A client opens the report and asks: "Google Ads shows 40 sales, Meta 35, and GA4 shows 50 in total. Which is right?" This comes up in almost every marketing meeting. The answer is simple, but explaining it takes care: each system assigns sales by its own rules.
Those rules are the attribution model. Last click gives all credit to the final click; data-driven splits credit across several touchpoints. The same sale can be credited to different channels under different models.
This article explains what attribution is, which models GA4 offers, the strengths and weaknesses of last click and data-driven and — most importantly — how to present the difference to a client without jargon.
Short answer: last click answers "what was the last step before the sale?" Data-driven tries to answer "how much did each channel contribute?" Both are useful, but for different decisions.
What attribution is
Attribution is the set of rules that decides which marketing touchpoints a conversion — a purchase, lead or call — is credited to. Customers usually visit several times before buying: from ads, search, email and social.
An example journey
A user in Baku is looking for a new laptop. They see an Instagram ad and visit the site. Two days later they read a comparison article on Google and come back organically. A week later they search the brand name, click a Google Ads ad and buy.
There are three touchpoints: a Meta ad, organic search and a brand Search ad. Which channel gets the sale depends on the model.
Why it matters
The model drives budget decisions. Looking only at the last click can make channels that create first contact — social ads, video, informational content — look worthless. Cut their budget and, a while later, last-click channels lose sales too.
Key events
Since 2024 conversions in GA4 are called "key events", while Google Ads still says "conversions". Clarifying which term belongs to which system avoids confusion with clients.
Attribution models in GA4
GA4 attribution settings live under Admin → Attribution settings; see GA4 Help on attribution. In 2023 Google removed several rule-based models — first click, linear, time decay, position-based — from GA4.
| Model | How it works | Useful for |
|---|---|---|
| Data-driven | Splits credit across touchpoints based on real paths | Overall budget split, channel contribution |
| Paid and organic last click | All credit to the last paid or organic click | Last-step analysis, simple reporting |
| Google paid channels last click | Credit to the last Google Ads click | Google Ads-focused accounts |
Direct traffic
In GA4's last-click models, direct visits get no credit when another channel is on the path. If a user first came from an ad and later typed the address, the ad gets the credit.
Lookback window
The lookback window sets how far back touchpoints count. GA4 lets you choose a shorter window for acquisition key events and a longer one for others. For long decision cycles — cars, real estate, courses — a longer window is more realistic.
Changing the model changes history
When the reporting model changes in GA4, past reports are recalculated too. Document the change and tell the client in advance, or last month's numbers "changing" will erode trust.
The last-click model
Last click is the simplest, most intuitive model: whoever owns the last click before the sale gets the sale. It was the industry standard for years and many clients still read numbers this way.
Strengths
- Easy to understand and explain.
- Clearly shows the channel closest to the sale.
- Relatively easy to compare across systems.
- Can be close to reality for short decision cycles.
Weaknesses
- Barely sees channels that create first contact.
- Inflates "last step" channels such as brand search and retargeting.
- Makes upper-funnel investment — social, video, informational content — look weak without reason.
A typical trap
A team looking at last click sees Meta bringing few sales and cuts its budget. A month or two later brand searches drop, and Google Ads sales with them, because first contact came from Meta.
This does not happen in every account, but the risk is real. Before cutting budgets on last-click data, also look at data-driven and the channels' effect on the overall business.
The data-driven model
Data-driven attribution (DDA) splits credit based on your account's real data rather than a fixed rule. It compares converting and non-converting paths and estimates how much each touchpoint changes the likelihood of conversion.
A simple explanation
Imagine one group saw an Instagram ad and searched on Google, and another only searched on Google. If the first group converts noticeably more often, the model gives Instagram a share of the credit.
The method comes from game theory, but clients do not need the maths. The key idea: the model tries to answer "what would change without this channel?" from data.
Data-driven in Google Ads
Data-driven is also the default in Google Ads; see Google Ads Help on data-driven attribution. The important difference: Google Ads only sees Google ad touchpoints, while GA4 sees all channels.
Strengths
- Makes upper-funnel contribution visible.
- Adapts to the account's real data.
- Gives Smart Bidding a more balanced signal.
Limitations
- Can feel like a black box.
- Less stable with little data.
- Sees only tracked touchpoints: offline ads, word of mouth and untracked clicks are missing.
- Paths of users who decline cookie consent are not fully visible.
Data quality comes first
No model fixes bad data. Messy UTMs, duplicate key events or counted test orders make attribution wrong. In our e-commerce tracking case study excluding test traffic and setting up funnel events correctly was step one.
How to present it to a client
Talking attribution with a client is less a technical topic than a trust topic. When numbers differ, the first thought is "someone is fooling me". The goal is to remove that feeling with a clear, simple explanation.
A one-sentence explanation
"Each ad platform sees the sale through its own eyes and counts its own share. GA4 sees all channels together and splits the sale between them. That is why the platforms add up to more than GA4."
The football analogy
The player who scores is the last click. But the midfielder who passed and the defender who started the attack also contributed. Data-driven tries to value the whole team. Most clients get this immediately.
Report structure
- Pick one source of truth: CRM or the store admin for total sales and revenue.
- GA4 for channel split: data-driven, cross-channel.
- Platform reports: only for in-platform optimization.
- Explain the difference upfront: a short note at the top of the report.
A client-facing table
| Question | Where we look |
|---|---|
| How many sales this month? | CRM / store admin |
| Which channels contributed? | GA4, data-driven |
| Which Google Ads campaign works best? | Google Ads |
| Which Meta ad drove more sales? | Meta Ads Manager, its own model |
| What was the last step? | GA4, last click |
Phrases to avoid
Instead of absolutes like "Google Ads brought us 40 sales", say "Google Ads was involved in 40 sales; GA4 credits roughly this many to it". This shows value without inflating numbers.
The new-customer question
Sometimes the key question is "which channel brings new customers?" Separate new and returning users in GA4 and flag first purchases in store data. In our coupons and creators case study this split showed that 361 of 387 purchasers in two months were buying for the first time.
Practical setup steps
To talk about attribution properly, the technical base must be in order. These steps create consistency between GA4 and ad platforms.
1. UTM rules
Write one UTM convention for all paid and email campaigns: lowercase utm_source and fixed utm_medium values (cpc, paid_social, email). Mixed spellings — "Facebook", "facebook", "fb" — split one channel into three in GA4.
2. Choose key events
Make only actions with real business value key events: purchases, lead form submissions, calls. Scrolls or page views should not be key events.
3. Link Google Ads and GA4
Link the accounts, keep auto-tagging on and import key events into Google Ads when needed. Do not count the same conversion as primary via both the Google Ads tag and a GA4 import — that double counts.
4. State the model in reports
At the top of each report, note the model, lookback window and date range. This prevents confusion when comparing numbers later.
5. Quarterly review
Each quarter, review conversion paths: how many touchpoints before purchase, and which channel comes first? This gives a deeper picture for budget allocation than last click. To learn GA4 setup from scratch, see our Web Analytics Pro course.
Example: one sale under three models
A concrete example is the clearest way to show the theory. The example below is illustrative, not real account data; data-driven shares are calculated differently in every account.
The path: an Instagram ad (Meta), then organic Google search, finally a Google Ads ad on a brand query and a ₼200 purchase.
| Touchpoint | Paid and organic last click | Google paid channels last click | Data-driven (illustrative) |
|---|---|---|---|
| Meta ad | ₼0 | ₼0 | Part of the credit |
| Organic search | ₼0 | ₼0 | Part of the credit |
| Google Ads (brand) | ₼200 | ₼200 | Part of the credit |
Under last click, all ₼200 goes to Google Ads. Data-driven splits it across the three touchpoints based on how much the Meta ad and organic search change conversion likelihood on similar paths.
Meanwhile Meta Ads Manager may credit the whole sale to Meta because the user clicked within 7 days, and Google Ads reports it too, so platform totals exceed real sales.
What decision follows
Looking only at last click, Meta plays no role in this sale; data-driven shows it started the path. Budget decisions should consider both views.
Cookie consent and measurement gaps
No model can count touchpoints it does not see. Browser restrictions, ad blockers and users who decline cookies create gaps; knowing them is essential for reading numbers correctly.
Consent Mode
Google's Consent Mode adjusts tag behaviour to the user's consent choice. Without consent, tags do not set cookies, but Google may provide modeled conversions. Consent Mode v2 is required for sites serving users in the European Economic Area.
For sites serving a local Azerbaijani audience the situation may differ, but a privacy-compliant consent banner is good practice. Confirm legal requirements with a lawyer.
Server-side tracking
Meta Conversions API and server-side Google Tag Manager help recover some events lost to browser restrictions, but must not be used to bypass consent.
Showing gaps in reports
State measurement limits openly: "GA4 does not see some users who decline consent, so real sales in the CRM are higher." This explains differences and builds trust.
Offline and call conversions
In Azerbaijan many sales close by phone, WhatsApp or in store and are not tracked automatically. Recording them in the CRM and importing them into GA4 or Google Ads brings attribution closer to reality — core work in our GA4 and tracking service.
Frequently asked questions
Why do Google Ads and GA4 show different conversion counts?
Each system has its own attribution rules, tracking method and date logic. Google Ads sees only its own ad touchpoints; GA4 sees all channels. The difference is normal and should be explained in the report.
Which attribution model should I choose in GA4?
Data-driven is usually recommended for reporting because it shows channel contribution in a more balanced way. Looking at last click as well helps analyze the final step.
Does data-driven work in small accounts?
It does, but with little data results are less stable. In that case, decide over longer periods and alongside total sales data.
Why do platform-reported sales add up to more than total sales?
The same sale can be claimed by both Google Ads and Meta, since each sees itself as part of the path. Use CRM or store data for total sales.
Does changing the attribution model affect past reports?
In GA4, yes: past data is shown under the new model. Document the change and tell the client in advance.
Are offline sales included in attribution?
Not automatically. To include offline sales or calls, import them from your CRM into GA4 or Google Ads. This needs a separate technical setup.
Conclusion
Attribution is not a question of "which number is right" but "which number for which question". Last click shows the step closest to the sale; data-driven tries to value the whole path.
To build trust with clients, choose one source of truth, explain the models simply and state the rules in every report. The technical base — UTMs, key events and duplicate-free tracking — underpins all of it.
Do your GA4 and ad platform numbers not add up? Let us find the reason with a tracking audit.
See the GA4 service