So, you want to build a multi touch attribution model in Leanlayer. Maybe you’re tired of hearing “first touch” vs. “last touch” arguments, or you just want to know what’s actually working in your marketing. Either way, this is for you. You’ll get a straight-shooting walkthrough—no jargon, no hype, just how to set up a real model that tells you what’s moving the needle.
If you’re new to Leanlayer, it’s a tool for building data pipelines and analytics without having to wrangle code every five minutes. Here’s the product if you want to poke around before diving in.
Let’s get started.
Why Multi Touch Attribution? (And When to Skip It)
Before you open up Leanlayer, it’s worth asking: do you really need multi touch attribution? If your sales cycle is dead simple, or your traffic is all word-of-mouth, you might get more clarity from simple reporting. But if you’re running multiple channels (ads, email, content, etc.) and want to see how those touches actually influence conversions, multi touch is the way to go.
What’s worth your time: - Understanding which channels actually contribute—versus ones that just look busy. - Fixing “last touch bias,” where the final channel gets all the credit. - Getting a clearer picture of the real customer journey.
What to ignore: - Overly complex models that nobody on your team understands. - Fancy visualizations if you’re still struggling to trust your data.
Okay, ready? Let’s build.
Step 1: Get Your Data House in Order
Attribution is only as good as your data. Garbage in, garbage out. Here’s what you need, minimum:
- Touchpoint Data: Every time a user interacts with your brand (ad click, email open, site visit, etc.), you need to log it. Source, timestamp, user identifier—don’t skip the details.
- Conversion Events: What counts as a “win”? Demo booked, purchase, sign-up? Define it and make sure it’s tracked.
- User Identity: You need some way to tie all those touchpoints and conversions to a single user. This could be an email, a user ID, or a cookie. If you can’t stitch these together, attribution won’t work.
Pro tip: If your tracking is a mess, fix that first. Attribution just magnifies data problems.
Step 2: Pipe Your Data Into Leanlayer
Here’s the thing—Leanlayer won’t magically make disconnected data work. You’ll need to get your touchpoints and conversions into their system.
How to do it: 1. Connect Your Sources. Leanlayer supports connectors for popular platforms (Google Ads, Facebook, HubSpot, GA4, etc.). Go to “Data Sources,” connect the accounts you care about. 2. Custom Data? If you’re tracking touchpoints in your own backend (think: custom events, internal CRM), use Leanlayer’s API or CSV uploader. 3. Check Your Imports. Don’t just assume it worked. Look at the raw data tables—verify you see user IDs, timestamps, channels, and conversion events.
What to ignore: Don’t plug in every single data source “just because.” Start with the channels that actually drive traffic and conversions.
Step 3: Define What Counts as a Touch
Not every interaction should count the same. Leanlayer lets you define what a “touch” is. This matters a lot.
Common touchpoints: - Ad clicks - Email opens/clicks - Page visits (landing pages, product pages) - Social engagement - Direct visits
Decide: - Does a pageview count, or only meaningful actions (clicks, form fills)? - Do you want to include every touch, or only the first from each channel? - How long is a “conversion window”? (i.e., how far back do you credit touches—7 days, 30 days, 90 days?)
How to set this up in Leanlayer: 1. Go to “Attribution Models” and click “Create New.” 2. Choose your event types to include as touchpoints. 3. Set your conversion event(s). 4. Adjust your lookback window (this is where you’ll decide how far back in time Leanlayer considers touches).
Pro tip: Keep it simple for your first model. You can always add more complexity later.
Step 4: Pick Your Attribution Model
Leanlayer comes with several attribution models out of the box. Here’s the real talk on them:
- First Touch: Gives all credit to the first interaction. Good for brand awareness, but ignores everything else.
- Last Touch: Gives all credit to the final step before conversion. (Most tools default to this. It’s easy, but often misleading.)
- Linear: Spreads credit evenly across all touches. Simple, but can make every channel look equally important (which is rarely true).
- Time Decay: More recent touches get more credit. Makes sense if your decision cycle is short or people move fast.
- U-Shaped / Position-Based: First and last touches get the most credit, middle touches get less. Good for longer journeys.
How to choose: - If you’re just starting, run Linear and Last Touch in parallel and compare. You’ll see what’s different. - If your sales cycle is long and involves lots of research, U-Shaped or Time Decay usually make more sense. - Don’t get hung up on “industry best practices.” Pick what matches your actual buying process.
How to set in Leanlayer: 1. In your attribution model setup, pick the model type. 2. For custom weights (like U-Shaped), Leanlayer lets you set percentages for first, middle, and last touches. 3. Name your model something clear, like “Q2 Linear Attribution” or “2024 U-Shaped Test.”
Step 5: Map and Validate Your Data
Even with the right model, bad mapping can ruin everything. Before you trust any output, check:
- Are all your channels mapped correctly? (Google Ads, Facebook, Organic Search, etc.)
- Are conversion events firing as expected?
- Are there “unknown” or “unassigned” touchpoints clogging up your model? (If yes, fix your source data.)
In Leanlayer: - Use the “Preview” or “Test Run” feature on your model. - Look at several real conversion paths. Do they make sense? Would a human agree with how the model attributes credit? - If something looks weird (like 50% of conversions coming from “unknown”), pause and troubleshoot before rolling out.
Pro tip: Don’t trust attribution output blindly. Always gut-check the results before sharing with your team or leadership.
Step 6: Analyze, Share, and (Yes) Iterate
Once your model is live, dig into the results. But don’t just look for what makes you feel good—look for real insights.
What to do: - Compare models side by side. Example: Does Paid Search look better in Last Touch than in Linear? Why? - Look for “hidden” channels that get undervalued in last-touch models. - Download or export the data (Leanlayer lets you do this) to slice and dice in your own tools if you prefer. - Share findings, but be clear about limitations. Attribution is a model, not gospel.
What not to do: - Don’t panic if your favorite channel looks bad. It might just mean you need to tweak touch definitions or lookback windows. - Don’t overfit your model just to “prove” a channel works. Let the data speak.
Common Pitfalls (and How to Dodge Them)
- Ignoring Data Quality: If you don’t have reliable user IDs, nothing else matters.
- Too Many Touches: If you count every pageview and email open, you’ll drown in noise. Focus on meaningful actions.
- Changing Definitions Mid-Stream: Once you set what counts as a touch or conversion, stick with it for a while to get consistent trends.
- Overcomplicating: If nobody on your team can explain the model, it’s too complex. Simple wins.
Wrapping Up: Keep It Simple, Test, and Adjust
Attribution is messy. Anyone promising perfect answers is selling you something. Start simple, get your data right, and use Leanlayer’s tools to see what’s actually happening—not just what you hoped would happen.
The best models are the ones you can explain in 30 seconds. Build, test, tweak, and don’t be afraid to throw out what’s not working. That’s how you get real insights—and avoid chasing your tail with analytics busywork.
Good luck, and remember: attribution should help you make better decisions, not just prettier dashboards.