You are investing in upper-funnel campaigns, but branded search keeps taking most of the conversion credit. Does that mean the earlier ads contributed nothing, or does the attribution model simply favor the final interaction? The answer matters, because attribution in Google Ads shapes more than a report. The model you select for a conversion action decides how Google distributes credit across eligible ad interactions, and that data lands in your conversion columns, where it can steer Smart Bidding, campaign evaluation, and budget decisions.

Google Ads now offers two attribution models for most eligible conversion actions, Data-driven and Last click. Choosing between them takes understanding what each one measures, how it affects bidding, and what neither one can prove.

Key Takeaways

Google Ads offers Data-driven attribution and Last click for eligible conversion actions.
Data-driven attribution estimates how credit should be distributed across eligible Google Ads interactions. It does not prove that an interaction caused the conversion.
The attribution model affects conversion reporting and any Smart Bidding strategy that uses the selected conversion action.
The Model comparison report shows how credit shifts between the two models, but it is not an incrementality test.
For businesses that generate leads or close sales offline, CRM and offline-conversion data often matter more than model selection.

What an Attribution Model Does

An attribution model is the rule or modeling system that decides how conversion credit is assigned to ad interactions along a measurable path. Picture a prospect who engages with a Google Ads YouTube ad, later clicks a non-brand Search ad, sees or clicks a Display remarketing ad, and finally converts after clicking a branded Search ad.

Under Last click, the final eligible Google Ads interaction receives all the credit. Under Data-driven attribution, Google may divide the credit fractionally among eligible interactions based on patterns it observes across converting and non-converting paths. That difference can materially change how campaigns look in reports, and it can change the conversion data that automated bidding relies on.

Attribution is not the same as incrementality, though. A model allocates credit among interactions it can observe. It does not build a control group or establish what would have happened without the advertising. Even Data-driven attribution is best read as a modeled estimate, not a causal measurement of each ad’s true contribution.

Which Google Ads Attribution Models Are Available in 2026?

Google Ads supports two models for eligible conversion actions: Data-driven attribution and Last click. Google began retiring First click, Linear, Time decay, and Position-based attribution in 2023 because those rules-based models were used for only a small minority of web conversions, and it explains the change in its notice about deprecated attribution models.

The previous models worked like this:

First click assigned all credit to the first interaction.
Linear divided credit equally among interactions.
Time decay gave more credit to interactions closer to the conversion.
Position based assigned 40 percent each to the first and last interactions and split the remaining 20 percent among the interactions in between.

These remain useful concepts for thinking about attribution, but they are no longer selectable for Google Ads conversion actions.

Data-driven attribution

Data-driven attribution uses Google’s machine-learning model to evaluate eligible conversion paths and estimate how credit should be divided among ad interactions. Google describes the model in its documentation on Data-driven attribution. Rather than following a fixed rule, it looks for patterns in paths that convert and paths that do not, then assigns fractional credit to the interactions associated with movement toward a conversion.

One conversion might appear as:

0.25 conversions for a YouTube interaction;
0.35 for a non-brand Search interaction;
0.15 for a Display interaction; and
0.25 for a branded Search interaction.

Those fractions represent Google’s modeled allocation. They should not be read as experimental proof that each interaction caused exactly that share of the outcome.

Does Data-driven attribution require a minimum number of conversions?

Not as a general rule. Google has expanded Data-driven attribution to most supported conversion actions regardless of their individual conversion or interaction volume, and DDA is the default for most new conversion actions. However, Google notes that some conversion-action types still require 3,000 ad interactions and 300 conversions within 30 days to become eligible, followed by 2,000 interactions and 200 conversions every 30 days to remain eligible. So the old 200/2,000 figures are not a universal rule for choosing a model in 2026, but they have not vanished either; they still apply to certain action types.

Volume still matters to interpretation. When account-specific evidence is thin, any model has less to work with, and the model leans more on broader aggregated patterns. That does not automatically make Last click more accurate. It means you should avoid treating a modeled allocation as precise causal evidence.

Advantages of Data-driven attribution

It can distribute credit beyond the final eligible interaction.
It may surface campaigns or keywords that Last click consistently undervalues.
It gives Smart Bidding a broader view of eligible interactions in the path.
It adapts to observed account and ecosystem patterns rather than one fixed rule.

Limitations of Data-driven attribution

Its calculations are less transparent than a fixed rule.
It can only evaluate interactions Google can observe and connect.
It does not capture every offline, organic, competitor, email, or person-to-person interaction.
It estimates attribution; it does not measure incrementality.

Last click attribution

Last click gives 100 percent of the credit to the final eligible Google Ads interaction before the conversion. Its main strength is interpretability: teams can see exactly why a campaign received credit, and the calculation does not depend on a less-visible model. Its main weakness is just as plain, because earlier interactions receive no credit even when they introduced the brand, educated the buyer, or brought the buyer back. That can make branded search and other closing interactions look disproportionately valuable.

Last click may suit organizations that deliberately want a simple closing-touch rule or need continuity with an established reporting method. But simplicity is not the same as accuracy. A short sales cycle or low conversion volume does not prove that the last interaction created the entire result.

Criterion	Data-driven attribution	Last click
Credit distribution	Fractional credit based on Google’s model	100 percent to the final eligible interaction
Method	Modeled from observed paths	Fixed rule
Interpretability	Lower	High
Earlier interactions	May receive credit	Receive no credit
Main limitation	Model is not fully transparent and is not causal	Systematically ignores earlier interactions

How to Choose Between Data-driven and Last click

For many advertisers, Data-driven attribution is the more informative starting point, because it does not automatically erase every interaction before the final click. It is especially relevant when campaigns work together across Search, YouTube, Display, Demand Gen, or other supported Google inventory.

Last click may still be appropriate when:

the business explicitly wants a simple closing-touch rule;
stakeholders need continuity with historical Last-click reporting;
a team wants an easily auditable internal convention; or
the model is used as one reporting view alongside broader measurement.

Before you change the setting, ask:

Are the relevant conversion actions configured and categorized correctly?
Are primary and secondary conversions being used intentionally?
Does the conversion represent a meaningful business outcome?
Does the conversion window reflect the normal time to purchase or lead completion?
Which bidding strategies use this conversion action?
How will stakeholders interpret fractional credit after the switch?

The model cannot compensate for incorrect tagging, duplicate conversions, poor lead definitions, missing offline outcomes, or values that do not reflect the business.

How Attribution Affects Smart Bidding

The selected attribution model affects how a conversion action is counted in the Conversions and All conversions columns. When that action feeds bidding, the credit distribution becomes part of the signal Smart Bidding works from. Current strategies include Target CPA, Target ROAS, Maximize conversions, and Maximize conversion value.

Enhanced CPC is no longer available for Search and Display campaigns. Google discontinued it in March 2025, and campaigns that were not proactively migrated now effectively use Manual CPC.

Switching from Last click to Data-driven can shift credit toward earlier eligible interactions. Campaigns and keywords that looked like weak contributors may start receiving fractional credit, while closing interactions receive less. That does not mean business performance changed at the moment of the switch; the immediate reporting difference is mostly a reallocation of credit. Over time, Smart Bidding may respond to the new distribution and make different auction-level decisions.

Google recommends reviewing bids and targets when you change attribution models, since existing targets were set against the old credit distribution. Evaluate the change over a period long enough to cover your normal conversion delay and produce a representative number of outcomes. A universal two-week or four-week rule does not fit every account.

Using the Model Comparison Report

Google Ads includes a Model comparison report that sets Last click and Data-driven side by side. Google documents it in its guide to attribution models. To open it, go to Goals, select Attribution, choose Model comparison, pick a dimension such as campaign, ad group, keyword, or device, and compare the two models.

The Cost per conversion and Conversion value per cost columns show how CPA and ROAS look under each allocation method, which helps identify campaigns or keywords that Last click undervalues. But the report answers a narrow question: how would Google distribute the same conversion credit under two models? It does not show whether the ads caused additional conversions, and a higher modeled ROAS does not by itself prove greater business impact. Incrementality requires a credible counterfactual, such as a conversion-lift study, holdout, geo experiment, or matched-market test.

How to Change the Attribution Model

Google’s current interface steps are:

Go to Goals.
Open Conversions, then Summary.
Select the conversion action you want to edit.
Choose Edit settings.
Open Attribution model.
Select Data-driven or Last click.
Save the change.

The setting applies to the selected conversion action. If you use cross-account conversion tracking, manage the model at the relevant manager-account level. Changing it affects the main conversion columns going forward and does not rewrite historical columns; Google provides current-model reporting columns so you can view how past conversions would look under the currently selected model. Before switching, document the date, active bidding strategies, targets, conversion windows, and recent performance, which makes the later reporting changes much easier to interpret.

Conversion Windows and Attribution

A conversion window defines how long after an eligible ad interaction a conversion can still be credited. The right window depends on the action and the normal decision cycle: a product bought within hours needs a shorter window than a B2B opportunity that takes weeks to reach a meaningful stage. Longer windows capture more delayed outcomes but can also tie conversions to interactions that happened much earlier.

Review your time-to-conversion data before choosing a window. Google explains how to examine this delay in its guide to how long customers take to convert. Keep the two settings distinct: the window determines which interactions and conversions remain eligible, while the attribution model determines how credit is distributed among the eligible ones. Both should reflect the real customer journey.

Google Ads Attribution and GA4 Are Not Identical

Google Ads and Google Analytics 4 both use Data-driven and Last-click approaches, but their settings and reporting scopes should not be collapsed into one system. In GA4, attribution can use paid and organic channels or Google paid channels depending on the selected conversion and reporting settings, and GA4 distinguishes event-scoped attribution from the acquisition logic used for user- and session-scoped reporting. Google explains these choices in its documentation on GA4 attribution settings. When Google Analytics conversions are shared with Google Ads, review the settings in both products so you understand why the reported totals or channel credit may differ. GA4 can give a wider view of measurable paid and organic interactions, but it still does not observe every influence on the customer or prove causal impact.

B2B Attribution: Optimize for Outcomes Beyond the Form Fill

For B2B advertisers, choosing between Data-driven and Last click is rarely the most important measurement decision. A platform may see the click and the lead form, while the CRM holds the outcomes that matter: qualified opportunity, sales acceptance, pipeline value, closed revenue, or disqualification. If bidding optimizes toward every form submission equally, it can generate more leads without generating better business.

Google Ads supports offline conversion imports and enhanced conversions for leads to connect eligible ad interactions with later outcomes. (Note that Google is combining enhanced conversions for web and leads into a single on/off setting in June 2026, so the setup path is changing.) A stronger B2B measurement structure usually includes consistent lead identifiers and consented first-party data, CRM stages with clear definitions, offline conversion actions for meaningful downstream outcomes, appropriate values for different stages, deduplication and data-quality checks, and conversion delays reflected in both reporting and bidding.

This will not produce a perfect view of the buying committee. Multiple people, offline conversations, dark social, organic research, and untracked interactions can still influence a deal. But it moves optimization closer to revenue than treating every form fill as equal.

Common Attribution Mistakes

Treating modeled credit as causal proof. DDA estimates how to allocate credit. When the question is whether advertising created incremental outcomes, use experiments.

Changing models before fixing conversion tracking. No model repairs duplicated tags, missing values, irrelevant primary conversions, or unqualified leads.

Comparing Google Ads and GA4 without aligning scope. The products can use different dates, channels, conversion settings, and attribution logic, so a discrepancy does not automatically mean either platform is broken.

Evaluating campaigns only through attributed ROAS. Margin, lead quality, pipeline, new-customer value, and incrementality can all change how you read an apparently strong ROAS.

Assuming Last click is more accurate because it is simpler. Last click is more transparent, but it encodes a strong assumption: every interaction before the final eligible click gets no credit.

Frequently Asked Questions

Which attribution models are available in Google Ads in 2026?

For most eligible conversion actions, Google Ads offers Data-driven attribution and Last click. First click, Linear, Time decay, and Position based were deprecated in 2023.

Does Data-driven attribution require 200 monthly conversions?

There is no universal 200-conversion rule for choosing a model in 2026. Google made DDA available to most eligible conversion actions regardless of volume, though certain conversion-action types still have eligibility thresholds. Lower volume still calls for caution when interpreting modeled credit.

Does changing the attribution model affect Smart Bidding?

Yes, when the conversion action is included in bidding. The model changes how credit is distributed in the conversion data available to strategies such as Target CPA, Target ROAS, Maximize conversions, and Maximize conversion value.

Does Google Ads DDA measure every marketing touchpoint?

No. It distributes credit among eligible interactions that Google can observe and connect. It does not automatically include every organic, offline, email, competitor, or person-to-person influence.

Can the Model comparison report prove that an upper-funnel campaign works?

No. It can show that Data-driven attribution assigns the campaign more credit than Last click. Proving incremental impact requires an experimental or otherwise credible causal design.

Should low-volume accounts use Last click?

Not automatically. Last click may be chosen for simplicity or reporting continuity, but low volume does not make its assumption more accurate. Review tracking quality, business needs, and how the output will be used.

Choose a Model, Then Validate the Business Impact

Data-driven attribution is a useful default for many Google Ads accounts, because it can recognize eligible interactions that Last click ignores. Last click stays valuable when a business deliberately wants a simple closing-touch convention. Neither model provides a complete or causal view of the customer journey, so the more important task is making sure Google Ads receives accurate, meaningful conversion data and that attribution reports are read alongside commercial outcomes and incrementality testing.

Black Propeller provides Google Ads management services across campaign strategy, conversion measurement, bidding, and performance analysis. If your account is optimizing toward the wrong conversions, or your CRM outcomes are disconnected from media reporting, contact the Black Propeller team for an assessment of the measurement structure behind your campaigns.

Attribution Models for Google Ads: A Practical 2026 Guide