April 5, 2026
.png)
A simple win attribution exercise revealed that our most expensive channel was also our worst performer. We shut it down and reinvested the budget where it actually works.
I've been in the startup game for a long time, and I'm embarrassed to admit how long it took us to do this. Over the summer, we sat down and did something that should have been table stakes from day one: win attribution.
Win attribution is straightforward. You take every closed deal and trace it back to the channel it came from. Where did this customer first find us? Was it an ad? A cold email? A conference? A referral? You do your best to answer that for every single close, and then you look at the pattern.
Once you know how many wins came from each channel, the math gets simple. You know what you're spending on each channel, so you can calculate cost of acquisition. You can go further and look at lifetime value by channel, and further still to calculate profitability by channel.
We hadn't done this exercise at Fraction until last summer. And looking back, the reason is the same reason most founders skip it: we were making assumptions. We had a general sense of what was working and what wasn't, and that general sense was wrong.
To understand why this mattered, you need to know that we were running several channels simultaneously: paid ads (Meta and LinkedIn), cold outbound email, conferences and in-person networking, referrals, and social media content.
Each of these required real time and real money. And without hard numbers, we were distributing budget based on gut feel. That's a dangerous way to run acquisition when you're a startup watching every dollar.
Marketing attribution as a discipline exists precisely because of this problem. Multi-channel strategies make it nearly impossible to understand what's driving results unless you sit down and trace every conversion back to its source. The more channels you run, the more critical the exercise becomes.
When we finally ran the numbers, paid ads turned out to be our most expensive channel by a wide margin. LinkedIn ads, despite us being a B2B company selling to exactly the kind of audience LinkedIn is designed to reach, had a cost per acquisition that was off the charts. Meta was somewhat cheaper, but still expensive relative to other channels.
This tracks with broader industry data. LinkedIn's average cost per click ranges from $5.58 to $10, making it one of the most expensive ad platforms available. For B2B SaaS companies targeting enterprise buyers, cost per lead can run $150 to $300 or more. And those are just leads, not closed deals.
Worse, we didn't see higher quality from paid ads. The lifetime value of clients acquired through ads was actually lower than clients from other channels. We were paying more to acquire customers who were worth less. That's a losing formula no matter how you slice it.
Cold outbound email came in at $7,500 per acquisition when it was working. For our business, that was a reasonable number. The problem is that cold email stopped working entirely.
Our cold outbound story mirrors what a lot of B2B companies have experienced. We used an agency for about a year and a half starting in late 2022, and it worked extraordinarily well. It genuinely helped build our business in the early days.
Then the wheels fell off in late 2023. The major email providers, led by Google, tightened their spam filtering significantly starting in February 2024. Google now requires bulk senders to implement strong email authentication, offer one-click unsubscribe, and maintain spam complaint rates below 0.3%. These aren't suggestions. Non-compliant email traffic is actively rejected with permanent failure codes.
On top of the filtering changes, we're now in the AI era on both sides of the equation. AI makes it trivially easy to generate cold email at scale, which means inboxes are flooded with AI-generated outreach. And AI-powered filters on the receiving end have gotten laughably good at categorizing those messages into the spam folder. Response rates have crashed essentially to zero.
I don't see a way of reviving cold email as a primary acquisition channel, and I'm not alone in that view. The combination of tighter platform enforcement and the flood of low-quality AI-generated outreach has fundamentally changed the economics.
Two channels stood out as effective: conferences and in-person networking, and social media content.
Networking and referrals proved to be high-quality, low-cost channels. The challenge with networking is scalability. Each person can only attend so many events and have so many conversations. But the deals that come through in-person relationships tend to be higher quality and higher lifetime value.
I actually don't think of referrals as a starting channel. They're a multiplier effect. The higher the quality of your service, the more referrals you get. It's payback for delivering a great customer experience. You still have to action it, run a good referral program, and take the steps to capture referrals, but the engine is powered by the quality of what you deliver. Industry benchmarks bear this out: referral programs consistently deliver the lowest customer acquisition costs across B2B channels.
Conferences turned out to be both effective and scalable, which is the combination you want. Meeting people in person, demonstrating expertise, and building relationships in a compressed timeframe produces real pipeline. Research suggests that 78% of B2B event organizers consider in-person events their most impactful marketing channel, and trade show leads that receive prompt follow-up convert at significantly higher rates than other sources. We're investing more here.
Social media content was the other standout. You can create social content for relatively low cost, and it scales in a way that in-person networking can't. When we factored in production costs, including the cost of our team's time (because you should always include that), it was still a low-cost, scalable channel. That's a rare combination.
The key with the cost calculation is honesty. You need to include not just the direct spend, like agency fees or ad budgets, but also the cost of your team's time. Every hour someone spends on a channel is an hour they could spend on something else in your business. When you account for that fully, the numbers shift in ways that might surprise you.
Because of this analysis, we shut paid ad spending down entirely. We went from a $30,000 per month Meta ads budget to zero. We reinvested that money into the channels that were actually working: conferences, content, and networking.
We're already seeing the reinvestment make a difference.
If you haven't done win attribution for your business, do it. It might take a couple of hours. AI can help you cut that time down if you have the data in a usable format. But check the figures. Get it done. The results might surprise you.
The exercise is simple: write down every closed deal, trace it back to its source, calculate cost of acquisition by channel, look at lifetime value by channel, and then calculate profitability by channel. When you see the numbers laid out in a spreadsheet, the right decisions become transparently obvious.
In our case, the right decision was cutting our most expensive channel and putting that money where it was actually earning a return. Don't wait as long as we did to find out.
Describe your software or AI project. Get a full scope with story-point pricing, sprint estimates, and a downloadable plan in minutes. No calls, no waiting.
Free and instant. Try the calculator now.
How long does a win attribution exercise take?
For most startups with a manageable number of closed deals, the exercise takes a few hours. The bulk of the time goes into tracing each deal back to its original source, which often means checking CRM records, asking salespeople, and sometimes making your best educated guess. The analysis itself, calculating acquisition cost and lifetime value by channel, is straightforward once the data is organized.
What if I can't trace every deal back to a single source?
You won't always have perfect data, and that's fine. Some deals involve multiple touchpoints, and the customer themselves might not remember exactly how they found you. Do your best with what you have. Even a rough attribution is dramatically better than no attribution, because it shifts you from assumption-based budgeting to data-informed budgeting.
Should I include my team's time in channel cost calculations?
Yes. This is one of the most common mistakes in channel cost analysis. If a founder spends 10 hours a week producing LinkedIn content, that time has a real cost. If an SDR spends half their week managing cold email sequences, that labor is part of the channel's expense. Excluding team time makes certain channels look artificially cheap and skews your decision-making.
Is cold outbound email really dead?
Cold email as a volume play is effectively dead for most B2B companies. The combination of stricter spam filtering from major providers, AI-powered content detection, and plummeting response rates has made the traditional spray-and-pray approach unviable. Highly personalized, low-volume outreach to warm prospects can still work, but that's a fundamentally different motion than what most companies mean by "cold outbound."
Why were paid ads our worst-performing channel despite targeting the right audience?
Targeting the right audience on LinkedIn or Meta doesn't guarantee quality conversions. Paid ad leads are often earlier in their buying journey, less committed, and more price-sensitive than leads who seek you out through content, referrals, or events. We found that the lifetime value of ad-acquired customers was meaningfully lower than customers from organic channels, which compounded the already-high acquisition cost.
How often should I redo this analysis?
At minimum, quarterly. Channel economics change, sometimes quickly. Cold email went from being one of our best channels to being completely ineffective within about a year. If we had been running attribution analysis regularly, we would have caught the decline faster and reallocated budget sooner.
Related: How Much Does It Cost to Build an App? · How Much Does It Cost to Build an MVP?
