Paid media audit case study: How Understory Agency found 74% of Northbeam's LinkedIn budget in the Audience Network

Nobody audits a paid account until something is obviously wrong. Budget goes out on schedule, the platform reports conversions, and the 2 numbers get assumed to describe the same thing. Meanwhile the defaults do a great deal of the spending. Placements nobody chose. Bidding strategies nobody set on purpose. Targeting expansions left switched on since launch.

Northbeam sells marketing attribution to ecommerce brands. By Zach Sherrod's own account the platform measures $15 billion in ad spend a year for its customers. When Understory Agency audited Northbeam's own ad accounts in January 2026, 74% of the previous 30 days of LinkedIn budget had gone to the LinkedIn Audience Network. The LinkedIn account had no conversion tracking configured at all.

Zach Sherrod is Head of Growth & Partnerships at Northbeam. He joined a lean go-to-market team that had never had SDRs and that ran in silos, automating what it could and short of hours to scale any of it. Most of the paid budget sat on Google search, where last-click made the spend easy to defend internally. Paid social was the channel he wanted and the one he had never been able to make work, including with previous agencies.

Northbeam wanted paid social producing, outbound running at real volume, and one system underneath both. Understory Agency audited the accounts in January 2026 and the program went live in February.

LinkedIn has since produced 217 leads, 84 of them qualified, and 37 deals. Zach Sherrod reports $2.3M in LinkedIn-sourced pipeline from that channel, including a single deal closed at $112,000. Google spend was cut 44% deliberately and cost per opportunity fell 35%. Outbound has sent more than 85,000 emails at a 1.2% bounce rate.

Here is how we built it.

Performance Summary (from February 2026)

• LinkedIn-sourced pipeline: $2.3M, reported by Northbeam
• Largest closed deal from LinkedIn: $112,000
• Google cost per opportunity: down 35%, on 44% less spend
• Google pipeline per dollar: up 41%
• Meta: 549 leads and 27 opportunities on $46,000 of spend
• Outbound: more than 85,000 emails at a 1.2% bounce rate

About Northbeam

Industry: Marketing attribution for ecommerce

Northbeam is a marketing attribution platform for ecommerce brands, headquartered in San Francisco. Growth teams use it to see what their advertising is actually producing across channels, and by Zach Sherrod's account it measures $15 billion in ad spend a year. Understory Agency has run Northbeam's paid media since February 2026, and the engagement later widened into GTM engineering, so one team now runs the ads and the outbound off the same audience data.

The Challenge

Northbeam had 2 problems, and the second one was the expensive one. The first was capacity. Zach Sherrod describes a lean and scrappy go-to-market team with no SDRs, working in silos and automating what it could. Most of the budget went to Google search because last-click attribution made that spend simple to justify. Paid social was the obvious gap, and previous agencies had not been able to close it.

The second problem was measurement, which is where the irony sits. An attribution company was buying media inside accounts that were mis-measuring themselves. There was no conversion tracking configured in the LinkedIn account. The campaigns were optimizing toward website visits. Elsewhere the tracking counted an email click and a closed customer as the same conversion, then counted several of them twice.

Zach Sherrod was specific on camera about why he picked Understory Agency. He is a RevOps person by instinct, and the Clay-built audience work and the allbound approach read as his own language. He wanted an agency fluent in B2B SaaS, having found that most of the alternatives came out of traditional lead gen and did not speak it. He had also seen good work from agencies with other ecommerce SaaS companies, so the ICP fit was already proven to him.

The numbers before we began

Understory Agency audits every paid account before touching it, and writes the audit down. The January 2026 audit of Northbeam's accounts found:

👉 74% of the last 30 days of LinkedIn budget in the LinkedIn Audience Network, and $100,000 over 90 days

👉 No conversion tracking configured in the LinkedIn account, which was optimizing toward website visits

👉 LinkedIn running on max-delivery bidding, at times charging over $50 per click

👉 Over $50,000 on brand search in 60 days on a max-clicks strategy, with 80% of Google spend going to terms that did not fit

👉 Fewer than 20 negative keywords added in 3 months on non-brand, and both non-brand campaigns running exactly 1 ad group each

👉 A video campaign spending its full $500 a day outside the target audience, because optimized targeting had been left on

👉 Conversion tracking treating an email click and a closed customer as the same event, with double and triple counting on top

The Strategy: Rebuild the measurement, then point 4 channels at one list

The Audience Network was the first thing switched off. It is enabled by default on LinkedIn, which is why so much budget ends up there without a decision behind it. In our experience that traffic is full of misclicks, with bounce rates close to 99% and no time on page. Max-delivery bidding went next, after clicks that at times cost over $50. Then the conversion tracking got rebuilt, because until an account agrees with the CRM about what a lead is, every other change is guesswork. Reconciling the 2 systems row by row surfaced a tracking bug on Northbeam's side. The ad platform recorded 42 leads where the CRM held 4, because the conversion event was firing on a button click before the form was ever submitted. A platform optimizing on that count keeps buying the audience that clicks and leaves.

Google went the other way, and this is the part worth reading carefully. We cut the spend 44% on purpose. Over $50,000 had gone to brand search in 60 days on a max-clicks strategy. 80% of the spend was landing on terms that did not fit the business. Fewer than 20 negatives had been added to non-brand in 3 months. So the work was subtractive. A negative keyword system went in. Both non-brand campaigns got real structure, having run on 1 ad group each. Budget came off brand terms that were going to convert anyway. The account is smaller now and it produces more. Pipeline per dollar is up 41% and cost per opportunity is down 35%, with non-brand at 14% of Google pipeline and climbing. The win here is efficiency at a lower spend level, and nobody should read it as absolute growth.

LinkedIn is where the budget went. Roughly $300,000 of LinkedIn spend has produced 217 leads, 84 of them qualified, and 37 LinkedIn-sourced deals, 1 of which closed at $112,000. Zach Sherrod reports $2.3M in pipeline from the channel. Part of what made it work sits outside the ad account: Northbeam's own executives started posting more on LinkedIn, and those posts became the thought-leader ads. Northbeam sells a product that has to create pain before it can capture demand, so an ad that argues a case in a named person's voice does more work than one that asks for a demo.

LinkedIn working is what freed the budget for Meta, a channel B2B teams usually write off. $46,000 of Meta spend has produced 549 leads and 27 opportunities, at roughly $100 per lead. Lead quality held as volume climbed, with a July qualification rate of 88%. Video creative carried it at an 18% click-through rate.

The most portable finding of the engagement came out of building the target list. The strongest qualifier turned out to be behavioral. A brand's count of active Meta ads predicted its spend better than any revenue data we could buy, and it is observable from outside the company. More than 50 active Meta ads, or 20%+ month-on-month growth in that count, typically means more than $100K a month on the channel. That list then fed the outbound program, which has sent more than 85,000 emails to roughly 49,000 brands and produced 76 positive replies at a 1.2% bounce rate. By Zach Sherrod's own count, 40% of those positive replies became deals, and he attributes $378,000 in open pipeline to the program. The 4 channels now run as 1 system off the same audience data:

• Google: non-brand campaigns with real structure and a maintained negative keyword system, with brand search no longer absorbing the budget.
• LinkedIn: thought-leader ads built from posts Northbeam's own executives write, running on deliberate bids inside the LinkedIn feed.
• Meta: video-led prospecting, funded by LinkedIn working.
• Outbound: email sent against the same enriched Clay list the ads run on, targeted by the active-Meta-ad-count signal.

The Results (from February 2026)

👉 $2.3M in LinkedIn-sourced pipeline, reported by Northbeam, from roughly $300,000 of LinkedIn spend

👉 217 LinkedIn leads, 84 of them qualified, and 37 LinkedIn-sourced deals

👉 1 LinkedIn deal closed at $112,000, the largest from the channel

👉 Google cost per opportunity down 35% and pipeline per dollar up 41%, on 44% less spend

👉 Non-brand at 14% of Google pipeline and growing

👉 549 Meta leads and 27 opportunities on $46,000, at roughly $100 per lead

👉 88% Meta qualification rate in July, with video creative at an 18% click-through rate

👉 More than 85,000 outbound emails to roughly 49,000 brands at a 1.2% bounce rate, producing 76 positive replies

👉 40% of those positive replies became deals, with $378,000 in open pipeline attributed by Zach Sherrod

Key Takeaways

• The LinkedIn Audience Network is on by default, and default settings spend real money. Read the placement split before anything else in the account. A majority of budget sitting outside the feed is common, and it is almost never a decision anyone made.
• An ad platform and a CRM should agree on how many leads exist. When they disagree, the conversion event is usually firing on the wrong trigger, and the platform is optimizing toward a number nobody in sales can see. Reconcile the 2 counts across a full period before trusting either.
• Cutting spend is a result. Pipeline per dollar and cost per opportunity both move without the account getting bigger, and a smaller account that produces more per dollar is the better account. Efficiency deserves to be reported as the win it is.
• Behavioral signals qualify better than firmographic ones. How many ads a company is running right now says more about its budget this month than a revenue estimate does, and anyone can count them from outside.

What Northbeam says

Conclusion

Northbeam came into the engagement with a lean go-to-market team and no SDRs. Google carried most of the paid budget. The LinkedIn account had no conversion tracking behind it at all. Northbeam now runs 4 paid and outbound channels off 1 enriched list, with the measurement rebuilt underneath all of them. Google is 44% smaller and produces 41% more pipeline per dollar. LinkedIn has produced 37 deals and, by Zach Sherrod's reporting, $2.3M in pipeline.

The same pattern turns up in most B2B SaaS accounts that have been running a while without an audit. Budget drifts into placements nobody selected. Brand search absorbs spend that non-brand should be earning. The conversion goal quietly stops matching anything the sales team recognises. Understory Agency reads the account first, and the audit is what the scope conversation gets built on.

Related reading

• Understory Agency paid media, the service line behind this engagement.
• Best LinkedIn ads agencies for B2B SaaS, the buyer's guide to this category.
• Northbeam, marketing attribution for ecommerce brands.
• Forthcoming in the paid social cluster: Doppel, Hyperbound and Vicarius.