How a B2B SaaS Company Reallocated $400K in Ad Spend After Switching Models

The company spent $1.8M on paid acquisition last year. Their marketing team was hitting ROAS targets. Pipeline coverage looked solid. The board was happy with CAC trending down quarter over quarter.

Then they ran a 90-day multi-touch attribution audit. What they found shifted $400K worth of budget allocation and changed how their entire growth team thought about channel mix.

This is what happened.

The Setup: What Their Data Looked Like Before

Before the audit, this company - a B2B operations software platform selling to mid-market logistics and supply chain teams - was running last-click attribution across all channels. Their Google Ads account showed strong ROAS. Their retargeting campaigns looked especially good, often showing 4-6x returns. LinkedIn was expensive and appeared to barely break even on a cost-per-conversion basis.

They'd already cut LinkedIn spend twice in 18 months based on these numbers. Each cut was followed by a gradual pipeline slowdown that the team attributed to "market seasonality" or "longer deal cycles." Nobody connected the dots.

Step 1: Rebuilding the Touchpoint Map

The audit started with a full touchpoint reconstruction across their top 200 closed-won deals from the prior 12 months. They pulled data from their CRM, ad platforms, website analytics, webinar platform, and sales activity logs.

The median deal had 11 tracked touchpoints across a 67-day sales cycle. The shortest was 4 touchpoints in 14 days. The longest was 31 touchpoints across 8 months for a 6-figure enterprise deal.

When they mapped which channels appeared at which stages of the funnel, LinkedIn showed up in the first two touchpoints for 71% of their closed-won deals. Retargeting showed up in the last two touchpoints for 83% of closed-won deals. These numbers looked very different when viewed through the lens of "which channel started the conversation" versus "which channel got the click before purchase."

Step 2: Choosing the Right Model for Their Cycle

They tested three model variants against their 200-deal dataset: linear attribution (equal credit to each touchpoint), time-decay (more credit to recent touchpoints), and a custom W-shaped model (40% credit split between first and last touch, 20% credit to the lead conversion event, 40% split across everything in between).

The W-shaped model produced numbers that most closely matched what their sales team already knew intuitively: LinkedIn was opening doors, demos were qualifying intent, and retargeting was closing people who were already close to a decision. The model matched the sales team's ground-level view of how deals actually closed.

That alignment matters. When your attribution data tells the same story your sales team tells, you can trust the numbers enough to act on them.

The Shift: What Changed in the Channel Mix

Total reallocation: $400K moved from bottom-funnel amplification to top and middle funnel investment.

What Happened to the Metrics

The first 30 days after reallocation looked rough. Branded search and retargeting were doing genuine bottom-funnel work, and reducing them caused some conversion metrics to dip. The marketing team got nervous. Finance asked questions.

The growth team had prepared for this. They'd forecasted a 45-60 day lag before the increased top-funnel investment would flow through to pipeline, and they'd communicated that to leadership before making the changes.

By day 75, pipeline coverage from organic and LinkedIn had increased meaningfully. Branded search conversions had only dropped about 12% despite a 40% budget reduction - which suggested a significant portion of that spend was capturing demand that would have converted anyway. Retargeting similarly showed diminishing returns: cutting 35% of spend only reduced conversions by 14%.

The Lesson: Attribution Isn't a Dashboard Setting

The most important thing this company did wasn't picking the right attribution model. It was building the case for change before making the change. They ran the 90-day audit first, showed leadership the delta between last-click and multi-touch numbers, explained why they believed the multi-touch view was more accurate, and got buy-in before touching any budgets.

Changing attribution models and then changing budgets is risky. You need to be able to answer "why do we trust the new numbers more than the old ones?" before you reallocate $400K based on them.

We spent 18 months cutting LinkedIn because the numbers told us to. The numbers were wrong. It took 90 days to find out and another six months to rebuild what we'd damaged. That's an expensive lesson in attribution hygiene.

The practical version of that lesson: run your multi-touch model in parallel with your existing model for at least 60 days before making any budget changes. If the two models agree, you haven't learned anything new. If they diverge significantly, that divergence is worth investigating carefully before acting on it.

Four hundred thousand dollars is a lot of budget to move. But it's nothing compared to what a year of misallocated spend costs in pipeline that never materialized.