Auditing a $650K/month account: cutting 30% of waste

The question that kicked off this audit: "If we had to cut $200K from the annual budget without losing pipeline, where would you cut it?"

This came from a CFO reviewing the marketing budget. Not a request I expected. My first instinct was to defend everything.

Then I actually ran the numbers.

Thirty percent of our spend was going to campaigns that had never contributed a single closed-won opportunity in twelve months. Not a low contribution. Zero.

We cut it. Pipeline didn't move.

How large accounts accumulate bloat

A $650K/month account doesn't start that way. It grows over years. Different campaign managers make different decisions. Agencies add campaigns for new initiatives that never get removed when the initiative ends. Seasonal campaigns run indefinitely past their season.

Every year the account gets a little more complicated. Every year there are more campaigns with "historical context" that feels important to preserve. Nobody wants to touch what someone else built.

The result is an account carrying campaigns from 2019 that were built for a product line that's been discontinued, a geographic expansion that never happened, and a test that ran for six weeks and never got turned off.

This isn't anyone's fault. It's entropy.

The audit methodology

You can't cut blindly. Cutting a campaign because it has low click volume might eliminate your most valuable brand protection. You need pipeline data.

The process I used:

Step 1: Connect keyword IDs to Salesforce opportunities. Pull all clicks from the past 24 months at the keyword level. Pull all closed-won opportunities from Salesforce for the same period. Join on GCLID or UTM campaign data where you have it. For everything else, join on domain and time window (click within 180 days before close).

This gives you a list of keyword IDs that appear in the histories of actual closed deals.

Step 2: Identify zero-pipeline keywords. Any keyword that has spent more than $500 in the past 12 months and appears in zero closed-won opportunity histories. Flag it.

Step 3: Isolate zero-pipeline campaigns. Any campaign where every keyword in that campaign is flagged. That campaign has never demonstrably contributed to revenue.

Step 4: Test the assist hypothesis. Before cutting, check if any flagged campaigns appear in the early touch history of won deals (even if they're not the last touch). Some campaigns do "awareness" work that shows up in Salesforce as an early interaction before the eventual conversion.

In our audit, we found that 90% of zero-pipeline campaigns had zero early-touch presence either. They were genuinely not contributing.

What we actually cut

Thirty-one campaigns. About $180K in annual spend.

Included: four "awareness" campaigns built to drive top-of-funnel traffic that had been running on auto-pilot for three years. Six geographic expansion campaigns in markets where we'd never closed a deal. A display retargeting campaign that was generating view-through conversions but had zero closed-won contributions after removing view-throughs.

Not included: any exact match keyword campaign where any single keyword appeared in a closed-won history. Even if the campaign looked "inefficient" by CTR or CPL metrics, if it could demonstrate pipeline contribution, it stayed.

The result

Annual spend: $650K → $465K. Closed-won pipeline volume: flat. Cost per closed deal: improved by 22%.

The CFO got the cuts. The sales team saw no change. The pipeline data was the entire argument.

The lesson: at scale, you're almost certainly carrying 20-35% of spend on campaigns that contribute nothing. The money to fund new experiments is already inside the account. You just have to do the work to find it.