I've been in sales for 14 years, 8 of them in B2B SaaS, and closed €4M+ along the way. The outbound engagements I run all end up following the same six steps in the same order, and it's the thing I explain on almost every first call, so it may as well have a name and a URL. The steps are boring on purpose, and each one exists so the step before it doesn't get wasted.
The six steps
- 01ICP from closed deals
- 02Buying signals
- 03Separate, warmed infrastructure
- 04Sequences routed by fit
- 05Reply routing
- 06The weekly tune
Pull the ICP out of deals you actually closed #
- What it is
- A written profile of who buys, built from your closed-won and closed-lost deals. Company type, sector, the seat of the person who signs, and the moment that made them need you. For my own business that's founders and sales leads at small B2B SaaS teams in the Netherlands, Belgium, DACH and the UK, and agencies and recruiters are excluded outright.
- Why it exists
- It's the document every later step inherits. The list, the signals, the copy and the channel choice all come out of it, so a vague ICP gets more expensive at each step after it.
- The failure it prevents
- Outreach going to whoever was easiest to find. "B2B SaaS companies in Europe" is a market, and you can't build a list from a market, so nobody does. This is where outbound usually breaks, and then the copy gets blamed for it.
- How you know it's done
- You can hand the document to someone who's never met you, and the list they build from it is one you'd sign off without edits. If two people on your team describe the ideal customer differently, it isn't done.
Wait for a buying signal before you send #
- What it is
- A watch layer on top of the ICP that flags an account the moment something changes: a funding round, a first sales hire, someone new in a buying seat, a tool they just adopted, a pricing page that doesn't exist. Every contact gets verified and scored before anything goes out, and that score decides how hard we go in step four.
- Why it exists
- Timing does most of the work in cold outreach. An email triggered by a real event could only have been written this week, and the reader can tell. Of everything I send, the trigger-event email is the one I trust most.
- The failure it prevents
- Pitching a perfect-fit company at a moment when it doesn't have the problem. And bounced email: sending to guessed addresses is the quickest way I know to burn a sending domain.
- How you know it's done
- Every contact in a send has a reason attached that you could say out loud in one sentence, and an address that's been verified.
Send from infrastructure that can't hurt you #
- What it is
- It's separate sending domains, never your company domain, with SPF, DKIM and DMARC set properly, a warmup period before the first real send, and a hard daily cap per domain. Plus a suppression list that's loaded before anything goes out: existing customers, open deals, anyone already in another campaign, and your first-degree LinkedIn connections.
- Why it exists
- Deliverability decides whether steps one and two count at all. A perfect message in a spam folder gets no replies, and inbox providers keep tightening their filters. Warmup is also the one thing in here you can't rush, so the domains get bought first, before the ICP is even finished.
- The failure it prevents
- A burned main domain, which means your invoices and your real sales follow-ups start landing in spam too. And the one that does more damage to a relationship: a cold email hitting someone who's already a customer, or the same person getting pitched twice from two campaigns.
- How you know it's done
- Authentication passes on every sending domain, warmup has run its course, the daily cap is enforced by the tool itself, and the suppression list is loaded and tested against a few names you know should be blocked.
Run sequences, with the channel picked by fit #
- What it is
- Short email sequences written around the prospect's problem, in the buyer's own language, with LinkedIn running alongside at a human pace. The score from step two picks the mix: the best-fit accounts get email and LinkedIn at the same time, the middle tier gets email first and LinkedIn later, and the rest get email only, because they haven't earned the scarce channel yet. For Venture Challenge that also meant a slower follow-up rhythm, built around how universities and accelerators actually make decisions.
- Why it exists
- The two channels do different jobs. Across 49,848 cold emails we measured a 0.89% reply rate. On LinkedIn, 45.7% of connection requests were accepted. LinkedIn is far stronger per touch, and it's the scarce channel, because LinkedIn caps how much of it you get. So you spend it where fit is highest and let email carry the volume.
- The failure it prevents
- Spending the scarce channel on accounts that barely fit, getting the LinkedIn account restricted by pushing it too hard, and follow-ups that say "just bumping this". Every touch needs a new reason to care, or it eats the goodwill the first one earned.
- How you know it's done
- Every tier has a sequence in the right language, each follow-up adds something the previous one didn't, LinkedIn runs with a minimum gap per person, and nobody you already know is in a cold sequence.
Get every reply to a human the same day #
- What it is
- Each reply gets classified (interested, not now, wrong person, unsubscribe), enriched and routed to whoever owns it, with a draft answer waiting for approval. The sequence pauses the moment someone answers, on both channels.
- Why it exists
- This is where the pipeline actually lives, and it's the part agencies skip. A warm reply cools off fast. If it sits unread while the sequence keeps firing, you've paid for the whole machine and dropped the one thing it produced.
- The failure it prevents
- The hot reply that sits in a shared inbox over a weekend. Or worse, the prospect who answers and then gets the next automated follow-up anyway. Nobody forgives an automated second one.
- How you know it's done
- Every reply from the past week has an owner and an answer, and you can't find a single prospect who replied and was still mailed afterwards.
Tune it every week, review it every month #
- What it is
- It's a fixed weekly slot: build new segments as the old ones run dry, change copy based on what actually got replies, check deliverability before it's a problem, and look at conversion between each stage. Once a month the whole thing gets reviewed on numbers: what was sent, what replied, what booked, what closed, and which segment we're retiring.
- Why it exists
- Outbound decays. Domains age, lists go stale, inbox providers change their filters without telling anyone, and a sequence that booked calls in spring goes quiet by summer. What the weekly review learns goes straight back into step one, which is why this is a loop and not a launch.
- The failure it prevents
- Two opposite ones. A setup that's handed over and left alone, which is worth less every week. And a motion that gets judged while it's still being set up, then switched off before it ever had a fair run.
- How you know it's done
- Honestly, never. The practical test: each week there's a short written note of what changed and why, and the monthly report shows you exactly the numbers I'm looking at. If a month doesn't justify the next one, you should be able to see that yourself.
What it has produced
Venture Challenge went from a founder doing all the outreach herself to 170 qualified leads in 90 days, and 25 teams enrolled at €5K each from a single cohort. IKI Health had leads scattered over spreadsheets and a founder who'd touched every one of them. In month one they had 30+ scheduled calls, and 2 high-ticket deals closed straight away.
The channel numbers in step four come out of the same engine, and they're published in full on the cold outreach benchmarks page: a 0.89% reply rate on 49,848 cold emails and 45.7% of LinkedIn invites accepted, counted across every campaign, the dead ones included.
Cite this method: Sell Successfully, The Signal-to-Reply Method, sellsuccessfully.io/method/signal-to-reply/. If you'd rather point at a single step, link its anchor, for example #reply-routing.
Free to reuse under CC BY 4.0: credit Sell Successfully and link this page.
Questions
Can I run this from my own company domain?
I wouldn't. Separate sending domains with their own warmup mean a deliverability mistake can never touch your main company email, and that's the domain your invoices and your real conversations depend on. It's cheap insurance and it's set up in the first month.
Where does AI come into it?
In the parts that are pattern matching: finding companies that fit the ICP, watching them for signals, verifying contact data, drafting first-line personalisation and classifying replies. The judgement calls stay human, and so do the calls themselves. Anyone selling you a fully autonomous AI SDR is selling you bounced email.
When is a company too early for this?
When you haven't closed enough deals to see a pattern, because then step one has nothing to work from. Do it by hand first: write to a handful of genuinely relevant people yourself, read every reply, and systemise what works once you can see it.
If you'd like me to tell you which of the six is breaking in your outbound, the free audit is where I'd start.