The Problem-Proof-Question refinement method behind cold emails that actually get replies, built from 250,000+ inboxes and 49 founder-led audits.
Before You Send Another 'Personalised' Email
you're personalising every email, and your reply rate is still on the floor.
you swapped in {first name}. you dropped in "loved your recent post." you let an AI tool scrape their LinkedIn and mention the university they went to. and the inbox went quiet anyway.
mate, this isn't the copy problem you think it is.
i've set up cold email infrastructure over 2,000 times in the last 12 months, and i run founder-led audits on live campaigns most days of the week (49 in the last 90). the same thing shows up on nearly every call: the emails are "personalised" and the replies still aren't coming.
the personalisation everyone's doing is the LAZIEST possible version, a merge token wrapped around a scraped fact. Google got tighter, the inbox got more crowded, and that version stopped earning a reply.
this playbook is the fix we actually use. it's built on one method, PPQ (Problem, Proof, Question), and one idea: personalisation by REFINEMENT instead of mail-merge tokens.
what you'll walk away with:
- a 60-second reply-rate self-check that tells you whether the inbox even sees you, and whether your problem is the list or the offer
- the refinement method that makes a cold email feel personal without one scraped "i noticed you went to [uni]" line
- the PPQ worksheet we write every first email against, kept to 80 words that earn the reply
- a before/after rewrite you do on one of your own dead emails
- the section-by-section audit that finds which line is sending you to spam (it's usually the one you'd never suspect)
- a one-page cheat sheet you can pin above your monitor
you can run this on a live campaign today, on the exact refinement process we use when a customer's replies dry up.
let's get your replies back.
Why Generic Emails Stopped Getting Replies
the version of personalisation you're doing used to work. it doesn't anymore.
in september 2025, Google purged the free legacy and nonprofit sending panels half the industry was running on, and it tightened its filters at the same time. bounce rates went through the roof overnight. the inbox got harder to reach, and the bar for "worth a reply" went UP.
at the same moment, AI made it trivial to blast out "personalised" emails. everyone got the same tools. so now the recipient sees ten emails a day that all open with "loved your recent post about..." and every single one gets deleted on reflex.
which is the trap everyone walked into. AI didn't make personalisation better, it made the lazy version universal, and the inbox learned to ignore it.
i'll be honest with you, we got caught doing it too. our own landing-page copy was AI slop for a while, and it converted worse than the page we'd written by hand. i say this on every audit call i take now: AI slop doesn't convert. then half of us go and write our marketing copy with it anyway.
so what gets a reply in 2026? the emails pulling replies now are built on tighter targeting, what i call personalisation by REFINEMENT.
refinement means targeting so tight that the email feels personal without you naming their dog. you write to one avatar about one problem they're DEFINITELY having right now. you show you've fixed it before, and you ask one simple thing. if the pain is specific enough, they resonate.
the frame for the rest of this playbook is simple. we're not chasing a clever first line, we're raising the one number that pays you: your reply rate.
everything from here is how you do that.
Score Your Reply Rate in 60 Seconds
before you touch your copy, run this read.
pull up your live campaign and grab three numbers. it takes about a minute, and it's the exact read i run live on audit calls every day. first you want your GLOBAL reply rate with every out-of-office counted into the total. then your MANUAL reply rate, meaning the real humans who typed back once you strip the OOOs out. and last, your POSITIVE reply rate, meaning how many of those humans were actually interested.
now read the three in order.
a global rate under about 2% is a placement problem, and personalisation won't touch it until you're landing. you can't get an out-of-office from the spam folder, it's physically impossible, so if the OOOs aren't coming either, you're not reaching the inbox at all. fix that first, because no decent email gets a reply from the spam folder.
a healthy global rate (4-6%+ with OOOs) sitting next to a near-zero manual reply rate points somewhere else. you're landing, but nobody's writing back. that's a list or a timing issue. you're either hitting dead data, or you blasted a public holiday when half your list was on OOO.
the third case is the one this playbook fixes. manual replies are coming in, but almost none are positive. you're landing fine and humans are reading, yet the message itself isn't connecting. your targeting and your offer aren't specific enough to that person. the bar you want here is at least a quarter of your manual replies coming back positive.
write your three numbers here:
global (incl. OOO): ______% manual: ______% positive (share of manual): ______%
if your positive share is under 25%, keep reading. the next three sections are built for you.
Refine, Don't Merge: The Personalisation That Gets a Reply
relevance is what gets the reply. this is how you build it in on purpose.
most "personalisation" is a costume, a {first name} up top and a scraped line about their last post. it looks personal, and it reads like every other templated email that pulled the exact same trick, so it converts like one too.
personalisation by refinement is the opposite move. instead of decorating a generic email with scraped facts, you NARROW who you're writing to until the email could only have been written to them. you're not writing to "B2B founders." you're writing to "cold email agencies running 50 to 500 inboxes whose reply rates just fell off a cliff after the september filter changes." when the pain is that specific, it feels personal because it IS relevant, and you don't need a scraped dog anywhere in it.
the way i say it on every call: "if you address somebody with a pain point that is specific enough, guess what? they'll resonate. that's all that matters."
here's the exact same outreach written both ways, the lazy version first.
FAKE personalisation (what everyone does): "hi {first name}, loved your recent post on scaling! noticed {company} is growing fast. i help companies like yours with lead gen. open to a quick chat?"
REFINED personalisation (what gets a reply): "saw {company} is hiring 3 SDRs this quarter. usually means outbound volume is about to jump, and that's exactly when sending reputation tanks and reps start blaming the copy. we keep inbox placement steady while you scale the team. worth me sending over how we'd set it up?"
notice what changed. the second one names no personal trivia at all. it just knows EXACTLY who it's talking to and what's about to break for them.
your turn (10 minutes). grab one email from a campaign that's getting ignored and rewrite it using refinement instead of scraping:
- who's the ONE avatar? go narrower than "founders", down to "founders of X doing Y who are hitting Z problem"
- what problem are they DEFINITELY having this month? tie it to a trigger like hiring, funding, a launch or a policy change
- rewrite the opener so it leads with THAT problem, and delete every scraped or flattery line
- cut the whole thing to under 80 words
diagnostic: read your new opener out loud. if it sounds like a pitch, rewrite it. it should sound like a message you'd send a mate who happens to have this exact problem.
key takeaway: get the aim tight enough and the email feels personal without a single scraped fact in it.
Write the Email With PPQ (Problem, Proof, Question)
PPQ is the framework you write every first email against.
once your aim is tight, PPQ is how you write the actual email. keep it to three sentences and 80 words. it's the only first-email framework i hand out, and it's the one behind the campaigns that book calls for our customers.
P (Problem): one problem the recipient is DEFINITELY having, stated as a fact about their world. for us that reads as "hiring 3 reps usually spikes your send volume, and that's when placement starts slipping."
P (Proof): one line that shows you've solved this before, kept short and specific. "we keep placement steady for 2,000+ senders doing exactly this."
Q (Question): one soft ask, and here's the bit people get wrong. the question is NOT "can we hop on a call?" it's an offer of value, like "want me to send over how we'd set it up?" it's the old marketing game of giving away the how, on the understanding that most won't do it themselves.
rules that keep it reply-worthy: • 80 words MAX. the preview pane only shows the first sentence or two, so the problem has to hit in line one. • one problem only. cram in two and you dilute both. • lowercase, casual, zero jargon. if it sounds like a pitch read out loud, rewrite it. • leave the links, attachments and calendly out of email one entirely. save the link for when they reply.
filled example (cold email agency avatar):
"saw you're onboarding a few new clients this month. that's usually when sender reputation slips and reply rates drop right as you're trying to prove ROI. we keep inbox placement steady for 2,000+ senders scaling exactly like this. want me to send over how we'd set it up for your accounts?"
under 80 words, and every line is doing a job.
your PPQ worksheet (fill this in for your one avatar):
P (Problem). the one thing they're definitely dealing with right now: _______________________________________________
P (Proof). one specific line that shows you've fixed this before: _______________________________________________
Q (Question). a soft, value-first ask that isn't "book a call": _______________________________________________
now assemble it, then read it out loud. if it runs over 80 words or it sounds like a pitch, cut until it doesn't.
key takeaway: the email has ONE job, and that's to earn the reply. everything else waits until they write back.
Find the Line That's Killing Your Replies
sometimes the aim is right and the replies still don't come. this is how you find the culprit.
a cold email is really four moving parts stacked on top of each other. up top is the problem statement doing the heavy lifting, with the proof statement sitting under it. below that comes the question that makes the ask, and the signature sits right at the very bottom. when a well-targeted email still gets ignored or lands in spam, ONE of those four is usually doing the damage, and you find it by redaction, taking one part out at a time.
run it like a science experiment. this is literally how we audit copy on live accounts.
first, prove the infrastructure isn't the problem. from one of your live sending accounts, send a plain email to a Google inbox and a Microsoft inbox you own. set the subject to "weekend" and the body to "how was your weekend?" if that lands in the inbox, your infrastructure is fine and the copy is the suspect.
then paste your ACTUAL campaign email into the same test. if the plain one landed and this one doesn't, the copy is what's hurting you. so now you redact, one section at a time.
start by taking out the signature and sending it again, then see whether it lands. if it does, the signature was your culprit. if it doesn't, put the signature back, pull the proof line out, and send again. keep working through the sections one part at a time until the email finally lands.
whichever part you removed when it started landing is the part you rewrite.
the one that catches everyone is the signature. we once traced a customer's entire deliverability collapse to their email signature. it took one Loom and thirty minutes to find the villain, a fancy signature with an image and a tracking link buried in it. so the standing rule we give every customer:
keep the signature plain text, your name and your company only. strip everything else out of it, images and banners and tracking pixels included.
your turn (10 minutes):
- send the plain "weekend" email from a live account to a Google and a Microsoft inbox you own
- if it lands, paste in your real campaign email and send again
- redact one section at a time until it lands
- rewrite the guilty section, and strip your signature back to text
key takeaway: when a targeted email still flops, isolate the one section doing the damage before you rewrite anything. 9 times out of 10 it's the signature.
The Reply-Rate Cheat Sheet
keep this one page pinned above your monitor.
THE PPQ FORMULA (every first email): • P (Problem): one problem they're DEFINITELY having, stated as a fact about their world • P (Proof): one specific line that shows you've fixed this before • Q (Question): one soft, value-first ask that isn't "book a call" • 80 words max, lowercase, no links in email one
THE REPLY-RATE READ (run before you touch copy): • global reply rate incl. OOO tells you about placement. under ~2% means you're in spam, because you can't get an OOO from spam. • manual reply rate tells you about list quality and timing • positive reply rate tells you about targeting and offer fit. the target is 25%+ of manual replies coming back positive • read it per-domain and averaged over a week, never off one single inbox
REFINEMENT OVER SCRAPING: • narrow the avatar until the email could only be for them • lead with the problem they're having THIS month, tied to a trigger like hiring, funding, a launch or a policy change • delete every "loved your recent post" and scraped-fact line • if it sounds like a pitch out loud, it gets rewritten
THE SECTION AUDIT (when a targeted email still flops): • the four sections to test are problem / proof / question / signature • plain "weekend" email lands? infra's fine, copy's the suspect • redact one section at a time until it lands, then rewrite that one • the #1 hidden culprit is the signature. keep it plain text, name and company only
BENCHMARKS WORTH MEMORISING: • healthy global reply incl. OOO: 4-6% • positive share of manual replies: 25%+ • first email length: under 80 words • signature: plain text, 2 lines at most
pin it, and run the read before you rewrite a single line.
Want this done for you?
We're the cold email infrastructure team behind 2,000+ businesses and over 250,000 inboxes, with 150+ five-star reviews. When reply rates drop, Richard runs the audit himself, 49 founder-led calls in the last 90 days.
Book a call and we'll walk you through exactly how we'd set your inboxes and copy up to actually get replies.
