GA4 Removed Your Attribution Models. Here's What to Do Now.

Looking for a tool-by-tool comparison? See 7 Best GA4 Attribution Alternatives.

What Google Took Away

In late 2023, Google removed four attribution models from GA4:

• First-click — 100% credit to the first touchpoint
• Linear — equal credit across all touchpoints
• Time-decay — more credit to recent touchpoints
• Position-based — 40% first, 40% last, 20% middle

What's left:

• Last-click — 100% credit to the final click before conversion
• Data-Driven Attribution (DDA) — algorithmic credit using Shapley values

Google's justification: the removed models accounted for less than 3% of conversions. Translation: most people were already using DDA or last-click. Google decided the other models weren't worth maintaining.

The Problem Google Didn't Mention

Does GA4 Data-Driven Attribution actually work? That depends on your traffic. What the announcement didn't say: DDA requires a minimum of 400 conversions per conversion type in a 28-day window to function.

Below that threshold, GA4 silently falls back to last-click.

No warning. No notification. No "your data-driven model couldn't run this period." It just quietly switches to last-click and reports numbers as if everything is fine.

Some sources cite 600+ conversions and 15,000+ clicks for DDA to work reliably. Google's own documentation (buried in help articles) confirms the 400-conversion minimum.

Who This Affects

Let's do the maths.

400 conversions in 28 days = roughly 14 conversions per day.

If your conversion rate is 2-3% (typical for B2B and mid-market):
- 14 conversions/day ÷ 2.5% conversion rate = 560 visitors per day minimum
- That's ~17,000 monthly sessions with a single conversion type

If you're spending $60K-$500K/year on ads and have a typical conversion rate, you're probably below the threshold for at least some of your conversion types. Lead form submissions, demo requests, trial signups—each one needs to independently hit 400/month.

The businesses most needing accurate attribution are structurally excluded from GA4's best model.

Companies spending $2M+/year? They clear the threshold easily. They also have budget for Haus, Measured, or Northbeam. GA4's DDA is a feature for companies that already have better options.

What "<3% of Conversions" Actually Means

Google's claim that the removed models represented "less than 3% of conversions" deserves scrutiny.

That statistic is across all GA4 properties globally. It includes:

• Massive e-commerce sites with millions of conversions (DDA works well for them)
• Properties that never changed from the default model
• Properties where the admin didn't know other models existed

It does not tell you whether YOUR team was using those models. If you were in that 3% — maybe you ran linear for baseline measurement, or position-based for your demand gen team — you lost a tool you relied on.

"Almost nobody used it" is Google's reason. "I used it and now I can't" is your problem.

Your Three Options

Option 1: Keep GA4, Accept the Limitations

Cost: $0
Effort: None
Best for: Companies with 400+ monthly conversions per type

If you clear the DDA threshold, GA4's remaining model is decent. It uses Shapley values — game theory that calculates each channel's marginal contribution. It's not perfect (documented bias toward Google channels), but it's better than last-click.

The trade-off: you can't compare models. First-touch told you what generated awareness. Time-decay told you what closed deals. DDA gives you one blended answer. If that's enough, stay.

What to watch for:
- DDA's documented bias toward Google properties (organic search, Google Ads)
- No visibility into whether DDA is actually running or falling back to last-click
- 90-day maximum lookback window (too short for B2B sales cycles)

Option 2: Keep GA4 for Analytics, Add Independent Attribution

Cost: $0-299/mo
Effort: 1-2 hours setup
Best for: Most companies, especially those below 400 conversions/month

This is what we recommend for most teams. GA4 is excellent at what it does — web analytics, traffic analysis, audience insights, Google Ads integration. Those features still work perfectly.

What GA4 can't do well is cross-channel attribution. It can't see your email performance fairly, it favours Google channels, and below 400 conversions it's just last-click.

Add a dedicated attribution tool alongside GA4:

Full disclosure: mbuzz is our product. We've listed alternatives at every price point because the right tool depends on your stack, not our revenue.

Option 3: Build Your Own in BigQuery

Cost: BigQuery compute costs (~$5-50/mo for most)
Effort: 20-40 hours engineering time
Best for: Teams with data engineers and specific model requirements

GA4 exports raw event data to BigQuery. You can rebuild any attribution model in SQL:

-- Simple first-touch attribution in BigQuery
SELECT
  FIRST_VALUE(source) OVER (
    PARTITION BY user_pseudo_id
    ORDER BY event_timestamp
  ) AS first_touch_source,
  COUNT(DISTINCT user_pseudo_id) AS conversions
FROM `your_project.analytics_XXXXX.events_*`
WHERE event_name = 'purchase'
GROUP BY first_touch_source

This gives you complete control, but you'll need a data engineer and custom visualisation. GA4 to BigQuery has a 24-48 hour delay.

If you have the engineering capacity, this is the most flexible option. If you don't, Option 2 is faster and cheaper than hiring for it.

Run Multiple Models in Parallel

The reason Google removing models matters isn't that any one model was perfect. It's that comparing models reveals truth.

When first-touch and last-touch disagree about a channel, that disagreement is information. It tells you the channel works differently at different funnel stages. Linear attribution giving a channel 15% credit while position-based gives it 35% tells you something.

One model gives you a number. Multiple models give you a range. The range is closer to reality.

What Should Change in Your Workflow

Regardless of which option you choose, the GA4 model removal forces two workflow changes:

1. Stop Using GA4 for Budget Allocation

GA4 should inform within-Google optimisation — which campaigns work, which keywords convert, which landing pages perform. It should not be the sole input for "should we spend more on Meta or Google?"

For cross-channel budget decisions, use a tool that sees all channels neutrally.

2. Check Whether DDA Is Actually Running

For each conversion type in your GA4 property, check:
- How many conversions in the last 28 days?
- Is it above 400?
- If not, you're on last-click whether you know it or not

There's no UI indicator for this. You have to check the numbers yourself.

Related Reading

• Why Did GA4 Remove Most Attribution Models? — the technical deep dive on what changed and why
• 7 Best GA4 Attribution Alternatives — full comparison of replacement tools
• How Much Are Your Ad Platforms Over-Reporting? — the data on platform attribution inflation

Holly Henderson

Co-Founder, mbuzz

Holly Henderson is Co-Founder of mbuzz. With 10+ years in marketing including roles at Westpac, Avon, and Forebrite, she's obsessed with making measurement actually useful.

Harvard Extension School Forebrite Westpac Avon