Short answer: how do you know which ad channel really drives sales? You cannot tell from a single dashboard, because Meta and Google each count the same sale according to their own attribution windows, and their totals add up to more than your real revenue. The right method has three layers: compare channels side by side in GA4 with data-driven attribution, track the whole picture with MER (total revenue divided by total ad spend), and run a geo or holdout incrementality test on any channel you doubt. Rebel Co. Group makes budget decisions based on the MER trend and incrementality results, not on platform ROAS.
At month end, the report says Meta Ads brought in 400,000 TL in sales and Google Ads 350,000 TL. That is 750,000 TL in total. Your e-commerce dashboard, however, shows total monthly revenue of 650,000 TL, and part of that came from organic, email and direct customers. When the ad dashboards add up to more than your real revenue, nobody is lying. Everyone is simply claiming the same sale.
Every e-commerce brand runs into this problem, and it can be solved. In this guide we cover where double counting comes from, what attribution models tell you, GA4 attribution settings, why platform reports differ from GA4, incrementality testing, the MER metric, UTM and server-side tracking discipline, and a weekly decision dashboard. The goal is not perfect measurement but measurement reliable enough to move your budget to the right channel.
Why Platform Dashboards Add Up to More Than Your Real Revenue
The dashboards add up to more than your real revenue because each platform claims every sale that touched its ads entirely for itself, without looking at what other channels contributed. A customer sees an ad on Instagram, two days later searches your brand on Google and clicks, and the next day buys through a link in an email. Meta claims the sale under its "7-day click, 1-day view" window, Google claims it under its "30-day click" window, and the email tool claims it too. One sale shows up three times in three dashboards.
On top of that comes view-through attribution: Meta also counts sales within 1 day from people who only saw the ad and never clicked. That person may already be a loyal customer of yours. The result: platform ROAS alone cannot answer the question "is this channel profitable?" It is only useful for tracking change within the channel itself, meaning how this month compares with last month.
Attribution Models Compared
An attribution model is the rule that decides how the credit for a sale is split across the channels a customer touched, and every model favors a different channel.
| Model | Who gets the credit | Who it favors | When to use it |
|---|---|---|---|
| Last click | 100% to the last click before the sale | Brand search, remarketing, email | Short decision cycles, single-channel businesses |
| First click | 100% to the customer's first touch | Discovery channels: Meta, TikTok, YouTube | When measuring new customer acquisition |
| Linear | Equal share to every touch | No one, it averages everyone | Many channels, unclear weighting |
| Time decay | More credit to touches closer to the sale | Bottom-of-funnel channels | Long decision cycles, high-priced products |
| Position-based | 40% to the first and last touch, 20% to those in between | Discovery and closing channels | Businesses where mid-funnel value looks low |
| Data-driven (DDA) | Machine learning compares converting and non-converting paths | The channel that actually contributes | Accounts with enough conversion volume |
Since 2023, GA4 reports only offer the data-driven and last-click models. First click, linear, time decay and position-based models have been removed. The table is for conceptual comparison.
What matters is not finding "the right model" but knowing which model your decision is based on. A brand that cuts its Meta budget based on last click switches off its discovery channel and three months later cannot understand why brand searches dropped.
Attribution Settings and Conversion Window in GA4
In GA4, attribution is configured in the Admin section, on the "Attribution settings" screen in the property column, and it involves two decisions: the reporting model and the lookback window. Data-driven attribution is the default reporting model and the right choice if you have enough conversion volume. Last click is switched on temporarily when you want a like-for-like comparison with the platforms.
The lookback window determines how many days back a conversion can be linked to a touch. For acquisition events you can choose 7 or 30 days, and for other conversions 30, 60 or 90 days. For products with short decision cycles 30 days is realistic, for furniture or high-priced electronics 90 days. Changing the window does not affect historical data, it applies to data after the change. Compare channels in the "Model comparison" report under Advertising, where you see the same conversion under last click and data-driven side by side.
Why Meta and Google Reports Differ from GA4
Platform reports almost always show more sales than GA4, and there are five structural reasons for this.
- View-through attribution: Meta counts sales from people who saw the ad without clicking, while GA4 only sees sessions that arrived through a click.
- Different attribution windows: Meta uses a 7-day click window and Google a 30-day click window, while GA4 applies the window you set, and if the last click was on another channel, the credit goes there.
- Cookie restrictions and iOS: because of iOS app tracking permission and browser cookie limits, platforms fill in missing data with modeling. Meta's "modeled conversions" figure is an estimate, not a real measurement.
- Cross-device paths: when a customer clicks an ad on a phone and buys on a computer, the platform connects them through the logged-in account, while GA4 often sees two separate users.
- Timestamps: platforms record the conversion on the click date, GA4 on the sale date. After a campaign ends, sales on the platform keep "rising".
Once you know these differences, you read the two numbers not as a contradiction but as two separate scales. The platform report tells you the trend within the channel, GA4 tells you how channels compare, and the till tells you the truth. We covered the character of each channel for budget decisions between Meta and Google in our article Meta Ads or Google Ads.
Incrementality Testing: Geo and Holdout
An incrementality test answers one question: if I had not run this ad, would this sale still have happened? Attribution models interpret the order of touches, while incrementality measures causation, and it is the method that should have the final say in budget decisions.
Geo test
You pick two regions of similar size, keep the campaign running in one and switched off in the other, and measure the revenue difference between them. Example: if revenue in Izmir and Ankara has moved in parallel over the past four weeks, switch off Meta in Ankara for three weeks. If Ankara revenue drops 12% relative to Izmir, and that campaign's Ankara spend was 60,000 TL while the lost revenue was 150,000 TL, the incremental ROAS is 2.5x. The Meta dashboard may be reporting 5x for the same campaign, so the real contribution is half.
Holdout test
In remarketing and existing customer campaigns, 10% of the audience is randomly set aside and shown no ads. You then compare the purchase rates of the group that saw ads and the group that did not. Most brands that show ads to loyal customers learn something painful here: the holdout group buys at almost the same rate, which means a remarketing campaign that looks like 8x in the dashboard has an incremental contribution below 1.5x. Meta's and Google's own "conversion lift" tests work on the same logic and can be set up with platform support in accounts with enough budget.
Reading MER and Platform ROAS Together
MER (marketing efficiency ratio) is total revenue divided by total ad spend, and it is the one number that stays independent of the attribution debate. If total monthly revenue is 1,200,000 TL and ad spend across all channels is 300,000 TL, your MER is 4.0. This number is not affected by double counting, because it uses the revenue in the till and the spend on the invoices.
The reading rule is this: if platform ROAS is rising while MER is falling, channels are stealing each other's sales, or organic and loyal customer sales are being credited to ads. If platform ROAS is flat while MER rises, your brand effect is growing and ads are also feeding organic sales. The decision to raise budget is based on the MER trend and your break-even MER: a store with a 40% gross margin has a break-even MER of 2.5, and an MER of 4.0 leaves a healthy profit margin. Measure channel-level ROAS with the ROAS calculator and real profit with the steps in our ad ROI calculation guide.
The platform dashboard tells you how the channel sees itself, MER tells you what the till sees. Budget decisions are made with the second one.
UTM Discipline and Server-Side Tracking
GA4 can only separate channels correctly if every ad link carries consistent UTM parameters. The rule set is short but strict:
- utm_source is always the platform name in lowercase: meta, google, tiktok, newsletter. "Facebook", "FB" and "meta" show up as three separate channels.
- utm_medium is the channel type: paid_social, cpc, email, affiliate. For Google Ads, auto-tagging (gclid) stays on and is not overridden with manual UTMs.
- utm_campaign is the campaign name itself and includes the date and goal, for example bf2026_remarketing_tr.
- utm_content is the creative or placement, utm_term the audience. If these two fields are left empty, you cannot compare creatives.
- In Meta, use the dynamic parameter template so every ad is tagged automatically and manual entry errors disappear.
Server-side tracking sends conversion data from your own server to the platform to get around browser restrictions. Meta's Conversions API, Google's Enhanced Conversions and server-side Tag Manager do this job. Both work alongside the browser pixel rather than replacing it, and they use hashed data such as email and phone numbers for matching. After setup, you need to monitor the event match quality score in Meta Events Manager and the enhanced conversions status in Google. According to industry reports, correctly configured server-side tracking noticeably increases the number of conversions the platforms see and speeds up algorithm learning, but it does not eliminate the gap between GA4 and the platforms.
Weekly Measurement Dashboard
One weekly table is enough to make decisions. The dashboard below places an e-commerce brand's numbers for the same week side by side on three scales.
| Metric | Source | This week | Last week | Reading |
|---|---|---|---|---|
| Total revenue | E-commerce dashboard | 310,000 TL | 280,000 TL | Real growth 11% |
| Total ad spend | Invoices | 78,000 TL | 70,000 TL | Spend up 11% |
| MER | Revenue divided by spend | 3.97 | 4.00 | Stable, scaling is efficient |
| Meta ROAS (platform) | Meta Ads | 4.6x | 4.1x | Rising, creative change |
| Google ROAS (platform) | Google Ads | 3.9x | 4.0x | Stable |
| Meta revenue (GA4, data-driven) | GA4 | 95,000 TL | 88,000 TL | About 55% of the platform figure |
| Google revenue (GA4, data-driven) | GA4 | 112,000 TL | 108,000 TL | About 75% of the platform figure |
| New customer rate | GA4 and CRM | 58% | 61% | Remarketing share growing, watch it |
| Brand search volume | Search Console | 4,200 | 3,900 | Upper funnel is working |
The figures are examples. The share of platform revenue that shows up in GA4 varies by channel and industry. What matters is that the ratio stays stable from week to week. If the ratio suddenly drops, it is the measurement that broke, not the channel.
Read the dashboard in this order: revenue and MER first, then platform ROAS, then the GA4 ratios, and finally the new customer rate and brand searches. Before raising a channel's budget, ask for a three-week MER trend and, if possible, one incrementality test result. We describe how our performance team works on the performance marketing agency page, and for measuring traffic that comes from AI search, see our article on measuring AI brand visibility. You can find how click costs are structured in our Google Ads costs guide.
No single dashboard tells you which channel sells. Three scales tell you together: the till, GA4 and incrementality tests. A brand that reads all three regularly shifts budget based on profit, not on shiny dashboard ROAS. To set up your measurement stack, attribution settings and decision dashboard together, request a free intro call.
Frequently asked questions
What is an attribution model?
An attribution model is the rule that decides how the credit for a sale or conversion is shared across the ad channels a customer touched along their journey. The last-click model gives the credit to the last touch before the sale, first click gives it to the first touch, and the data-driven model compares converting and non-converting paths to give each channel a share that matches its real contribution. The model you choose determines which channel looks "successful" and therefore where you shift your budget.
Which attribution model should I use in GA4?
If you have enough conversion volume, data-driven attribution, the default setting, is the right choice. GA4 reports now only offer the data-driven and last-click models, and older models such as first click, linear and time decay have been removed. Switch to last click temporarily only when you want a like-for-like comparison with platform reports. Set the lookback window to match your product's decision time: 30 days for fast-moving consumer goods, 90 days for high-priced products.
Why does Meta ROAS differ from Meta revenue in GA4?
Meta also counts sales within 1 day from people who saw the ad but did not click, while GA4 only sees sessions that came through a click. In addition, Meta uses a 7-day click window, connects cross-device journeys through logged-in accounts, fills in missing data with modeling and records the conversion on the click date. If the last click was on another channel, GA4 gives the credit there. As a result, Meta revenue in GA4 typically shows up at roughly half to two thirds of the platform figure.
What is MER and how is it different from ROAS?
MER, the marketing efficiency ratio, is total revenue in a given period divided by total ad spend across all channels. ROAS is the return a single channel or campaign calculates based on its own attribution, whereas MER uses the real revenue in the till and the real spend on the invoices, so it is not affected by double counting. If platform ROAS is rising while MER is falling, channels are stealing each other's sales. Budget decisions are made with the MER trend.
Can you run an incrementality test on a small budget?
Yes, but with a geo or holdout test you set up yourself rather than the platforms' official conversion lift tests. In a geo test you pick two cities whose revenue moves in parallel, switch the campaign off in one for 2-3 weeks and measure the revenue difference. In a holdout test, 10% of the remarketing audience is excluded from ads. On a small budget, start with one campaign and one channel, and test the most expensive campaign that looks best in the dashboard, because that is usually where the biggest double counting sits.
Is server-side tracking (Conversions API) a must?
In e-commerce it is now effectively a must. Because of iOS tracking restrictions and browser cookie limits, accounts that rely only on the pixel lose part of their conversions, the algorithm learns from incomplete data and costs go up. Meta Conversions API and Google Enhanced Conversions work alongside the browser pixel and improve matching with hashed data such as email and phone numbers. After setup, monitor the event match quality score in Meta. Server-side tracking does not close the gap between the platforms and GA4, it only completes the data the platform sees.