- Cold calling outperforms cold email today because the prep work that used to make it unscalable is now done by agents.
- Most teams gave up on the phone for the wrong reason — they assumed it didn't scale, when really the research before the call was the bottleneck.
- A 5-step implementation: precise ICP and personas, agent-built lists, agent-driven research, agent-written openers, the 60-second human script.
- Cody and Magnus run this every day at Evergrowth. 28% connect rate, 68% meeting conversion, ~190 meetings per 1,000 dials.
- The biggest pitfalls aren't the call. They're volume, scripts, and the missing QA loop.
Why cold calling scales now
I sold over the phone for years before I built anything. So when people tell me cold calling doesn't scale, I push back.
It scales just fine. What didn't scale was the work before the call.
The reason cold calling fell out of fashion isn't that the channel stopped working. It's that the prep got harder. Buyers got noisier. Lists got dirtier. And the pre-call research that makes a cold call sound like a peer-to-peer conversation took longer than most reps could afford to spend. So everyone migrated to email, where you could send 500 messages in a click and call it pipeline.
Then email broke. Spam policies tightened. AI flooded inboxes. Reply rates collapsed across the board. And the channel everyone abandoned, the phone, is suddenly the one that earns attention.
Here's the part most teams haven't caught up with. The pre-call research that used to gate the channel is now agent work. A team running the playbook below produces a research-backed call list of 500 accounts faster than the same team builds a sequenced email campaign. That changes the math entirely.
If your reps book a meeting on 65 to 70% of their connected calls, and the prep work isn't the constraint anymore, the channel that scales is the one your buyers still pick up.
How this looks at Evergrowth
This is how our team operates today. Cody and Magnus run this motion every day, and they close the deals they book.
Here's what their dashboard looks like over a recent 60-day window.
Cody connected on 162 of 635 dials. Magnus connected on 118 of 365. Combined: 1,000 dials, 280 connected calls, roughly 190 meetings booked.
Frontspin handles the dials. TitanX scores phone numbers for pickup likelihood. Evergrowth agents handle everything before the dial.
The 5-step implementation playbook
The order matters. Skipping a step or doing them out of order is the most common reason cold calling motions die.
Step 1Precise ICP and personas
This is the work nobody wants to do, and the only work that compounds.
If your ICP is "B2B SaaS, 50 to 500 employees," you don't have an ICP. You have a filter. A real ICP includes qualification logic, disqualifiers, and the signals that separate a fit account from a noise account.
Same for personas. "VP of Sales" is a job title. The persona is what they own, what they care about, what makes them push back, and what makes them pick up the phone. This is the foundation the agents are trained on. Get it wrong and every step downstream amplifies the mistake.
Built in the Agent Training Center · Pairs with the top-of-funnel playbook for outbound prospecting.
Step 2Agent-built lists
Once your ICP and personas are precise, Account Qualification and Contact Finder produce accounts and contacts that fit. The output is a shortlist your reps can call, not a database export they have to clean.
The shift is from filtered to qualified. A filter says "company has between 50 and 500 employees." Qualification says "this company runs a dedicated outbound sales team, sits on Salesforce, and posted three SDR roles in the last 90 days." Only one of those is worth a dial.
For accounts that should be called the moment something changes — a new exec, a funding round, a job posting that signals a strategic shift — wire this up against the signal-based playbook instead of running it as a static list.
Step 3Agent-driven account research
This is the step that decides whether your cold call sounds like a cold call.
Account Research reads the company's website, content, hiring activity, recent news, and product surface area. It pulls out one usable angle per account. Not "they raised a Series B." Something specific enough to open a real conversation. A new VP of Sales who used to run a team that bought a competitor. A job ad asking for a skill that signals a strategic shift. A new product line that breaks an assumption your messaging is built on.
Run this on a schedule, not ad hoc. The research stays fresh, and the rep walks into every dial with current context. Contact Research does the same job at the persona level, and Account Planning turns the angle into a sequenced next step. When research is the trigger — a 10-K change, an event invite, a customer logo move — the event-based playbook is the natural extension.
Step 4Agent-written cold call opener
Play Copywriting takes the research output and writes a 30-second opener grounded in it. Not a script the rep reads word for word. A briefing the rep reads before the call.
The opener has one job. Earn the next 30 seconds. It does that by referencing something specific the rep wouldn't know unless they'd done their homework. The agent did the homework. The rep delivers it. The same engine drives the openers used inside the top-of-funnel and signal-based playbooks — only the trigger differs.
The full workspace context behind the opener
Step 4 only works because every prior layer of context lives in the same view. By the time the rep clicks into a contact, the Agent Training Center on the left has already trained the system on value props, key accounts, ICP verticals and personas. The middle of the workspace shows two stacks of agents working on the live record — account-level (Qualification, Research, Planning) and contact-level (DISC profile, LinkedIn insights, Contact Finder) — and the Sales Rep Output column on the right is what the rep actually picks up before the dial: the Play Copywriting opener, the Digital Twin chat with full account context, and the Voice Roleplay practice run against a trained AI buyer persona.
RevOps trains and governs on the left. The rep uses and executes on the right. The Play Copywriting agent on the right reads everything in the middle and writes the cold call track for this exact contact.
This is the artifact the rep reads before they dial. Not a script to recite — a briefing built from every layer of context above.
Account Qualification
Account Research
Account Planning
Contact Qualification
Contact Finder
Play Copywriting
Digital Twin
That's why the opener lands. The rep isn't improvising. The system did the prep — and every agent in that view contributes one layer of context to the same account record.
Step 5The 60-second human script
This is the easiest part to teach and the hardest part to execute.
After the opener earns the first 15 seconds, the rep asks one question grounded in the research. Then they shut up.
Most cold calls die because the rep keeps selling. Ours work because the rep knows when to stop.
A connected call that runs 1 minute 49 seconds on average is doing this right. Long enough to qualify and book. Short enough to respect the prospect's time.
Reps practice the motion before they dial. Digital Twin lets them stress-test the angle in chat with full account context, and Voice Roleplay runs them against a trained AI buyer persona so the live call isn't the first time they say it out loud.
What to watch out for
Three traps that kill the motion if you don't see them coming.
Once the playbook works, the temptation is to triple the dial count. Don't. That breaks the research quality, breaks the opener quality, and within two weeks the connect rate slides back to industry average. The playbook scales by adding reps, not by squeezing more dials per rep.
Reps who memorize the opener word for word sound like they're reading a script. Because they are. The opener is a briefing, not a teleprompter. Train reps to internalize the angle and deliver it in their own words. The research has to be precise. The delivery has to sound human.
Capturing what happens on the call is the part most teams skip, and it's the part that compounds. We run the call transcripts through an agent that pulls the context, updates the account, and drafts the next step. That keeps the playbook from drifting because every conversation feeds back into the workspace, and the next dial on that account starts with what the last one revealed. If you don't close that loop, the playbook degrades quietly. Meeting count alone won't tell you why.
What this means for your team
This is the same playbook every Evergrowth customer can run. The agents and the workflow get deployed in days, not months. The hard part is Step 1, and that's your work, not ours.
If your cold calling motion isn't producing numbers like the ones above, the issue is rarely the rep. It's the context they walk into the call with.
The playbooks behind this motion
The cold calling workflow above is one execution of the Newbiz Gap playbook — agents qualify, research, find contacts and write the play for accounts that aren't in your CRM yet. Below is the live workflow, followed by every playbook our team and customers run today.
Newbiz Gap
Import company names not in your CRM. Agents qualify, research, find contacts, and write plays from scratch.
Newbiz Recycling
Re-qualify dormant CRM accounts with fresh research, re-verified contacts, and updated outreach.
Ex-Customer Recycling
Monitor customer contacts for job changes. CS gets departure alerts, Sales gets warm intro opportunities.
Signal-based Outreach
Define your signal criteria. When a match is found, agents trigger the full workflow automatically.
Event (Pre & Post)
Upload attendee or booth lead lists. Agents qualify, research, and write event-context outreach.
Champion Monitoring
Track when pipeline contacts change companies. Agents research the new company and write warm outreach.
Scheduled Research
Schedule recurring agent runs on accounts and contacts. Surface new signals and create CRM alerts.
Inbound Lead Qualification
New CRM lead from a form fill or demo request? Agents auto-qualify, research, and write follow-up.
Frequently asked questions
Does cold calling still work in 2026?
Yes. Cold calling outperforms cold email today because the prep work that used to make it unscalable is now done by AI agents. With agent-driven research, qualification and openers, a connect rate around 28% and a meeting conversion above 65% are reproducible — well above what most email programs deliver.
Why did teams give up on cold calling in the first place?
Most teams concluded the channel didn't scale. The real bottleneck was the research before the call — pre-call homework took longer than reps could afford. So teams migrated to email, where 500 messages went out in a click. The phone never stopped working. The prep work just stopped fitting in a workday. The channel evolution piece tells that fuller story.
What is a good cold call connect rate and meeting conversion rate?
Industry averages hover around 5% connect and 20–30% meeting conversion of connected calls. With an agent-built list, agent-driven research and an opener written from that research, our internal team runs at a 28% connect rate and a 68% meeting conversion — roughly 190 booked meetings per 1,000 dials.
How long should a cold call be?
Around 1 minute 49 seconds on average for a connected call. Long enough to deliver a researched opener, qualify, and book the next step. Short enough to respect the prospect's time. Calls that drift past three minutes usually mean the rep kept selling instead of asking one good question grounded in the research.
What kills a cold calling motion most often?
Three things. The volume trap — tripling dial count once it's working, which collapses research quality. The scripted-opener trap — reps memorizing the opener word for word, which makes them sound like a script. And the QA gap — failing to feed call transcripts back into the workspace so each next dial starts from what the last one revealed.
