Last Click vs Data-Driven Attribution: Which Model Should You Use for PPC?

An attribution model defines how conversion credit is distributed between different touchpoints along the purchase path.

Example: a user sees a display ad (touchpoint 1), clicks on a Shopping ad (touchpoint 2), returns via a remarketing ad (touchpoint 3), and purchases. Who gets credit for the sale?

- **Last Click**: 100% to remarketing (point 3) - **First Click**: 100% to display (point 1) - **Linear**: 33% to each touchpoint - **Time Decay**: more credit to recent touchpoints (remarketing > Shopping > display) - **Data-Driven**: % calculated by AI based on account historical data

**Why it's critical for your campaigns:** Google Ads and Meta use your attribution model to optimize bids. If you use Last Click, the algorithm will over-invest at the bottom of funnel (remarketing, branded) and under-invest at the top (discovery, display, prospecting). Result: you cut your acquisition sources by focusing only on capture.

**When to use it:** - Short purchase cycle (< 3 days) - Few touchpoints before purchase (< 2 on average) - Account with low conversion volume (< 30/month): Data-Driven can't function - Industries where purchase decision happens in one session (impulse, low price)

**Known biases:** - Overvalues branded keywords and remarketing ads - Undervalues prospecting and awareness campaigns - Distorts channel comparison (Search vs Display vs Social)

**Impact on decisions:** In Last Click, you'll likely cut your prospecting campaigns "because they don't convert directly," while overvaluing your branded campaigns. In reality, prospecting campaigns feed the pipeline that remarketing captures.

**Recommendation:** use Last Click only if your purchase cycle is short or you don't have the volume for Data-Driven.

**How it works:** Google's AI analyzes all conversion paths in your account to calculate the real contribution of each touchpoint. It compares paths that convert with those that don't, and assigns credit accordingly.

**Prerequisites:** - Minimum 300 conversions in the last 30 days (ideally 1000+) - Minimum 3,000 clicks in the last 30 days - Without these thresholds, Google automatically reverts to the Data-Based Available model (hybrid)

**When to use it:** - E-commerce accounts with > 300 conversions/month - Long purchase cycles (7+ days) with multiple touchpoints - Multi-channel: Search + Shopping + Display + YouTube

**What it concretely changes:** Prospecting campaigns and generic keywords recover a fairer credit. Branded and remarketing campaigns see their credit decrease. Initially, your apparent performance metrics will degrade (lower ROAS, higher CPA in reports): this is normal, it's the signal that you were going too far in last-click attribution.

**Adaptation period:** 4-6 weeks for Smart Bidding to adapt to the new attribution model.

| Situation | Recommended model | Reason | |---|---|---| | E-commerce, > 500 conversions/month, multi-channel | Data-Driven | Sufficient volume, multiple touchpoints | | E-commerce, < 100 conversions/month | Last Click | Not enough data for DD | | B2B lead gen, long cycle (30+ days) | Data-Driven or Linear | Multiple critical touchpoints | | B2C lead gen, impulse purchase | Last Click | Decision in 1 session | | Branded campaigns only | Last Click | Single logical touchpoint | | Multi-channel account with defined funnel | Data-Driven | Cross-channel attribution needed |

**Meta Ads case:** Meta Ads works with attribution windows (1-day click, 7-day click, 28-day click + 1-day view) rather than attribution models in the Google sense. The default recommendation is 7-day click for most e-commerce. Switch to 1-day click if your purchase cycle is very short (< 24h) to avoid overvaluing Meta's impact.

**Step 1: Verify conversion volume** In Google Ads: Tools > Measurement > Conversions > Check "Data available for Data-Driven" status

**Step 2: Anticipate the impact on reports** Inform your client or management: performance numbers will change. Prepare a before/after comparison.

**Step 3: Make the transition during a quiet period** Avoid high-activity periods (sales, Christmas). Choose a period with a stable baseline.

**Step 4: Let Smart Bidding adapt** For 4-6 weeks, avoid sudden budget or bid adjustments. Let the algorithm recalibrate.

**Step 5: Compare over 60 days** Compare the 60 days post-migration with the 60 days pre-migration for the same periods of the year. Look at real ROAS (revenue / spend) not the reported ROAS in the interface.