Paid media case study: How Understory Agency rebuilt Knit's neglected ad account and reached $180 per demo request

Paid media is usually the first thing an early-stage marketing team stops maintaining. The account stays live and the budget keeps clearing. Nothing looks broken from the outside. What happens underneath is that the platform's defaults take over. Spend drifts toward whatever is cheapest for the algorithm to buy. Lead volume stops tracking with the money going out, and the report still says the campaigns are running.

Knit is an AI-native consumer insights and market research company in Austin. Logan LeBouef was their Head of Marketing at the time, running a marketing team of 2 inside a company of around 40 people. Paid media was one of the things he owned personally, and he got to it when he found the time, checking in about every 2 weeks.

Knit sells into an enterprise ICP with a $90,000 average selling price and a sales cycle of 3 to 9 months. Paid media had to produce demo requests from that kind of buyer on a schedule Knit controlled, at a cost that made sense against that deal size.

In the half-year before Understory Agency arrived, paid media at Knit produced 7 leads and 1 closed deal. Understory Agency audited the account in May 2025 and took it over in July. The rebuild moved Knit's impression share on their own brand name from 18% to 75%. Demo requests on LinkedIn came in at roughly $180, from logos including JP Morgan and Yahoo. The competitor campaign produced 1 opportunity worth roughly $45,000 with a verbal commitment.

Here is how we built it.

Performance Summary (May 2025 audit through the rebuild)

• Brand impression share: 18% to 75%
• Cost per demo request on LinkedIn: roughly $180, from logos including JP Morgan and Yahoo
• Largest single opportunity: roughly $45,000 with a verbal commitment, from the competitor conquest campaign
• Paid media output in the half-year before Understory Agency: 7 leads and 1 closed deal
• Campaigns live at audit: 1, with the last meaningful LinkedIn activity in January 2024
• Cost per branded click at audit: $8 to $10, against a $2 to $3 target

About Knit

Industry: AI-native consumer insights and market research

Knit describes itself as an AI-native research agency that combines the rigor and expertise of a traditional agency with the speed and efficiency of AI. The company is headquartered in Austin, Texas, and runs both quantitative and qualitative research for enterprise insights teams. Understory Agency audited Knit's paid account in May 2025 and took over paid search and paid social in July 2025.

The Challenge

Understory Agency sees this shape of account regularly at Series A and Series B, and nobody involved has made a bad decision. Knit's marketing team was 2 people inside a company of around 40, and paid media was one of many things Logan LeBouef owned personally. He checked the account roughly every 2 weeks, which is enough to keep it running and short of what it takes to optimize it. The result showed up in the pipeline. Paid media produced 7 leads and 1 closed deal across an entire half-year, and Logan went looking for a partner with better recommendations than the agency Knit had used before.

He had been watching Understory Agency for months before any conversation started. Adam Robinson at RB2B had mentioned them, which put Understory Agency's own LinkedIn content in front of him. What held his attention was that the content gave away how the work is actually done. Then Alex Fine sent him a cold email. Understory Agency's own outbound program is what turned Knit into a client, which is the mirror of the Operatus engagement, where the content program sold the content service. The call that followed opened on the account itself. Knit granted access to the ad platform. Understory Agency's paid lead walked through what the audit had found, and Logan stopped the walkthrough about 5 minutes in.

The numbers before we began

Understory Agency audits every paid account before scoping an engagement, and writes the audit down. The May 2025 audit of Knit's account found:

👉 7 leads and 1 closed deal from paid media across an entire half-year

👉 18% impression share on Knit's own brand name, against an 80% to 90% benchmark

👉 $8 to $10 per branded click, where $2 to $3 is the target

👉 Around 80% of paid non-brand clicks completely irrelevant to what Knit sells

👉 1 campaign live in the whole account, with the last meaningful LinkedIn activity in January 2024 and historical costs per click there that often exceeded $50

👉 Account-default conversion goals, counting phone calls, page views and dead content downloads alongside real demo submissions

👉 The wrong CRM connected to the ad platform, and every keyword on broad match including brand

👉 Most historical conversions arriving through the Search Partner Network, which the audit classed as junk or spam leads

👉 LinkedIn retargeting audiences with no job-title filters at all

The Strategy: Take the brand back first, then rebuild the funnel around it

Understory Agency started with the cheapest volume in the account, which was Knit's own name. An 18% impression share on your own brand means most of the people who already know you and search for you are being shown somebody else. Knit's brand term was also running on broad match, so it was matching generic category searches and paying category prices. That is the mechanism behind $8 to $10 branded clicks against a $2 to $3 target. The rebuild moved every keyword onto phrase and exact match and set target impression share bidding on the brand campaign. 3 negative keyword lists went in underneath to keep category traffic out. Search Partners came off at the same time, because most of the account's historical conversions had arrived through that network and the audit had read those as junk or spam. Removing it lowered reported conversion volume, which is the right trade when the conversions were never real.

The account was also counting the wrong things. Google Ads was optimizing against account defaults, so a phone call, a page view and a dead content download carried the same weight as a real demo submission. The CRM wired into the ad platform was the wrong one entirely, so nothing downstream of the form was reaching the algorithm. Understory Agency cut the account to a single primary conversion action and connected Knit's actual CRM. Lead-stage data then went back into the platform, which is what makes value-based bidding possible. A platform optimizes toward whichever counted event is easiest to produce, and that is almost never the demo request.

With the brand defended, the search account got a funnel. Dynamic search ads went in at the discovery end, where a buyer is still describing the problem in their own words and has no vendor in mind. Competitor campaigns went in at the other end, where a buyer is comparing named vendors and is closest to a decision. Logan describes that head-to-head layer as the one that performed incredibly well, and it is where the engagement's largest single opportunity came from, at roughly $45,000 with a verbal commitment.

LinkedIn needed more than a tune-up. Knit's last meaningful activity there was January 2024. Historical costs per click had often exceeded $50. The retargeting audiences carried no job-title filters at all, so the account was paying to re-reach anyone who had ever loaded a page. Understory Agency rebuilt the channel:

• Audience Network removed, so budget stayed on LinkedIn's own feed instead of the third-party inventory the platform enables by default.
• Manual bidding in place of the automated bid, which is how a small budget stays inside a job-title-filtered audience.
• Retargeting audiences rebuilt across page groups, video viewers, post engagers and lead-form openers.
• Thought-leader ads for Knit's executives, plus conversational ads and document ads.
• An account-based top of funnel built around an ungated AI research report.

The social account Understory Agency inherited had run on one-off pushes. A new release got teased. A white paper got shared. Each campaign ended with its own announcement, so nothing compounded and nobody was ever re-reached. The rebuild ran a full funnel underneath the report, with the top of it gathering interest and the retargeting and conversion layers catching that interest further down. Demo requests came in at roughly $180, from logos including JP Morgan and Yahoo. Against a $90,000 average selling price and a sales cycle of 3 to 9 months, that is the number that decides whether the channel is worth funding.

The Results (May 2025 audit through the rebuild)

👉 Brand impression share on Knit's own name: 18% to 75%

👉 Demo requests on LinkedIn at roughly $180, from logos including JP Morgan and Yahoo

👉 1 opportunity worth roughly $45,000 with a verbal commitment, out of the competitor conquest campaign

👉 From 1 live campaign to a full-funnel program across Google Ads and LinkedIn, with Search Partners off and 3 negative keyword lists live

👉 A single primary conversion action in place of account defaults, with Knit's real CRM connected and lead-stage data imported for value-based bidding

👉 LinkedIn back in market after a gap since January 2024, with Audience Network removed and retargeting rebuilt across page groups, video viewers, post engagers and lead-form openers

Key Takeaways

• Bid on your own name properly before anything else. Impression share on your own brand is the cheapest volume in any account, and 80% to 90% is the working benchmark. A brand term left on broad match will match generic category searches and pay category prices for them.
• Count 1 conversion action. Platform defaults will happily count phone calls, page views and content downloads next to a real demo request, and the algorithm optimizes toward whichever of those is easiest to produce.
• Check what the platform switched on for you. Google's Search Partner Network and LinkedIn's Audience Network are both on by default, and either can account for a large share of what an account reports without producing anything a sales team can work.
• An account nobody maintains still spends every day. The defaults keep buying against settings chosen for a different stage of the business, so the cost of leaving it alone compounds for every month it stays live.

Conclusion

Knit had the budget and the intent. A marketing team of 2 inside a 40-person company does not have the hours to rebuild an ad account on top of everything else, and platform defaults fill that gap on their own terms. Understory Agency audited the account in May 2025 and took it over in July. Impression share on Knit's own brand name went from 18% to 75%. Demo requests on LinkedIn came in at roughly $180 from logos including JP Morgan and Yahoo, and the competitor campaign produced 1 opportunity worth roughly $45,000 with a verbal commitment.

The same account shape turns up across Series A and Series B B2B. 1 campaign is live because 1 campaign was easy to keep alive. The brand term is on broad match. The conversion goals are whatever the platform suggested on setup day. Understory Agency runs the program as the team, so the rebuild happens without anyone in-house having to find the hours for it.

Related reading

• Understory Agency paid media, the service line behind this engagement.
• Best B2B SaaS paid media agencies, the buyer's guide to this category.
• Knit, AI-native consumer insights and market research.
• Forthcoming in the paid search cluster: Doppel, RemoFirst and Hiver. Also Operatus, which began the way this engagement did, with Understory Agency's own marketing producing the client.