Online Ad Measurement: Metrics, GA4 Tracking, Attribution and Dashboards

Online ad measurement means recording what each ad actually caused, from the people who saw it and the people who clicked, through to the sales or leads that came back, and then comparing those numbers with what you paid. The core metric set most advertisers use is CTR, CPC, conversion rate, CPA and ROAS, and the data comes from two sides: each ad platform's own reporting, and a web analytics tool such as GA4 installed through Google Tag Manager.

This article does not go deep on ROI or break-even ROAS formulas, because those are topics of their own. It walks through the whole system instead: choosing metrics, installing tracking, comparing numbers across Google, Meta and TikTok, understanding why the numbers never match between tools, choosing an attribution model, and finishing with a dashboard your team can open every week.

Which metrics should you use to measure ads

Ad metrics fall into three layers along the funnel. The top layer tells you how far an ad was seen and whether it caught attention. The middle layer tells you whether the people who arrived did what you wanted. The bottom layer tells you whether the money was worth it. Looking at only one layer often leads to the wrong decision, such as a campaign with a very high CTR that nobody buys from, or a campaign with a cheap CPA that brings in low-value customers.

Basic formulas worth remembering

• CTR (Click-through rate) = clicks divided by impressions, times 100. It shows how well the message and creative pull people in.
• CPC (Cost per click) = cost divided by clicks. It is the cost of bringing one person to your site.
• Conversion rate = conversions divided by clicks (or interactions), times 100. It shows how well the landing page and the offer work.
• CPA (Cost per acquisition) = cost divided by conversions. Use it for businesses whose goal is a lead or a sign-up.
• ROAS (Return on ad spend) = conversion value divided by ad cost. Use it for businesses that can send sales value into their systems.

A hypothetical example to make this concrete: a campaign gets 50,000 impressions and 1,000 clicks, spends 10,000 baht, and brings 40 orders worth 60,000 baht in total. That gives a CTR of 2%, a CPC of 10 baht, a conversion rate of 4%, a CPA of 250 baht and a ROAS of 6x. These numbers only illustrate the arithmetic. They are not a standard for any industry.

Choose the main metric by business model

An online store that can pass order value into GA4 and the ad platforms should treat ROAS as its main metric. Businesses that close through calls or chat, such as clinics, real estate or B2B, usually anchor on the CPA of qualified leads. Awareness campaigns should be judged on reach, frequency and cost per reach, rather than expecting immediate conversions. What matters is setting one main metric per campaign and letting the others act as diagnostics, instead of giving every metric equal weight.

How to set up tracking with GA4 and Google Tag Manager

Every number in a report is only as trustworthy as the tracking behind it. This order of work fits most websites:

1. Create a GA4 property and a web data stream, and note the measurement ID that starts with G-. Set the property's time zone and currency to match the ones used in your ad accounts.
2. Install the Google Tag Manager container on every page of the site, in both the head and the body, as GTM instructs.
3. Create a Google Analytics tag in GTM, put the measurement ID in the Tag ID field, and set the trigger to Initialization - All Initialization Events, as the Tag Manager help guide describes.
4. Create event tags for the actions that matter, such as form submissions, call-button clicks, LINE chat clicks or purchases. For ecommerce, send the data through the data layer so value and item details travel with the event.
5. Test with Preview in the GTM workspace, which opens Tag Assistant so you can check that tags fire at the right moment and carry every parameter. Then check again in GA4 DebugView.
6. Click Submit and Publish, and give the version a name that says what changed, so you can roll back if something breaks.
7. Mark key events in GA4 for the events that are business goals, and link GA4 to your Google Ads account.

The parts that usually get missed

Only a handful of mistakes cause most broken numbers. The first is duplicate firing, for example installing the Google tag both in the site code and in GTM, so page views or purchases count twice. The second is counting conversions on a thank-you page that people can refresh. The third is leaving UTM parameters off Meta and TikTok ad links, so that traffic lands in the referral or direct channel instead. On the Google Ads side, keep auto-tagging switched on. For links from other platforms, add utm_source, utm_medium and utm_campaign using the same convention in every campaign.

The other piece is the platform-side pixel and API. Meta has the Meta pixel alongside the Conversions API, and TikTok has the TikTok Pixel alongside the Events API. Sending data from the server side reduces losses from browsers that block cookies, but deduplication has to be configured correctly, or conversions will be counted twice.

Consent mode and modelled data

Once a site shows a cookie consent banner, some users will decline. GA4 has a behavioral modeling feature for consent mode that uses machine learning to estimate how those users behave, but Google's help page sets conditions. Consent mode must be on across all pages, the property needs at least 1,000 events per day with analytics_storage set to denied for at least 7 days, at least 1,000 daily users with analytics_storage set to granted on at least 7 of the previous 28 days, and the reporting identity must be set to Blended before modelled data appears. Google adds that meeting these thresholds does not guarantee eligibility. Many small sites will not meet that threshold, so expect GA4 numbers to undercount the share of users who decline cookies.

Why Ads Manager and GA4 numbers do not match

This is the most common question in online ad measurement. The short answer is that each system counts differently on purpose, and a gap does not mean either one is broken. The main causes:

• Different attribution models. GA4 spreads credit across every channel, including organic ones (direct visits get credit only when the whole path is direct), while Meta or TikTok look only at their own ads. If a customer sees a Meta ad and then searches on Google before buying, both platforms may claim the sale.
• Different attribution windows. Each platform looks back over a different period when it counts conversions. The table below has the details.
• View-through. Meta and TikTok can count conversions from people who saw an ad but did not click, within the view-through window you set. GA4, which works from sessions on your site, cannot see those unclicked ad views.
• A click is not a session. One click may never become a session if the person closes the page before the tag loads, or a browser blocks the script. A session can also come from a click that happened long before.
• The reporting date. Ad systems usually tie a conversion back to the date of the click or interaction, while GA4 reports show the time the event happened.
• Time zone and currency. Each account has its own time zone setting. If the ad account and the GA4 property use different time zones, daily totals drift apart at the day boundary.
• Consent and modelling. Each platform estimates missing data its own way, so the results are estimates built on different assumptions.

In practice, do not try to make every number match exactly. Decide which system is the reference for which question. For example, use GA4 or your back-office system (CRM or order system) as the neutral figure for total sales, and use the ad platforms' numbers to compare campaigns within the same platform.

Can you measure across Google, Meta and TikTok consistently

You cannot make them agree to the last digit, but you can make them fair to compare. Start by knowing how each platform sets its windows. This table summarises the mechanics from each platform's official help pages as of October 2026.

How to make them comparable

Step one is to use GA4 as the neutral view, because it is the only system that sees every channel under the same rules, provided your UTMs are consistent. Step two is to compare within platforms with windows set as close together as possible. If you want to compare Meta with TikTok, for instance, look at 7-day click on both and break view-through out into its own column. Step three is to check against real totals in your back-office system. If the conversions all the platforms claim add up to far more than actual sales, the platforms are double counting each other. That is normal, but you need to know it before you move budget.

For the question "if I stop this channel, how many sales would actually disappear", no attribution model gives a complete answer. A closer method is an incrementality or lift test that splits people into a group that sees the ads and a group that does not. That takes a fair amount of budget and time, so it suits the channels that take the largest share of spend.

How to choose an attribution model

Since November 2023, GA4 has removed the first click, linear, time decay and position-based models. Google Ads removed the same set and moved conversion actions that used them to data-driven. That leaves a narrower set of choices:

• Data-driven uses your own account data to spread credit across touchpoints. It suits accounts with enough conversions for the model to learn from. One thing to know: GA4's help page says conversions can be reattributed for up to 7 days after they happen, so the most recent figures can still move.
• Last click gives all the credit to the final click. It is easy to understand and stable, but it gives too little weight to channels that create demand early on, such as video or social.
• Google paid channels last click in GA4 gives credit to Google's ad channels first. Use it when you want a view close to Google Ads.

A workable rule is to use data-driven as the main model once conversion volume is high enough, and keep last click as a comparison view. If the two views paint very different pictures for a campaign, that campaign is probably doing upper-funnel work and should also be judged on other metrics. Once you choose, stick with it for a while. Switching models back and forth makes month-over-month trends impossible to compare.

How to build an ad reporting dashboard

A good dashboard answers the same questions every week in a few minutes. The tool most teams use is Google's Data Studio (formerly called Looker Studio), which has connectors for GA4 and Google Ads and supports blending data from several sources. Meta and TikTok data has to come through third-party connectors, or be exported as files or Google Sheets and imported.

A suggested structure

1. Top row: summary figures for the period, covering spend, total conversions, and CPA or ROAS against the previous period.
2. By channel: a table showing spend, clicks, conversions and CPA or ROAS for Google, Meta and TikTok, stating clearly which system each column comes from and which window it uses.
3. The neutral GA4 view: conversions by session source/medium, to compare with what the platforms claim.
4. Daily or weekly trend: a line chart of the main metric, so you can see when a change happened.
5. Notes: a space to log events such as tag changes, landing page changes or a promotion starting, because numbers without context are easy to misread.

Data retention to set first

Standard GA4 properties can keep user-level and event-level data for 2 months or 14 months, while GA4 360 properties can keep event data for 26, 38 or 50 months. The setting affects explorations and funnel reports, not the standard aggregated reports. If you want to compare behaviour year over year in explorations, set it to 14 months on day one, because data past the retention period is deleted automatically. If you need to keep data longer than that, the option is to export it to BigQuery.

What this means for Thai marketers

Many Thai businesses close sales through a LINE Official Account, by phone, or in Page chat, which means the real conversion happens off the website. If you measure only website purchases, you will see ad results that are lower than reality. Track LINE button and call button clicks as events and, where possible, record the lead source in your back-office system so you can match it later. Also check that the ad accounts and GA4 are both set to Thailand time and to baht, so daily totals and values do not drift apart for no reason. Finally, if your site asks for cookie consent, declined cookies directly affect how much data you can collect, so consent mode setup belongs in the measurement plan from the start.

Frequently asked questions about online ad measurement

Where should online ad measurement start

Start by defining what a conversion is for your business, then install GA4 through Google Tag Manager so that event is counted correctly before you look at any other metric. Without a clear conversion definition, CTR or CPC can only tell you that ads were clicked, not that they made money.

Should I use ROAS or CPA

Use ROAS when you can send sales value into your systems, and CPA when the goal is a lead or a sign-up that has no money value at the moment it happens. A business with both can use them side by side, split by conversion action.

Meta shows far more conversions than GA4. Is that wrong

Usually it is not, because Meta counts view-through and credits its own ads, while GA4 counts only sessions that reach the site and spreads credit across every channel. Check that your UTMs are complete and that Conversions API deduplication is set correctly. If both are right, the remaining gap is the difference in counting methods.

How often should I change the attribution model

Not often. Pick one main model and keep using it so you can compare trends month to month. If you want a different view, use model comparison reporting instead of switching the account setting.

Do I need a dashboard if the platforms already have reports

Yes, because each platform's reports see only their own channel. A dashboard that pulls everything into one place, with the source of every number stated, lets you make cross-channel budget decisions from the same picture.

Summary

Practical online ad measurement rests on correctly installed tracking, a clear conversion definition, windows you know the settings of, and a dashboard that states where every number comes from. If you want help auditing or setting up GA4, see our Google Analytics 4 setup and migration service. For campaign measurement on each channel, see our Google Ads service, Facebook Ads service and TikTok Ads service. To bring data from several sources together, see our data integration and consolidation service.