Cold DM Case Study: A SaaS That Booked 40 Demos

The starting point

Northpeak, a fictional analytics SaaS, had seven employees and a product that helped e-commerce operators understand ad attribution. Paid acquisition was too expensive at their stage, and inbound was thin because they had little content history. They had 200 qualified LinkedIn connections of potential buyers and a list of 1,500 target operations leads scraped from public profiles of Shopify brands. The goal: book demos without hiring an SDR, because every dollar of margin mattered at their size and a sales hire would have eaten the entire experiment's budget before it proved anything.

The constraint shaped the strategy. With no budget for ads and no headcount for a sales team, the founder had only time and relevance. That forced discipline: every message had to count, which is exactly the condition under which cold DM outperforms paid channels. Scarcity of resources produced focus, and focus produced the reply rate that paid channels could not match for the same cost, because paid channels do not reward a tightly written sentence the way a human reader does.

The targeting

They narrowed to one segment: heads of e-commerce operations at Shopify brands doing $1M-$10M annually. That focus let them write one sharp observation about a problem those operators share: messy attribution across Meta and Google as spend scales. Broadening would have diluted the message and the reply rate, because a generic message to 'e-commerce' cannot name the specific pain that makes someone reply. The tighter the segment, the sharper the opener, and the sharper the opener, the higher the reply, which is the entire game.

The message and sequence

The opener named the attribution problem and offered a tiny, no-commitment deliverable: a two-line teardown of how a similar brand fixed it. By asking for a look rather than a demo, they lowered the reply barrier to almost nothing, which is exactly why the sequence performed. The follow-up added a concrete proof point, and the breakup gave permission to say no, which paradoxically earned more replies than the middle touch because it removed the pressure of a sales conversation the prospect was avoiding.

They used the follow-up schedule to keep spacing consistent and avoid the mistake of following up too fast, which is the error that makes prospects feel chased and opt out mentally before you have earned the right to ask. Consistency of cadence signalled a professional, not a desperado, and that perception protected the reply rate across the whole sequence rather than just the first message.

The funnel numbers

Forty demos from 900 sends is a 4.4% demo rate, which sits in a healthy range for well-targeted LinkedIn outreach. The 27% close rate reflected tight fit, not a discount, because the people who replied were already experiencing the exact problem the product solved. That fit is the lesson: targeting drove the close as much as the product did, and the DM merely surfaced a need that already existed rather than manufacturing one.

What drove the result

• Narrow ICP meant every message referenced a real shared problem.
• Low-friction ask (a 2-line teardown) beat a direct demo request in message one.
• Consistent three-touch follow-up captured replies the first message missed.
• They tracked every stage, so they knew the bottleneck was closing, not opening.

What they would do differently

With hindsight, they would have started the warm-up two weeks earlier and parallelized it across two sender accounts to avoid the mid-quarter volume ceiling. They also noted that the Instagram leg converted lower but cheaper, and they would have allocated more top-of-funnel there while keeping LinkedIn for the high-intent close. Small tweaks, but they illustrate that even a working campaign has a next iteration waiting, and the data from the first quarter told them exactly where to spend the second.

Lessons you can copy

You do not need a big team or budget to book demos with DMs. You need a narrow target, a relevant observation, a small ask, and the discipline to follow up. Model your own funnel with the revenue goal calculator before you start so you know how many sends your goal actually requires, and you will avoid the disappointment of quitting at 200 sends when the math said you needed 900. The calculator turns hope into a plan you can execute and defend to a boss or a co-founder.

Building the attribution proof

The 'similar brand' claim only worked because it was specific and true. Northpeak had one real customer story where they untangled Meta and Google attribution, and they reused that single proof across every opener rather than inventing different results for different prospects, which is how a one-customer company sounds like it has a track record. The lesson for early-stage SaaS is to mine your first three customers for one sharp, quantified story each, then let that story do the persuading while the DM stays short and the ask stays small, because the proof carries the weight the message deliberately leaves out.

Scaling the playbook to a team

The moment Northpeak hired its first SDR, the founder's instinctive relevance had to become a written system, or the reply rate would have collapsed under someone new. They documented the ICP, the observation sources, the opener, and the follow-up schedule in one SOP, then trained the SDR against it for a week before letting volume climb. The outreach SOP template is the exact artifact they wish they had started earlier, because writing the system down is what let the founder step away from every send without watching quality slip, and it is the difference between a founder-led hack and a repeatable growth channel the business can rely on.

• Write the ICP, signal sources, and opener into one SOP.
• Train the new sender against it for a week before scaling.
• Keep the single proof story until a second one is real.
• Audit the first 100 sends for adherence, not just outcomes.

What a second quarter looked like

With the first quarter proven, Northpeak’s second quarter doubled sends on LinkedIn and shifted Instagram to pure top-of-funnel, lifting demos to 80 and clients to 22 without adding headcount. The repeat was possible only because the system was documented in an SOP; the founder could step back while the sequence ran. The revenue goal calculator let them predict the second-quarter numbers before they happened, which is the whole point of treating outreach as a forecastable system rather than a hope.

Channel breakdown: LinkedIn vs Instagram

Northpeak ran LinkedIn as the high-intent close and Instagram as cheap top-of-funnel. LinkedIn converted higher per send but cost more attention; Instagram was lower friction but needed more volume. Knowing the split let them allocate founder time where it paid and use Instagram to feed the LinkedIn list with warmed names.

The demo-to-close motion

Booking the demo was only half the work. The founder ran every demo personally, which kept the close rate high because the conversation stayed about the prospect’s actual attribution problem rather than a rehearsed pitch. The DM earned the meeting; the founder’s genuine expertise closed it, and that division of labor is what made 27 percent of demos become customers.

Why the founder ran the demos

Northpeak could have routed demos to a junior, but the founder ran every one because the close depended on genuine expertise about attribution, not a script. A founder-led demo kept the fit conversation honest and the close rate at 27 percent instead of the lower rate a disconnected closer would have produced. The lesson for early-stage SaaS is that the DM earns the meeting and the founder’s real knowledge closes it; handing the close to someone who does not understand the problem wastes the reply you worked hard to get.

Replicating the playbook in a new vertical

• Keep the one-ICP discipline; pick a second vertical only after the first compounds.
• Swap the proof story to a result from the new vertical, not a borrowed one.
• Reuse the same three-touch sequence; only the signal changes.
• Re-model the funnel with the revenue goal calculator for the new list size.

What the second segment taught them

Northpeak’s second vertical was usage-based pricing tools, a cousin of attribution software. The same three-touch sequence worked, but the proof story had to be swapped to a result from that world, because a borrowed proof from e-commerce would have read as fake to a pricing-tool founder. The lesson is that the system is portable but the specifics are not: keep the skeleton, swap the signal and the proof, and re-model the funnel because the list size and reply rate differ by vertical. The revenue goal calculator handled the re-forecast in an afternoon, which is why they could expand without guessing.

Suggested image brief

Quick checklist

• One narrow ICP defined with a shared problem
• Opener references a real, specific signal
• First ask is small (a look, not a demo)
• Three-touch follow-up scheduled
• Every funnel stage tracked
• Daily volume within safe limits
• Founder or expert runs the demo
• Results modeled before launch

Frequently asked questions