Marketing Mix Modeling
for the spend your dashboard can't explain.
Your CRM says most pipeline is "direct traffic." Last-click is flattering Google and punishing LinkedIn. Your CFO wants a defensible answer before the next budget cycle. That's the gap MMM closes — three questions at a time.
Is this channel actually working?
If we paused it tomorrow, would pipeline drop? Brand search, retargeting, and display usually fail this test. The channels your CMO worries about usually pass it.
Where does spend stop earning its keep?
Every channel has a point where more money stops buying more pipeline. We find that point per channel — so you know what to scale and what to hold.
Where should the next dollar go?
The output is a budget, not a report. A quarter-by-quarter spend plan your CFO will sign off on and your team can actually ship.
Four engagements, four different reallocations.
Clients anonymized; numbers are lifted from the actual Meridian / correlation models we delivered.
HR Tech
"Is the mid-market LinkedIn budget actually working?" The CMO had a hunch it wasn't. We were asked to either defend it or reallocate it — with math the CFO would sign off on.
- 3.5×LinkedIn vs GoogleLinkedIn worked 3.5× harder than Google per dollar across the full program — $1.3M in spend, 4,000+ leads. That was the easy part.
- 0MM dealsMid-market LinkedIn closed zero deals. £61K in explicit mid-market targeting produced 0 direct leads, and the 85 MM leads LinkedIn did pick up from halo never became revenue. The channel was planting seeds; other channels were harvesting.
- ~8%of MM demandPaid media only moved the needle 8% on mid-market. The other 92% was brand, content, and organic. So we pulled the MM budget out of paid and pushed it into the things that were actually working.
FinTech
"What do we actually get for our LinkedIn spend?" Last-click wasn't flattering. The team suspected LinkedIn was pulling more weight than the dashboard showed. We tracked 27 weeks of spend against every downstream channel to find out.
- ~45%of direct trafficNearly half of weekly direct-traffic swings moved with LinkedIn spend. People were seeing the ads and typing the URL later in the week. Classic brand recall — invisible to last-click.
- ~⅓of branded searchAbout a third of branded-search volume tracked with LinkedIn, too. People saw the ad, remembered the name, and Googled it a few days later. Google Ads was getting the credit for demand LinkedIn created.
- –8%flat-spend weekThanksgiving was the natural experiment. LinkedIn spend flat that week, site traffic dropped 8%. The pattern wasn't coincidence — when LinkedIn paused, traffic followed.
HealthTech
"LinkedIn vs Google — which one do we cut?" Leadership had been debating the split for a year. We built a model that tied every marketing dollar to real Salesforce pipeline — and ended the argument.
- $10.6Mfrom $293K$293K in paid media produced 204 Salesforce opportunities worth $10.6M in ARR. Cost per opportunity: $1,435. Pipeline-to-spend math the CFO could verify against the CRM.
- 3× vs 1.5×volume vs qualityLinkedIn brought 3× more leads. Google leads closed 1.5× more often. The channels were doing different jobs — LinkedIn filled the top, Google converted the bottom. Both earned their keep.
- +213%branded search liftBranded organic search more than tripled during high-LinkedIn weeks. A textbook halo effect — Google Ads was being credited for pipeline LinkedIn actually created.
Privacy Tech
LinkedIn was the biggest line item and the hardest to defend. Last-click attribution kept under-crediting it. 14 months of weekly data gave us enough signal to settle the question.
- 4×LinkedIn vs GoogleEvery LinkedIn dollar generated 4× the contacts a Google dollar did. The channel under pressure was the one doing most of the work. "Cut LinkedIn" would have been the wrong move.
- 3×direct-traffic liftIn high-LinkedIn weeks, direct traffic tripled (from ~3 to ~9 contacts per week). People were seeing the ads and coming to the site directly — showing up as "direct" in the CRM.
- 1.3×true impactFor every LinkedIn-attributed contact, there was roughly one more hidden in "direct." The dashboard was under-counting LinkedIn by about 30%. Adjust for that, and the economics looked very different.
"LinkedIn creates demand. Other channels capture it."
Same finding surfaced in all four engagements — different industries, different buyers, different budgets. If your attribution only rewards last-click, you're systematically underfunding the channel doing the actual work of building awareness.
Clients are anonymized. Numbers are pulled from the actual engagements. Every project ships with the full reallocation plan, the assumptions behind it, and a model that refreshes quarterly as new data comes in.
6–10 weeks from kickoff to a budget your CFO will sign.
No black box, no locked vendor platform. You'll see every assumption in the model, why each recommendation looks the way it does, and what would have to change for it to flip.
The four phases
Most MMM engagements take 6–10 weeks — from the first data pull to a reallocation plan your CFO can sign and your team can execute.
Data pull
We pull 12–24 months of weekly spend, leads, and pipeline from your ad platforms, HubSpot / Salesforce, and Search Console. Two weeks.
Build the model
We fit a channel-by-channel response curve — how long effects last, where each channel plateaus. Validated against holdout weeks or past experiments. Two–three weeks.
Reallocate
We translate the model into a budget. Quarter-by-quarter, per-channel, with the constraints you actually operate under (spend cap, CAC floor, pipeline target).
Prove the halo
We check whether big channels drive branded search and direct traffic you're not crediting them for — and queue a geo-holdout test to prove causation.
Want this run against your data?
If you're spending $1M+ a year on paid media and your attribution stack is starting to lie to you, MMM usually pays for itself in the first reallocation.
Talk to 42 Agency