GA4 attribution models in South Africa are the reason most SA SME marketing reports quietly lie. A campaign that looks like it produced 40 leads on the Google Ads dashboard produces 22 in GA4 and 34 in a report an agency sends by email. All three are technically correct. They are just using different rules to answer “which channel gets the credit for this sale”.
This piece is the honest, non-technical version: what each GA4 attribution model actually does, which one to pick for what decision, and how to compare paid channels without funding the wrong one.
What GA4 attribution actually is
Every conversion has a story. Someone saw a Meta ad on Monday, ignored it. Searched your brand on Wednesday, clicked an organic result, left. Saw a Google Search ad on Saturday, clicked, converted. Who gets the credit for that sale? All three touchpoints played a role. An attribution model is the rule that decides how much credit each gets.
Universal Analytics had six models to pick from. GA4 has effectively two: data-driven and paid/organic last-click, plus a handful of cross-channel models available in the reporting comparison views but not in the main reports.
The GA4 attribution models available in 2026
| Model | How it splits credit | Best used for |
|---|---|---|
| Data-driven (DDA) | Machine learning on your actual conversion paths; each channel gets fractional credit based on its measured contribution to conversions. | Cross-channel comparison; the default in GA4 for most SA businesses with enough data. |
| Paid and organic last-click | Last non-direct paid or organic click gets 100% credit; direct visits get nothing unless there was no other touchpoint. | When DDA is unavailable, or as a sanity-check against DDA. |
| Google Paid Channels last-click (comparison view) | 100% credit to the last Google-paid channel in the path. | Comparing Google Ads campaigns to each other, in isolation. |
| First click (comparison view) | 100% credit to the first channel in the path. | Understanding brand and awareness channels. |
| Linear (comparison view) | Equal credit split across all touchpoints in the path. | Rough view of channel contribution when DDA data is thin. |
| Time decay (comparison view) | More credit to touchpoints closer to the conversion. | Short sales cycles where recency matters most. |
| Position-based (comparison view) | 40% to first, 40% to last, 20% split across the middle. | Businesses where both discovery and closing channels matter separately. |
Which model to pick for what decision
The right model depends on the question you are actually answering:
- “Which channel should I put more budget into?” Data-driven attribution. It is the fairest cross-channel comparison GA4 offers.
- “How is my Google Ads doing in isolation?” Use the Google Ads dashboard (which uses its own DDA or last-click) alongside GA4 DDA. Compare, do not blend.
- “Is my Meta retargeting doing anything?” Meta Ads Manager attribution for platform decisions; GA4 DDA for the honest total.
- “Where do new customers first find me?” First-click model in a GA4 comparison view. Shows discovery channels honestly.
- “Which touchpoint closes the sale?” Last-click or time-decay. Best for short cycles like local services.
- “Should I run brand campaigns?” Compare DDA vs last-click. If brand shows in DDA but not last-click, it is doing the assisting work last-click hides.
Why the numbers never match across platforms
Google Ads and Meta Ads Manager will always show more conversions than GA4. Reasons that show up on almost every SA account:
- Attribution windows differ. Meta looks back 7 days on clicks, 1 day on views. Google Ads defaults to 30. GA4 uses the lookback window you configured (default 90 days for acquisition, 30 for conversion).
- Each platform counts conversions from touchpoints it saw. Meta cannot see the Google Ads click that came 2 days later; Google Ads cannot see the Meta view that came before.
- Dedup logic. GA4 deduplicates properly across channels; the platforms do not.
- Modelled conversions. Both Google Ads and Meta now use modelled data to fill iOS and cookie-limited gaps; GA4 does its own modelling separately.
Small differences (10 to 20%) are normal and expected. Gaps of 30% or more usually mean a broken tag, a missing enhanced conversions setup on Google Ads, or a Meta CAPI that never actually got wired up. The GA4 WordPress conversion tracking mistakes post covers the recurring reasons in detail.
A working setup for SA SMEs
What we actually run for most SA clients:
- GA4 attribution: Data-driven, kept as the default.
- Google Ads: Data-driven (its own), with enhanced conversions turned on.
- Meta: Manual retargeting rules with the Conversions API firing purchase and lead events with a real value, not R0.
- Weekly report: DDA numbers from GA4 for the cross-channel view; platform numbers for platform-level tuning; a plain paragraph explaining any gap over 20%.
For the wider view on marketing attribution and what SA SMEs get wrong reading it, we covered it in where your leads are really coming from. This piece is the technical GA4 side, so you can actually make sense of what the numbers are telling you.
