Your Call Mining Tool Won't Fix a Broken Sales Feedback Loop

Three out of every five sales teams we track complain that customer feedback from calls never reaches product. It's a chronic pain point — severity 4.1 out of 5, according to our platform data. And it's the dirty secret that undermines the entire premise of Jason Lemkin's latest SaaStr piece.

Lemkin profiles Attention.com, a Series B company automating go-to-market workflows from call transcripts. The thesis is elegant: your prospect conversations are the highest-signal first-party data you own, and most teams throw it away. Attention's co-founder Anis Bennaceur lays out three loops to run off that data — refreshing your ICP monthly from closed-won calls, writing outbound copy that mirrors buyer language, and letting a proactive agent queue next actions.

It's a compelling argument. I buy most of it. But our data suggests a missing piece that makes the whole machine less effective than it could be. The bottleneck isn't call mining technology — it's the broken handoff between sales and product.

We track 47 problems in the Sales & Marketing vertical. The most severe? "Sales team doesn't share customer feedback with product" at 4.1/5. Right behind it: "Inbound leads are not distributed fairly to reps" (3.8/5) and "No systematic way to mine call transcripts for buyer signals" (3.6/5). Notice the pattern. Teams are aware they need to mine calls, but the structural incentive to keep insights within sales — and the lack of a feedback loop to product — means even the best AI transcript analysis becomes an island.

Lemkin's article treats the pipeline problem as if it's purely about generating more outbound. But what about the inbound leads that get mishandled? What about the product feedback that never makes it from a rep's brain to the roadmap? Our data shows 16 problems in the Communication category alone, with average severity 3.6/5. Users report that "key details from sales calls get lost or forgotten" and that there's "no systematic way to mine call transcripts for buyer signals."

Attention's proactive agent — which re-engages lost deals and surfaces at-risk accounts — is a step toward solving this. But it still lives inside the sales org. The real unlock comes when that intelligence flows into product, marketing, and customer success. Imagine an agent that not only drafts an email from a transcript but also pings the product team with a feature request based on the prospect's exact words. That's the full loop.

The author also asserts that "growth is stalling for a lot of B2B companies." Our data paints a more nuanced picture. In the SaaS industry, we track 1,028 problems total. While "retaining customers" and "churn" are top concerns, there are still 287 problems related to "scaling sales" — a sign that many are still growing, just facing specific bottlenecks. So the urgency Lemkin creates might not apply equally to every segment.

For builders considering a tool like Attention, or building their own, here's my advice: don't invest in AI call analysis until you've fixed the feedback pipeline from sales to product. Otherwise you'll get better transcripts — and better outbound — but miss the compounding effect of aligning your entire company around what buyers actually say.

For investors, this is a signal to look for companies that solve the cross-functional loop, not just the outbound optimization. The big opportunity isn't another conversation intelligence tool; it's the layer that routes insights to the right teams automatically.

Lemkin's piece is worth reading — especially the practical plays on refreshing ICP and scoring outbound copy. But when you finish, look inward. Is your sales team holding a treasure chest of product insights that nobody else sees? If so, the best AI in the world won't help until you open the lid.

We track 47 sales & marketing problems across industries. The most common thread? Not lack of data. It's lack of data sharing.