The short answer
For 95% of UK Google Ads accounts in 2026, the answer is data-driven attribution (DDA). It's the default for new conversion actions, it's required by most Smart Bidding strategies, and Google has deprecated nearly every other option. The remaining 5% who should still consider last click are accounts with under 300 conversions/month, single-touch funnels (one click, one purchase), or businesses where strict media-mix modelling sits outside Google Ads.
The bigger mistake isn't picking the wrong model - it's switching models too often. Each change triggers a 14-21 day Smart Bidding learning period. Pick a model based on your funnel and stick with it for at least 90 days before evaluating.
What attribution actually means in Google Ads
Attribution is how Google Ads decides which click to credit when a user converts after multiple ad interactions. Imagine a user clicks a Search ad on Monday, a Shopping ad on Wednesday, and a brand search ad on Friday before buying. Three clicks, one conversion. Which click "caused" the sale?
The attribution model is Google Ads' answer to that question. The chosen model determines:
- How conversions are split between campaigns in reporting
- Which keywords look profitable vs unprofitable
- Which campaigns Smart Bidding allocates more budget to
- How tCPA and tROAS targets are calculated
Get this wrong and you'll defund the campaigns that were doing the upper-funnel heavy lifting, leaving only the bottom-funnel brand campaigns that grab the credit at the end.
The Google Ads attribution models in 2026
As of 2026, Google Ads supports two attribution models for cross-channel conversions:
| Model | How it works | Best for |
|---|---|---|
| Data-driven (DDA) | Google ML splits credit fractionally across all touchpoints based on observed conversion patterns | Almost all accounts with 300+ conversions/month |
| Last click | 100% credit goes to the final click before conversion | Single-touch funnels, low-volume accounts |
That's it. Google deprecated First Click, Linear, Time Decay and Position-Based attribution between September 2023 and September 2024, removing them from the UI entirely. See Google's official attribution model documentation for the latest list.
Data-driven attribution explained
Data-driven attribution uses Google machine learning to assign fractional credit to every touchpoint in the conversion path, based on the actual contribution each click made. The model compares converting paths to non-converting paths and learns which combinations of clicks reliably move users towards conversion.
Practical example: a user sees a Display ad (no click), clicks a Search ad on a generic keyword, clicks a Shopping ad three days later, then clicks a brand Search ad and buys. DDA might credit:
- Search (generic): 0.35 conversions
- Shopping: 0.45 conversions
- Search (brand): 0.20 conversions
Last click would have given the brand Search 1.0 conversions and the others 0.0 - making generic Search and Shopping look unprofitable when in reality they did most of the work.
What DDA needs to work
- Account-wide volume of 300+ conversions in the trailing 30 days (Google's published threshold)
- Conversions need to span multiple campaigns/channels - DDA can't learn from a single touchpoint
- Conversion tracking must be reliable - DDA amplifies bad signals just like Smart Bidding
Below 300 conversions, Google falls back to last click automatically and shows a warning in the conversion action settings.
When last click still wins
Three scenarios where last click is the right call in 2026:
1. Low-volume accounts
Under 300 conversions/month account-wide, DDA doesn't have enough data to model paths. Last click is more predictable, less noisy. Once you cross 500-1,000 conversions/month, switch.
2. Single-touch funnels
Some businesses genuinely have one-click-one-purchase patterns. Impulse-buy ecommerce under £50, emergency services (locksmith, plumber), most B2C lead gen for transactional services. If 80%+ of converting users have only one ad click in their path, DDA and last click produce nearly identical results - last click wins for simplicity.
3. You're running rigorous media-mix modelling outside Google Ads
Larger advertisers (£100k/month+) often run incrementality tests and MMM (media-mix modelling) to determine true contribution. In that setup, Google Ads' attribution becomes secondary - last click is fine because the real attribution lives in another system.
The deprecated models (don't use, can't use)
Between 2023 and 2024 Google removed these from the Google Ads UI:
- First click - 100% credit to first click. Deprecated September 2023.
- Linear - equal credit to every click. Deprecated September 2023.
- Time decay - more credit to clicks closer to conversion. Deprecated April 2024.
- Position-based - 40% to first, 40% to last, 20% spread across the middle. Deprecated September 2024.
Why deprecated? Google's argument: these are rule-based models that ignore actual user behaviour. DDA learns from real data and consistently outperforms rule-based models in tests. The counter-argument: Google removed transparency and forced everyone into a black box. Both are partly true. Either way, the choice is now binary.
You may still see deprecated models referenced in older blog posts, third-party reporting tools (e.g. Looker Studio templates), or GA4 - which retains a wider model selection. In Google Ads itself, only data-driven and last click remain.
How attribution affects Smart Bidding
Attribution choice directly drives Smart Bidding behaviour. Smart Bidding uses the attributed conversion value per click to set bids - so if attribution shifts credit, bids shift too.
Switching from last click to DDA typically:
- Increases bids on upper-funnel keywords (generic Search, Shopping non-brand)
- Decreases bids on bottom-funnel keywords (brand Search, exact-match high-intent)
- Spreads spend more evenly across the funnel
- Improves overall conversion volume by 6-10% on average (Google's published data)
The shift can feel uncomfortable - "why is brand spend dropping while generic spend grows?" - but it usually reflects reality: the generic clicks were always doing more work than last click gave them credit for.
Best attribution model by business type
| Business type | Recommended model |
|---|---|
| Ecommerce £100+ AOV | Data-driven (DDA) |
| Ecommerce impulse buy under £50 | Last click acceptable; DDA if volume allows |
| B2B lead gen, considered purchase | DDA + offline conversion import |
| Emergency services (locksmith, plumber) | Last click |
| SaaS with free trial | DDA on trial signup; OCI on paid conversion |
| Lead gen, low volume (under 300/month) | Last click until volume scales |
How to switch attribution models safely
If you're moving from last click to DDA (or vice versa), follow this process:
- Verify volume - confirm 300+ conversions in the last 30 days account-wide for DDA. Google blocks the switch if you're under threshold.
- Note current performance - export the last 30 days of campaign performance as a baseline.
- Switch the conversion action - Google Ads → Goals → Conversions → click the action → Edit settings → Attribution model.
- Don't change anything else for 21 days - no budget changes, no new campaigns, no bid strategy switches. Smart Bidding needs a clean learning period.
- Evaluate at day 30 - compare cost per conversion, conversion volume and revenue. Expect ±10% volatility in week one.
- Evaluate again at day 90 - DDA's full benefit shows over a longer window. Snap judgements at week 2 are wrong.
The most common mistake: switching back and forth in panic. Each switch triggers another learning period. Pick one, commit for 90 days.
Google Ads attribution vs GA4 attribution
GA4 has its own attribution settings, separate from Google Ads. By default GA4 uses data-driven attribution for "Google Paid" channels and last click for "all channels". You can change GA4's settings under Admin → Attribution settings.
Critical: Google Ads attribution is what drives bidding. GA4 attribution is what populates the GA4 reports you look at separately. They can - and often do - show different conversion numbers for the same period. That's normal. Don't try to force them to match.
If you import GA4 conversions into Google Ads, the GA4 attribution model affects what Google Ads sees. Use the same model in both places (DDA in both, or last click in both) for consistency.
