Google Ads offers two attribution models today: data-driven, which is the default for most conversion actions, and last click, which gives 100% of the credit to the last ad clicked. Google retired first click, linear, time decay and position-based models in 2023. It stopped letting advertisers pick them in June 2023 and switched any conversion action still using them to data-driven in September 2023, according to its official notice. The model you choose matters more than most reports suggest, because it changes the "Conversions" column that Smart Bidding optimizes toward.
This guide covers what each remaining model does, how data-driven attribution decides credit, how to change the model step by step, what happens to Target CPA and Target ROAS when you switch, and where Google's attribution stops being useful for lead-gen teams.
Which attribution models does Google Ads offer in 2026?
Google's attribution models help page lists two models for most conversion actions. There is also an external attribution option for advertisers who calculate credit in another tool and upload it.
| Model | How credit is assigned | Status | Best fit |
|---|---|---|---|
| Data-driven (DDA) | Credit split across ad interactions based on your account's own conversion paths | Default for most conversion actions | Accounts with several campaigns that touch the same buyer |
| Last click | 100% to the last clicked ad and keyword | Available | Simple accounts, or when you need numbers that match a last-click CRM view |
| External attribution | Credit calculated outside Google Ads and imported | Available, not affected by the 2023 change | Teams with their own attribution system |
| First click, linear, time decay, position-based | Rule-based splits | Retired in 2023 | No longer selectable |
Google said fewer than 3% of Google Ads web conversions used the four retired models. So for most advertisers, the real choice is data-driven or last click. If you want a refresher on how rule-based models work in general, see our guide to multi-touch attribution models.
How does data-driven attribution decide who gets credit?
Data-driven attribution compares the paths of people who converted with the paths of people who did not. Interactions that show up more often before a conversion get more credit. The model is built for your account, not shared across advertisers.
Google's data-driven attribution page says every conversion action is eligible regardless of volume. For better accuracy, it recommends at least 200 conversions and 2,000 ad interactions in supported networks within 30 days. The model still works with less data, but credit is less precise. Supported networks include Search (with Shopping), YouTube, Display and Demand Gen.
Two limits are worth knowing:
- It only sees Google ad interactions. A Meta ad, a WhatsApp conversation or a sales call that happened before the Google click gets no credit inside Google Ads.
- It only sees the conversions you send. If your conversion is "form submitted", data-driven attribution optimizes toward form submissions, not toward sales.
Data-driven vs last click: what actually changes in your reports?
Switching from last click to data-driven does not create or remove conversions. It moves credit between campaigns, ad groups and keywords. Google notes three visible effects:
- Decimals appear in "Conversions" because credit is split.
- Upper-funnel campaigns, such as broad or generic keywords, YouTube and Demand Gen, often gain credit. Brand search often loses some.
- Recent days can show a temporary dip, because conversions are reported by interaction date and the time lag can grow.
Example (hypothetical numbers): a B2B software company gets 100 demo requests a month from Google Ads. Under last click, brand search gets 60 and a generic "crm software" campaign gets 40. After switching to data-driven, the same 100 demos split 48 to brand and 52 to generic, because many buyers clicked a generic ad first and searched the brand name later. Spend is $3,000 on brand and $6,000 on generic.
| Campaign | Spend | Conversions (last click) | CPA (last click) | Conversions (DDA) | CPA (DDA) |
|---|---|---|---|---|---|
| Brand search | $3,000 | 60 | $50 | 48 | $62.50 |
| Generic search | $6,000 | 40 | $150 | 52 | $115.38 |
| Total | $9,000 | 100 | $90 | 100 | $90 |
The account CPA stays at $90. What changes is which campaign looks efficient, and therefore where Smart Bidding and your budget decisions push money.
How does the attribution model affect Smart Bidding?
The attribution model applies to the conversion action, and bid strategies use the "Conversions" column built from it. That means Target CPA, Target ROAS, Maximize conversions and Maximize conversion value all bid according to the model you chose. Google's page on managing attribution model changes warns that if you switch models and keep the old targets, you may overbid on some campaigns and underbid on others.
Google's recommended adjustment is simple. Compare "Cost / conv." with "Cost / conv. (current model)" for a recent stable period, then move each campaign's Target CPA by the same percentage. For Target ROAS, do the same with "Conv. value / cost". In the example above, the generic campaign's CPA moves from $150 to about $115, a drop of roughly 23%. If its Target CPA was $150, the matching new target is about $115.
Google also advises excluding the last few weeks from that comparison (to avoid conversion lag) and waiting at least the "average days to conversion" before judging the result.
How do you change the attribution model in Google Ads?
- Click the Goals icon in your Google Ads account.
- Open the Conversions menu and select Summary.
- Click the conversion action you want to change.
- Click Edit settings.
- Open the Attribution model dropdown and choose Data-driven or Last click.
- Click Save, then Done.
- Before the change, export "Cost / conv." and "Conv. value / cost" for a stable two-week period so you can recalculate targets.
- After the change, update Target CPA or Target ROAS by the percentage difference, then leave the campaigns alone through the learning period.
If the conversion is imported from Google Analytics 4, check the GA4 attribution settings too, because the property's settings affect what gets imported. Our post on GA4 attribution settings mistakes covers the common traps, and attribution windows across Meta, Google and GA4 explains why the lookback window changes totals too.
Should you choose data-driven or last click?
For most accounts, keep data-driven. It is the default, Google's bidding is designed around it, and it gives some credit to the generic and video campaigns that start demand. Choose last click when:
- The account has one campaign or one funnel step, so there is little to split.
- You need Google Ads numbers to reconcile with a last-click report from finance or your CRM, at least temporarily.
- You are running a clean test and want a fixed rule while you measure incrementality another way.
Whichever you choose, do not switch back and forth. Each switch moves credit, changes the targets that bidding needs, and makes month-over-month comparisons harder.
Where does Google Ads attribution stop being enough?
Any in-platform model answers a narrow question: which Google ad interactions deserve credit for conversions Google can see. For a lead-gen or sales-led business, that leaves three gaps.
- Other channels. Google cannot see the Meta video, the TikTok ad or the Click-to-WhatsApp conversation that also touched the buyer. Every platform credits itself, which is why the totals rarely add up. We explain the pattern in why ROAS is lying.
- Lead quality. If the conversion is a form fill, both models reward cheap forms. Sending qualified-lead and closed-won stages back with offline conversion imports fixes what bidding optimizes toward.
- Revenue. Google Ads does not know which deals closed, at what value, weeks later.
This is the gap a cross-channel tool fills. Admira connects Meta, Google Ads, TikTok and Click-to-WhatsApp leads to revenue in your CRM, so you can compare a Google campaign's cost per sale with Meta's on the same basis, and its Conversion API sends CRM events back to the ad platforms server-side. It does not replace Google's bidding model, and it is not a marketing mix modeling consultancy. See how multi-touch attribution software compares channels outside any single platform.
When this doesn't apply
- Very small accounts. With a handful of conversions a month, the difference between models is noise. Fix conversion tracking first.
- App campaigns and some special conversion types. Available models can differ by conversion source, so check the dropdown for each action.
- Pure ecommerce with online checkout. Shopify-style stores may get more from an ecommerce attribution tool than from a lead-to-revenue setup.
FAQ
Is last click still available in Google Ads?
Yes. Last click remains available alongside data-driven. Only first click, linear, time decay and position-based were retired in 2023.
What happened to conversion actions that used linear or time decay?
Google switched them to data-driven attribution in September 2023. You can change any of them to last click if you prefer.
How many conversions does data-driven attribution need?
None as a hard minimum. Google recommends about 200 conversions and 2,000 ad interactions in 30 days for more accurate credit.
Will changing the model reset Smart Bidding?
It changes the conversion data bidding uses, so expect an adjustment period. Update Target CPA or Target ROAS by the percentage difference Google shows in the "(current model)" columns.
Does data-driven attribution include Meta or TikTok ads?
No. Google Ads data-driven attribution only distributes credit across Google ad interactions it can observe.
Can I compare models before switching?
Yes. The "(current model)" columns and the attribution reports let you see how cost per conversion would shift before you commit.
Which model should a lead-gen business use?
Usually data-driven, combined with importing qualified leads and sales from your CRM so the model optimizes toward revenue, not form fills.
.svg)
%20(2).png)
