse by Nurdd for your own Meta and Google Ads reporting, book a walkthrough. Pulse is in private access and has no free trial. You can also start with the free Nurdd tools.