Where we are
b2 built the list: eight named accounts, enriched, ranked, with the bad-fit rows already killed. Tonight is stage 2 of the deal loop - Reach. The deliverable is one email you will actually send to a real person this week, scored 5 out of 5 on the levers and under 80 words. b4 preps the call it earns you.
The ladder: five levers, one email 6 min live
Every lever is a different kind of homework, and they are not worth the same. Trigger event more than doubles the rate on its own because it is the only one that proves a human looked. The rest compound on top of it. The last rung, No AI tells, is the only one where you add nothing at all - you delete, and the number goes up by five points.
LiveWhat each lever actually costs you4 min▶
Reps skip levers by price, not by value. Here is the honest cost of each one:
| Lever | What it costs | Where it comes from |
|---|---|---|
| Trigger event | 4-6 minutes per prospect | Their news page, LinkedIn posts, job reqs, earnings call, a lease or permit filing |
| Named pain | Once per segment, not per prospect | Your last 10 discovery calls. The buyer's words, not your category's |
| Proof point | Once per segment | One comparable customer, one number, one time period |
| Specific ask | Seconds, once you stop asking for time | A question they can answer in one line from a phone |
| No AI tells | Seconds - you are deleting | The checklist in Part 2. Under 80 words |
Two of the five are per-segment work you do once and reuse for a quarter. Only the trigger is genuinely per-prospect, and it is the one worth the most. That ratio is the entire argument for AI in outreach: let it do the finding, keep the writing.
LiveWhat AI drafts, what you write3 min▶
The working split that survives contact with a real week:
- AI finds and summarises - scan the account's last 90 days, surface candidate triggers with dates and links, tell you which of your segment pains plausibly applies.
- You choose the trigger and write the first line. This is the sentence the entire email lives or dies on. It is 20 words. Write it yourself.
- AI tightens - cut to under 80 words, strip jargon, flag tells, offer three subject lines.
- You own the ask and the send. Nothing goes out unread. Ever.
Here is a company: Northwind Logistics, 3PL, Singapore, ~900 staff. Find every dated, public thing that happened to them in the last 90 days: funding, leases, facility openings, exec hires, job postings, earnings remarks, product launches, regulatory filings. For each: the date, one sentence of what happened, the source link, and which of these three pains it plausibly creates: (1) decisions lost between meetings during a ramp (2) new sites out of sync with HQ (3) new managers with no shared operating record Rank by how recent and how specific. Do not write an email.
Write a cold email to the VP of Operations at Northwind Logistics selling our AI meeting intelligence platform. Make it professional and persuasive, and personalise it.
The second prompt has no trigger to work with, so the model fills the gap with the only thing it has: your product. That is the whole reason generated outreach sounds like generated outreach.
The eight tells, and the two seconds you get 6 min live
Your buyer does not read the email. She scans the first line, the shape of the paragraphs, and the ask. Every tell below is something she has now seen a hundred times this month, and each one is a signal that no human spent any time on her. The right-hand column is the same email once the tells are gone - shorter, more specific, and noticeably less polite.
Self-studyWhy "personalisation at scale" turned into an insult3 min read▶
The benchmark data genuinely supports personalisation: deeper personalisation and tighter cohorts lift reply rates by multiples, and advanced-personalisation campaigns report roughly double the replies of generic ones. Teams read that, buy a tool with forty merge fields, and get worse results. Here is the gap.
- Depth means a human read something. One specific, dated, checkable fact. It cannot be produced by a field lookup because the value is the reading, not the data.
- Merge fields simulate depth. "{{first_name}}, as a fellow {{industry}} leader in {{city}}" contains four facts and zero attention, and buyers have been trained by volume to spot the pattern instantly.
- Empty fields are catastrophic, not cosmetic. The failure is visible to the buyer and unambiguous: she is a row in a spreadsheet, and now she has proof.
The anti-lever in tonight's simulator is calibrated to behave the way this actually behaves: it lowers the rate at every rung, and on the one prospect who stated her pain publicly it produces the worst outcome on the board.
Self-studyBefore you send: the rails that decide whether you may4 min read▶
Copy quality is a performance question. Whether you are allowed to send is a legal one, and it is not optional. The three regimes most likely to touch a B2B list, in plain terms:
| Regime | Applies to | The rule that bites |
|---|---|---|
| PDPA (Singapore) | Personal data of individuals in Singapore, plus the Do Not Call registry for phone numbers | Consent or a valid exception for the data; check DNC before calling or texting a Singapore number |
| CAN-SPAM (US) | Commercial email to US recipients | Accurate headers, no deceptive subject lines, a physical postal address, and a working opt-out honoured promptly |
| CASL (Canada) | Commercial electronic messages to Canadian recipients | Consent-first, with limited implied-consent windows. Penalties are high and the burden of proof sits with the sender |
| GDPR (EU/UK) | Personal data of people in the EU or UK | You need a lawful basis. B2B legitimate interest is possible but must be documented, balanced, and easy to object to |
- The suppression list is a rail, not a preference. A prior opt-out in your CRM must score zero sends, forever. That is the eighth row in tonight's sequence and the reason it is capped at 0%.
- AI raises the stakes. Volume that used to take a week now takes a minute, so a compliance mistake that used to touch 40 people now touches 4,000.
- Get it checked. This card is orientation, not legal advice - your list, your jurisdictions, your counsel. Leader session a5 covers the org-level version.
Finish the climb, then break it ★ 14 min · everyone runs it
The simulator starts tonight where b1 left you at the halfway point - trigger and pain on, 9.5%. Add the last three levers and read the email text at every step, not just the number. Then do the thing every outbound team eventually does: switch on the tokens.
Read the 9.5% version out loud. It is already better than most of what your buyer receives. It is still 77 words of you.
+ Proof point: 11.9%. One customer, one number, one time period. Notice how little space it takes.
+ Specific ask: 14.2%. The old ask wanted 15 minutes of her calendar. The new one wants one line from her phone.
+ No AI tells: 19.2% and 71 words. Nothing was added. Read what got deleted.
Now switch on ⚠ Max personalisation tokens: 10.6%. Read the subject line before the number.
Switch it off. Turn off Trigger event instead and watch the single biggest drop on the board. That is the lever you must never skip.
The sequence, with the rails on ★ 10 min · explore
Now the whole week: eight prospects, all five levers already on, sequence average 10.1%. Tonight the interesting rows are the ones that stay red no matter what you write - because two of them are targeting mistakes and one of them is a compliance failure, and only one of the three is fixable by trying harder.
Read Jonas Weber's row (1.8%, capped). Right company, wrong seat. Which stage of the deal loop failed?
Read Anna Kovac's row (1.2%, capped). Perfect email, company below your price floor. This one costs you domain reputation for nothing.
Read Marcus Hale's row (0.0%, capped). Prior opt-out. This is the only row where a zero is the correct, professional outcome.
Turn every lever off and back on and confirm those three never move. Copy cannot fix targeting, and nothing fixes an opt-out.
Turn on ⚠ tokens and read Sarah Lim's row - the prospect who said her pain out loud on an earnings call.
The three capped rows are where most outbound budget actually goes. Sending to wrong seats, under-ICP accounts, and stale lists is invisible on a dashboard that counts sends, and it is the reason platform-average reply rates fell from about 5.1% to about 3.4% while every team reported "more activity". Your reply rate is a targeting metric wearing a copywriting costume.
Try it yourself - this week ◐ 40 min total
- Take the five accounts from b2. Spend six minutes each finding one dated, public trigger. Write the date and the source link next to each. If you cannot find one in six minutes, that is data about the account.
- Write one email per account at 5/5 on the levers, each under 80 words. Use the finding prompt from Part 1 - and write the first line yourself, always.
- Run all five through the eight-tell checklist. Count how many tells survived your first draft. Most people find three.
- Check your list against the rails: any Canadian recipients, any prior opt-outs, any DNC-registered Singapore numbers. Fix before sending, not after.
- Send them. Log what came back in your CRM with the trigger you used - b8 turns that into a pattern and b9 turns it into a forecast.
Official sources covered
This session teaches the working core of the published cold-outreach evidence plus the compliance regimes that govern it. Benchmarks are directional - vendor samples skew to their own users - and this page says so on the simulator itself. Legal specifics stay with your counsel. This page covers:
Three questions before you go 🎯 ◐ 90 seconds
1 · Which lever is genuinely per-prospect work, and therefore the one reps skip most?
Pain and proof are per-segment work you do once and reuse for a quarter. The trigger is the one that costs six minutes per prospect - and it is also the heaviest lever on the board, which is why "let AI find, you write" is the whole operating rule of this session.
2 · Your sequence average is stuck at 10.1% even with all five levers on. What is the most likely cause?
Three of the eight rows are capped by targeting, not writing. Reply rate is a targeting metric wearing a copywriting costume - which is why b2 comes before b3 in this track.
3 · A prospect opted out of your emails eight months ago and is still in the CRM. What is the correct handling?
Opt-outs do not expire on their own, and routing around one through another domain is worse, not cleverer. This is the single row in the simulator where 0% is the professional outcome - and AI-era volume is exactly why the discipline matters more than it used to.