The average customer journey involves 6 to 8 touchpoints before conversion. If you’re only crediting the last click, you’re ignoring 85% of the marketing that drove the sale. Multi-touch attribution solves this by mapping the full path.
Why Single-Touch Attribution Fails
Single-touch models are simple. First-click gives all credit to the first interaction. Last-click gives all credit to the final one. They’re easy to implement and easy to understand.
They’re also wrong.
First-click overvalues awareness channels. It credits the CTV ad but ignores the retargeting, email, and search clicks that actually moved the prospect to convert.
Last-click overvalues conversion channels. It credits the branded Google search but ignores the 5 impressions that built the awareness to search in the first place.
Both models lead to bad budget decisions. You’ll overfund whatever gets credit and underfund everything else.
Multi-Touch Attribution Models Explained
There are four main approaches to distributing credit across touchpoints. Each makes different assumptions about which interactions matter most.
Linear Attribution
Every touchpoint gets equal credit. If a customer had 5 interactions before converting, each gets 20%.
Pros: Simple, fair, easy to explain. No touchpoint gets overlooked.
Cons: Treats a random display impression the same as a high-intent search click. That’s rarely accurate.
Best for: Early-stage measurement when you don’t have enough data for more sophisticated models.
Time-Decay Attribution
Touchpoints closer to the conversion get more credit. The first interaction might get 5% credit. The last might get 40%.
Pros: Reflects the reality that recent interactions typically have more influence. Works well for long sales cycles.
Cons: Undervalues awareness channels that start the journey. Can make top-of-funnel spend look inefficient even when it’s essential.
Best for: Businesses with long consideration periods. Legal intake is a good example: someone might see a CTV ad three weeks before they actually call.
Position-Based (U-Shaped)
The first and last touchpoints each get 40% credit. The remaining 20% is split evenly among middle interactions.
Pros: Acknowledges that discovery and conversion are the two most important moments. Doesn’t ignore the middle.
Cons: Arbitrary split. Why 40/40/20 and not 30/30/40? The percentages don’t adapt to your actual data.
Best for: Most businesses starting with multi-touch. It’s a good default that balances awareness and conversion.
Algorithmic (Data-Driven)
Machine learning analyzes your actual conversion data and assigns credit based on statistical impact. Google Analytics 4 calls this “data-driven attribution.” It looks at paths that converted vs. paths that didn’t and calculates each touchpoint’s incremental contribution.
Pros: The most accurate model. Adapts to your specific business and customer behavior. No arbitrary rules.
Cons: Requires significant data volume (typically 600+ conversions per month). Black-box: harder to explain to stakeholders. Can be inconsistent with small datasets.
Best for: Mature advertisers with enough conversion data. If you’re spending $50,000+/month across multiple channels, this is where you should be.
Implementing Multi-Touch Attribution
Getting MTA right requires infrastructure. Here’s what you need.
Tracking Foundation
Website tracking. Every page, every form, every click needs to be captured. UTM parameters on all campaign URLs. Google Tag Manager or equivalent for event tracking.
Call tracking. For any business where phone calls matter (legal, healthcare, home services), you need dynamic number insertion (DNI) on your website. Each visitor session gets a unique phone number that ties back to their source. Without this, phone leads are a black hole in your attribution.
CTV/OTT tracking. Connected TV doesn’t have clicks. Attribution relies on household matching, IP-based attribution, or incremental lift studies. This is the piece most marketing attribution setups miss.
CRM integration. Attribution data needs to flow into your CRM so you can track from first touch all the way to closed revenue, not just to the lead.
Data Connection
The hardest part of MTA is connecting touchpoints across devices and channels. A prospect might:
- See your CTV ad on their Samsung TV
- Google your name on their phone
- Visit your website on their laptop
- Call from their phone
That’s four devices, one person. Cross-device identity resolution is what ties them together. Options include:
Deterministic matching. Uses login data to connect devices. Most accurate but limited reach. Only works when users are logged in (Google, Facebook, Amazon ecosystems).
Probabilistic matching. Uses signals like IP address, location, device type, and browsing patterns to infer connections. Broader reach but less precise.
Household-level matching. Groups all devices on the same IP/network into one household. Popular for CTV attribution since the whole household sees the ad.
Attribution Windows
How far back should you look? The attribution window matters.
- 7-day window: Good for impulse purchases and direct response
- 14 to 30-day window: Standard for most B2C
- 60 to 90-day window: Necessary for legal, where someone might see an ad months before they need a lawyer
- View-through windows: Typically 1 to 7 days for display, 7 to 14 days for CTV
Multi-Touch Attribution for Legal
The legal intake journey is one of the best use cases for multi-touch attribution. Here’s why.
Long consideration cycles. People don’t see one ad and call a lawyer. They research, compare, wait, and then act when they’re ready. A 60 to 90-day attribution window is standard for PI firms.
High case values make every touchpoint matter. If a signed case is worth $10,000 to $50,000+, you need to know what started that journey. Was it the CTV ad? The Google search? The retargeting display ad?
Multiple channels, one conversion. Legal advertisers typically run CTV, Google Ads, display retargeting, and SEO simultaneously. Without multi-touch, you’ll credit Google Ads for everything (because it’s usually the last click) and conclude that CTV “doesn’t work.” That’s a measurement failure, not a channel failure.
The typical legal conversion path:
- CTV ad builds awareness (no click, household-level impression)
- Display retargeting reminds the prospect (maybe a click to the site)
- Branded Google search (the prospect remembers the firm name)
- Website visit and phone call via call tracking
- Intake, sign-up, case
Last-click gives Google 100%. Multi-touch shows that CTV started the whole thing.
For legal-specific attribution setup, see our detailed guide on marketing attribution for law firms.
MTA vs. MMM vs. Incrementality
Three measurement approaches. They’re complementary, not competing.
Multi-touch attribution (MTA) is user-level. It tracks individual journeys. It’s granular and real-time but struggles with impressions that don’t get clicks (like CTV) and with privacy restrictions that limit tracking.
Marketing mix modeling (MMM) is aggregate-level. It uses statistical regression on historical spend and outcomes to measure channel-level impact. It accounts for external factors like seasonality, economic conditions, and competitor activity. But it’s slow (needs months of data) and can’t optimize in real time.
Incrementality testing uses experiments. Hold out a group from seeing ads, then compare conversion rates. It’s the gold standard for causal measurement. But it requires enough volume to run statistically significant tests, and it’s expensive (you’re not showing ads to the holdout group).
The best measurement programs use all three. MTA for day-to-day optimization. MMM for strategic budget allocation. Incrementality for validating the other two.
Getting Started
If you’re currently on last-click attribution, here’s how to move to multi-touch:
Step 1: Audit your tracking. Make sure every channel has proper UTM parameters, pixels, and event tracking. Fix gaps before you build models on bad data.
Step 2: Implement call tracking. If phone calls matter (they always do in legal), get call tracking software with dynamic number insertion running before anything else.
Step 3: Start with position-based. The 40/20/40 model is a good first step. It’s simple, easy to explain, and better than single-touch.
Step 4: Build toward algorithmic. Once you have 3 to 6 months of clean multi-touch data and enough conversion volume, test data-driven attribution. Compare its credit distribution to your position-based model and see what changes.
Step 5: Add CTV attribution. Most MTA setups ignore CTV because there’s no click to track. Household-level matching or incremental lift studies fill this gap. This is especially critical for legal advertisers spending on streaming.