How this track works
Ten sessions, one running territory. You are an AE at Cadence - AI meeting intelligence for operations teams - carrying eight named accounts into a new quarter. The anchor account is Northwind Logistics: they signed a new Jurong distribution-centre lease five weeks ago and have three warehouse-ops roles open. Tonight is the frame. b2 builds the account research and the prospect list, b3 writes the outreach (and is where tonight's simulator lives for real), b4 preps the first call, b5 runs it and captures it, b6 handles the objections, b7 writes the proposal and the follow-up, b8 fixes the CRM, b9 forecasts honestly, and b10 assembles the whole deal and scores it. Every session ships something you can use on Monday morning.
The deal loop: five stages, one line through them 7 min live
Every deal you will ever run passes through the same five stages: Target (is this account even real?), Reach (earn a reply), Discover (learn what is actually true), Advance (make the next step easy), Record (make it findable by someone who is not you). AI is genuinely good at a different slice of each one, and genuinely dangerous at the rest. The line below is the whole course in one picture - the sessions just zoom in on one column at a time.
LiveThe five levers of an email that earns a reply4 min▶
The deal loop is how the quarter is shaped. The five levers are what one email needs, and they are the five ways cold outreach dies:
- Trigger event - a dated thing that happened at their company. Timeline-based hooks outperform problem-based hooks by roughly 2.3x in reply rate. This is the single biggest lever you own.
- Named pain - the problem in the buyer's words, specific enough that guessing wrong would be obvious. Category language ("visibility", "efficiency") is not pain.
- Proof point - one comparable customer and one number. Not a logo wall.
- Specific ask - a question answerable in one line from a phone. "Do you have 15 minutes Thursday or Friday?" asks the busiest person in the building to do calendar admin for a stranger.
- No AI tells - under 80 words, no "I hope this email finds you well", no "I noticed that you are", no leverage/streamline/synergy, no three-paragraph wall.
There is a sixth toggle in tonight's simulator - max personalisation tokens. It is not a lever. You will find out what it does in Demo 1, and get the full treatment in b3.
The 3.7x is real, and so is the catch. Gartner surveyed 1,026 B2B sellers and found those who effectively partner with AI are 3.7x more likely to meet quota. The same research house also found AI now saves sellers nearly five hours a week - and that 72% of sales organisations fail to reinvest that time in anything high-value. The hours come back automatically. The quota does not.
Self-studyTarget is upstream of everything, and nobody wants to hear it3 min read▶
Reps ask for better copy. Copy is stage 2. Almost every "our outreach is not working" conversation is actually stage 1 wearing a disguise, and you will feel this yourself in tonight's Demo 2 - three of the eight prospects have a hard ceiling that no sentence you write can raise:
- Wrong seat - right company, person with no budget and no pain. Best case they forward it.
- Under the ICP floor - a company that cannot buy at your entry price. A perfect email to them is still a no, and it burns a domain reputation you need.
- Prior opt-out - the row that must score zero. Sending here is a compliance failure, not a missed number (b3 and leader session a5 cover PDPA, CAN-SPAM and CASL properly).
AI makes it trivially cheap to email all three. That is exactly why the discipline matters more now than it did when a list took a week to build.
The trust line: what AI must not cross 6 min live
Here is the tension nobody selling you sales-AI will put on a slide. Adoption is near-universal and the productivity case is real - and Gartner predicts that by 2030, 75% of B2B buyers will prefer sales experiences that prioritise human interaction over AI. Meanwhile sending got free, volume went vertical, and platform-average reply rates fell from about 5.1% to about 3.4%. Your buyer is drowning in machine-written mail and has become extremely good at spotting it. Three things must stay unmistakably human, or the rest of the automation stops working.
LiveThe productivity trap: five hours back, nothing to show for it4 min▶
The most useful finding in the current sales research is not about output quality at all. Gartner: AI gives sellers back nearly five hours a week, and 72% of sales organisations fail to reinvest that time in high-value activities. The hours reappear as more admin, more tabs, and more emails sent to the same tired list.
- What actually correlates with growth - Gartner found organisations giving sellers AI-enabled next best actions are 2.6x more likely to achieve commercial growth, and those that emphasise upskilling sellers on AI are 2.4x more likely to see good revenue growth. The tool is not the intervention. The habit is.
- The agent forecast, read honestly - Gartner predicts AI agents will outnumber sellers 10 to 1 by 2028, while fewer than 40% of sellers will say agents improved their productivity. Plan for the tools arriving whether they help or not.
So the question this track keeps asking is not "what can AI do in sales?" It is: what did you do with the five hours? Every session ends by naming where the saved time goes.
Self-studyWhat the benchmark data actually says about cold email3 min read▶
Four numbers worth carrying into b3, all from published 2025-2026 benchmark reports on large cold-email samples:
- The average is falling. Platform-average reply rates around 3.4%, down from roughly 5.1% two years earlier. Top-quartile senders still land 15-25%. The spread between average and good got wider, not narrower.
- Timeline hooks beat problem hooks by about 2.3x. "You opened a DC five weeks ago" outperforms "struggling with visibility?" - which is why Trigger event is the heaviest lever in tonight's simulator.
- Short wins. Best-performing cold emails run under 80 words. Your instinct to explain the product is the thing killing the reply.
- Follow-ups matter more than the first send. A 3-7-7 style cadence captures the large majority of total replies by around day 10. One-and-done sending throws away most of your own work (b7).
Benchmarks are directional, not gospel - vendor samples skew to their own users. Treat them as the shape of the curve, and measure your own.
Watch one email climb ★ 12 min · everyone runs it
One real prospect: Mira Chen, VP Ops at Northwind Logistics, five weeks into a new Jurong DC. Below is the email with every lever off - the version that arrives in her inbox forty times a week. Turn the levers on one at a time and read what changes in the actual text, not just the number. 2.8% → 19.2% is the ladder. Then, at the top, turn on the sixth toggle and watch what happens.
All levers off: 2.8%. Count the tells the simulator lists. Read the email out loud - it is about you, not her.
+ Trigger event: 6.3%. One dated, public fact replaces the entire "I noticed that you are" opener. Biggest single jump on the board.
+ Named pain: 9.5%. Watch the jargon paragraph get replaced by something a VP Ops would recognise as her own Tuesday.
+ Proof point and + Specific ask: 11.9% then 14.2%. Note that the ask got smaller as the number got bigger.
+ No AI tells: 19.2%, and the word count collapses under 80. Nothing was added here. Only removed.
Now turn on ⚠ Max personalisation tokens. Read the subject line and the first sentence before you look at the number.
The 8-prospect sequence - your course-long benchmark ★ 10 min · explore
Same levers, now across the whole week of sending: eight real-shaped prospects, including the ones you would rather not think about. The aggregate ladder is 1.7 → 3.6 → 5.2 → 6.4 → 7.6 → 10.1. Notice that your best email averages 10.1% while Mira alone hits 19.2% - the gap is targeting, and no amount of writing closes it. Three rows are capped on purpose. Find them.
All levers off: 1.7% average. Every row red. This is what a "we sent 400 emails" week actually looks like.
+ Trigger: 3.6%. + Named pain: 5.2%. Watch which rows move and which do not budge at all.
+ Proof: 6.4%. + Ask: 7.6%. + No tells: 10.1%. Three rows are still red at the top of the ladder.
Read those three rows' text: wrong seat, under the ICP floor, prior opt-out. Say out loud which stage of the deal loop each one belongs to.
Now turn on ⚠ Max personalisation tokens at 10.1%. Read Sarah Lim's row specifically - the one prospect who said her pain out loud on a public call.
The token trap is the most common real mistake in modern outbound. The research says personalisation depth lifts replies - so teams buy more merge fields and spray them at everyone, which is not depth, it is decoration. Depth means a human read something. Decoration means a machine filled a slot, and the day a field is empty your buyer sees "Hi {{first_name}}" and learns exactly what she is to you. Sarah Lim's row is the whole lesson: the prospect most worth an email is the one a template insults hardest.
Try it yourself - this week ◐ 25-35 min total
- Run the full ladder on Demo 2 until you can predict which rows move before you click. Write one sentence on why the three capped rows never move.
- Take the last cold email you actually sent. Score it out of 5 on the levers - trigger, pain, proof, ask, no-tells. Most reps score 1 or 2. That is the point.
- Rewrite that email with all five levers on and get it under 80 words. Send it to one real prospect this week and note what comes back.
- Draw the deal loop from memory - Target, Reach, Discover, Advance, Record - and mark where your last three lost deals actually died. If two of them died at Target, b2 is your session.
- Answer the course question once, honestly: if AI gave you five hours back this week, where did they go?
Official sources covered
This track teaches the working core of the current sales-AI evidence base - Gartner's seller and buyer research, published cold-email benchmark reports, and the compliance regimes that actually govern outreach - with a live simulator on the pages where a number tells the story better than a paragraph. Vendor certifications stay with the vendors. This page covers:
Three questions before you go 🎯 ◐ 90 seconds
1 · In Demo 1, which single lever produces the largest jump in predicted reply rate?
Trigger takes 2.8% to 6.3% on its own. Timeline-based hooks outperform problem-based hooks by roughly 2.3x in the published benchmarks, which is exactly how the simulator is weighted. It is also the lever that requires actual homework, which is why most reps skip it.
2 · At the top of the ladder the sequence averages 10.1% while Mira alone sits at 19.2%. What explains the gap?
Target is upstream of Reach. A wrong-seat contact, a company below your price floor, and a prior opt-out are all stage-1 problems, and the opt-out row must score zero: sending there is a compliance failure, not a missed number.
3 · You turn on "Max personalisation tokens" at the top of the ladder and the number drops. Why is that the honest result?
Personalisation depth genuinely lifts replies - depth meaning a human read something specific. Merge fields simulate that at scale and buyers have learned to spot the simulation. The Sarah Lim row shows the worst case: the one prospect who stated her pain publicly is the one the template insults hardest.