Marketing attribution is hard. Most attribution models have real flaws. None of them perfectly capture how a B2B deal actually formed. But imperfect attribution beats no attribution every time, and the teams that run a consistent model outperform those that operate on instinct.

Here is how to build a multi-touch attribution model that is good enough to make better budget decisions.

Step 1: Understand What You Are Actually Measuring

Attribution is an answer to one question: which marketing activities contributed to a deal closing?

The reason this is hard is that B2B deals involve multiple people, multiple touches over months or years, and a buying process that marketing cannot fully observe. A CMO who consumed four blog posts and attended a webinar before a $500K deal closed: how much credit does each of those touches deserve?

There is no objectively correct answer. Attribution models are decision-support tools, not scientific measurements. The goal is consistent, directional accuracy: can you tell which channels and programs are producing pipeline more efficiently than others? That is sufficient to make better budget decisions.

Step 2: Choose Your Model

For most B2B companies, one of three models is appropriate as a starting point.

U-shaped (First and Last Touch Weighted): 40% credit to the first touch that created the lead, 40% credit to the last touch before the opportunity was created, 20% spread across all middle touches. This model is the easiest to implement and is appropriate for companies with short to medium sales cycles and relatively simple buying journeys.

W-shaped (First, Lead Creation, Opportunity Creation): 30% credit to first touch, 30% to the touch that created the lead, 30% to the touch that created the opportunity, 10% spread across remaining touches. Better for longer sales cycles where the lead creation and opportunity creation moments are distinct and important.

Full-Path: Credits first touch, lead creation, opportunity creation, and close touch, with remaining credit distributed across all other touches. Most accurate but most complex to implement. Appropriate for mature RevOps organizations with clean CRM data and a history of reliable attribution tracking.

Start with U-shaped. It is implementable in any major CRM or MAP within weeks. Move to W-shaped or full-path when the data hygiene and CRM infrastructure support it.

Step 3: Clean the Data First

Attribution is only as good as the data it runs on. Before activating any model, audit your CRM for: leads with no source field populated, contacts with no marketing touches logged, opportunities with no associated contact, and campaigns with no response tracking.

A 30% attribution coverage gap (30% of closed deals have no marketing attribution) means your model is systematically understating marketing's contribution. Fix data hygiene before drawing conclusions from the model.

Run a data audit across the last 12 months of closed-won deals. For each deal: how many marketing touches are logged, what channels are represented, what is the time span of the engagement? This audit reveals both data quality issues and your actual attribution baseline.

Step 4: Configure the Model in Your CRM or MAP

Most major platforms (HubSpot, Salesforce + Pardot/Marketing Cloud, Marketo) have multi-touch attribution built in or available through an add-on. Configure the model you selected in Step 2. Map campaign types to touch categories. Confirm that all major channels (paid, organic, email, events, direct) are being captured.

Run the model against your last 6 months of closed-won data. Compare the results to your current first-touch or last-touch view. The differences reveal where your current model was over- or under-crediting specific channels.

Step 5: Use the Model for Decisions, Not Perfection

The most common misuse of attribution data is treating it as exact. It is not exact. Use it for relative decisions.

If paid LinkedIn consistently shows a 40% higher pipeline-per-dollar than paid search in your attribution data, that is a real signal to reallocate budget toward LinkedIn. If a specific content type (case studies, ROI calculators) shows high attribution across multiple models, that is a signal to invest more in that format.

Never make a budget decision based on a single attribution data point. Look for consistent patterns across three or more quarters. Attribution outliers are noise. Attribution trends are signal.

Step 6: Build the Budget Review Process Around It

Once the model is running, change your budget review process to use it. Quarterly budget reviews should include: channel ROI by attribution model, program performance by cost-per-pipeline-sourced, and year-over-year attribution trend by major channel.

This process trains the organization to think about marketing investment in terms of pipeline outcomes rather than activity volume. It also creates accountability for poor-performing channels: if a channel shows up consistently at the bottom of the attribution ROI ranking, it warrants a reduction or reallocation.

FAQ

Q: What is multi-touch attribution? A: Multi-touch attribution is a marketing measurement model that distributes credit for a closed deal across multiple marketing interactions throughout the buyer's journey, rather than assigning all credit to the first or last touch. It provides a more accurate picture of which channels and programs contributed to pipeline and revenue.

Q: Which attribution model is best for B2B? A: There is no single best model. U-shaped is the most practical starting point for most B2B companies. W-shaped adds value for organizations with distinct lead creation and opportunity creation moments. Full-path is most accurate but requires mature data infrastructure. Start with the simplest model your data quality can support and add complexity as data hygiene improves.

Q: How do I handle offline touches in attribution? A: Log offline touches (events, in-person meetings, phone calls) as manual activities in your CRM with a consistent naming and categorization convention. This enables them to be captured in the attribution model. Common gaps include trade shows (often logged as a list import rather than individual activities) and executive dinners (often not logged at all).

Q: What CRM or MAP supports multi-touch attribution natively? A: HubSpot has multi-touch attribution built into its paid tiers. Salesforce supports multi-touch attribution through Einstein Attribution. Marketo, Eloqua, and Pardot support it through built-in reporting or third-party integration. Dedicated attribution platforms (Bizible/Marketo Measure, Dreamdata, Rockerbox) provide more sophisticated models for organizations that need full-path or custom attribution.

Q: What is attribution coverage rate and why does it matter? A: Attribution coverage rate is the percentage of closed-won deals that have at least one marketing touch logged. If 40% of your deals have no marketing attribution, your pipeline sourced metric understates marketing's actual contribution by approximately 40%. Before trusting attribution data, run a coverage audit.

Q: How do I convince a skeptical CFO that attribution data is reliable? A: Show the methodology, acknowledge the limitations, and present trends rather than point-in-time data. A CFO who sees consistent patterns across 8 quarters is more convinced than one shown a single quarter's data. Also show what the data implies for investment decisions: "our attribution model suggests reallocating $200K from channel A to channel B would produce approximately $1.2M in additional pipeline." Decision relevance builds credibility faster than methodological perfection.

Jeff Pedowitz | President and CEO, The Pedowitz Group | Marketing Operations | HubSpot Services