Read Time: 5 Minutes
You spent money on marketing.
Your CEO asks what it did.
You open the attribution report... and it gives you a clean answer that is wrong.
Short answer: No. Use your attribution model as one input, not the thing that sets the budget. Early on, call every new customer and ask how they found you. Check every lead with sales each week. Then look at brand signs like search and surveys next to the model before you move money.
This one comes from my podcast with Jessica Gilmartin. She started as an investment banker, then started a chain of yogurt stores. She was CMO and CRO at startups, helped grow Asana from about $150M to $600M in revenue, and was CMO and then CRO at Calendly. She now runs her own marketing and sales consulting practice for first-time CMOs, CROs, and CEOs.
No. Jessica calls this her most important rule. Every model misses things, and some miss a lot. A last touch model can't see brand, content, or the things a buyer saw months ago. So use it as one input. Then tell your CEO and CFO up front that it is not how you set the budget.
Here's how she put it. When she sat with a CEO and a CFO, she made sure "we are not basing our budget decisions just on this attribution model. That's probably the most important thing."
And she said it out loud. She was "very upfront that that is not the right way to make budget decisions."
Attribution tools still help. Just don't let them set your budget.
Ask your customers. If you have a small number of deals, call every new customer and ask how they found you and what they saw before they talked to sales. Put the answers in one sheet. After 10 calls, you'll see the same few things come up. That is your first model. No data scientist needed.
Jessica's first piece of advice: "don't focus so much on making attribution perfect."
At a company called Honor, deals were big and slow. Buyers saw a lot of marketing before they signed. So she called them. "I would actually get on the phone with every new customer and I would just ask them, What did they do?"
The same few things came up again and again. "They would never have shown up in an attribution model, but I knew that they were critical."
Here's the call. 10 minutes.
Her take on doing it by hand early on: "don't rely on scalable things, just brute force things, and that's totally fine."
Because nobody can see what happens after the handoff. Marketing pays to make leads, sends them to sales, and gets no feedback. Was the lead good? Did a rep call? How fast? Without that, marketing can't fix its targeting and sales can't prove the leads were bad. Both sides end up guessing and blaming.
Jessica sees this all the time. "marketing spends a ton of money and time generating leads and they go into a black hole."
Her clients fix it with a simple weekly check. They look up every lead and what happened to it. Then they go back to sales with 1 question:
"Hey, we gave you what we consider to be 50 really good leads out of a hundred. Only 30 of them were acted upon. What happened to the other 20?"
That question leads to real changes. Faster follow up rules. Better targeting from marketing.
The weekly lead check. 5 minutes.
A short list both teams trust. Start with one shared source of truth in your CRM, not a pile of sheets. Agree on what each stage means, down to the exact rules. Then pick 10 to 15 numbers everyone looks at each week. Too much data hides the signal, so keep the list short.
Jessica says clean data is rare. "It's pretty unusual to find that the data is in a good place."
And dashboards come second. "you can't build good dashboards unless you have a shared source of truth between product, sales, and marketing."
Stage names are where most teams get stuck. She spent about 2 months with one client just on what each stage means. Even then, "even when you define them, you find out that you still haven't actually agreed on them".
The fix is to write it down. This stage means these exact 5 things. Then "it's getting it into Salesforce. And then you can do the reporting."
Look at signs that show demand is growing. Jessica tracks views, shares, organic search, AI search, and brand surveys every 3 months next to her model. None of them is perfect. Together they show if your marketing is working while the model only sees the last click before a deal.
She starts most teams on a first touch or last touch model. Her view: "everyone knows that that is a bad model, but that's probably the best way to start it."
The problem with last touch is what it can't see. "you are missing all of the top of the funnel, all the brand stuff".
So she adds other signs. Is organic search going up? Are AI search mentions going up? "We do brand surveys every three months."
As you grow and spend more, you can move to multi-touch models, and later to media mix models. Even then, keep the rule. The model informs. People decide.
Long enough to see if a focused plan worked. Pick a few things, do them well, and tie them to one goal for a set time, like moving into a new market. Then judge the goal. If you get more deals from that market, it worked. You don't need to tie each deal to one ad.
Jessica sees teams try too many things. "they kind of chase the shiny objects."
Her advice: "you have to do a few things and do them really, really well."
Say the whole company is moving into healthcare. Content, ads, webinars, and launches all point there. If healthcare deals go up, it worked. Her words: "Now, can you absolutely attribute it to one thing or another? Probably not. But does it matter? Also, probably not."
One more thing. If you hire an agency, don't wait for them to hand you a plan. "if you're expecting agencies to provide the strategy, then it's always going to fail." You set the plan. "you need to provide the strategy and they execute on that channel."
More on this: UTM Tracking Into Salesforce: A MarketingOps System You Can Trust, How do I build a Salesforce UTM taxonomy AEs will use?, and The 5-Minute Lead Rule
Your attribution model will always miss something.
So call your new customers. Check your leads with sales every week. Watch brand signs next to the model. And never let the model set the budget alone.
It's a fine place to start, as long as you know what it misses. It can't see brand, content, or early touches. Pair it with customer calls and other signs like organic search before you make budget calls.
Start with 10. Ask the same 5 questions each time and put the answers in one sheet. By 10 calls, the same few sources tend to show up again and again.
Someone in marketing or ops who can see every lead in the CRM. They mark the good ones, check which ones sales worked, and bring the gaps to the sales leader each week.
When you have enough deals and spend that calls can't keep up. Even then, the weights in a multi-touch model are a guess. Use it to learn what helps. Don't use it alone to set the budget.
Views, shares, organic search, AI search mentions, and a brand survey every 3 months. None of them is perfect. Together they show if demand is growing.
Want more of this? The Simply Scale newsletter drops every week: Subscribe here
Need help with Salesforce? Book a call: https://calendly.com/jordan-gosimplyscale/30min?utm_source=blog&utm_campaign=marketing-budget-attribution-model
Let's talk about how we can build a Salesforce system that grows with your business.
