The Outbound Rulebook

The Outbound Rulebook is the running order we follow when we put a new B2B outbound campaign live, written down in the order the work actually happens. Thirteen chapters on the parts that decide whether a campaign works: domains and authentication, warm-up, the customer profile, the list, the message, reply handling, EU compliance, the weekly numbers, and a 30-day plan you can run yourself.

Each chapter carries one rule, the reasoning behind it, the method, the mistakes we see most often, and a checklist you can work through before you send anything. The whole book is on this page. The 44-page PDF carries the same text for printing and sharing.

Get the 44-page PDFFree, one email field

Chapter 0Who this is for, and the one promise

You run a staffing agency in Hamburg, an IT consultancy in Warsaw, a machine shop in Kaunas that wants Nordic customers, or a sales team in Brisbane. Somebody told you cold email still works. Somebody else told you it is dead. Both of them sent from the company domain, bought a list, hit 5,000 people in a week, and stopped when the replies came back empty and the invoices stopped reaching customers’ inboxes.

This book is the other path, in the order we run it for clients: domains, mailboxes, authentication, warm-up, the customer profile, the list, the message, the cadence, reply handling, EU compliance, the weekly numbers, and a 30-day plan you can run yourself.

The promise: by the end you can set up outbound that does not burn your main domain. Your invoices keep landing. Your recruiters keep getting replies from candidates. Your quotes keep reaching purchasing managers. The cold campaign runs beside all of that on separate infrastructure, and if it fails, it fails without taking your company email with it.

What this book will not do is promise you a fixed number of meetings. Anyone who does is guessing. Reply rates move by country, sector and season, and the spread is enormous. In Lithuania we have seen campaigns pull 12.7% replies. In the UK and the Netherlands, 2% counts as a good result. Same method, same operator, different market.

What goes wrong

• People treat outbound as a copywriting problem. It is a plumbing problem first, a targeting problem second, and a copywriting problem third.
• People run it for four weeks, see two meetings, and quit. Two months is the minimum honest test. Three is better.
• People build the machine and then nobody answers the replies. We have watched clients sit on five interested replies for two days while the prospect went and googled a competitor.

Checklist before you turn the page

• You can name, in one sentence, what your company sells and who buys it.
• You have somebody who can answer a reply within the hour during working days.
• You have a budget for at least two months of sending, not one.
• You accept that month one produces infrastructure and data, and month two produces meetings.

Chapter 1The math first

Why

Every argument about subject lines is a distraction until you know the shape of the funnel. The funnel decides how many contacts you need, which decides how many mailboxes you need, which decides how many domains you buy and how long the warm-up takes. Get the math wrong at the start and you build a machine that is either half idle or permanently over its send limits.

The math also settles the patience question. When a client asks whether outbound is working after three weeks, we show them the arithmetic. Three weeks at 40 sends a day is 840 emails. At a 2% reply rate that is 17 replies, of which maybe 4 are interested. Four conversations is not a verdict on anything.

The how

Six numbers, in order.

1. Sends. Every message that leaves a mailbox, opener and follow-ups counted separately. A four-step sequence to 1,000 people is roughly 3,000 to 3,500 sends, because replies and unsubscribes pull people out along the way.
2. Delivered. Sends minus bounces. Keep bounces in the low single digits. Our own campaigns run between 1% and 2.5%. The consultancy run we cite later came in at 1.2%.
3. Replies. Everything that comes back, including “no thanks” and “wrong person”. Published 2025 and 2026 benchmark studies put the B2B cold average between 3% and 6%, well-run campaigns at 8% to 12%, and anything sustained below 1% as a data or deliverability fault. Those are industry rules of thumb, not your forecast. Our own reads vary more: a Lithuanian consultancy at 4.9% across 6,924 people, a Polish manufacturer selling into the Nordics at 1.1% across 6,182, our own recruitment run at 1.1% across 8,156 in nine countries, and inside that run a Polish blue-collar staffing campaign at 7.1% across 337.
4. Positive replies. The share of replies that want something. A quarter to a half is the working range. The Lithuanian consultancy hit 59% of replies positive, which is unusually high and came from an offer with a deadline attached. The Nordic manufacturer hit 20%. Our own recruitment run hit about 10%, which told us the message was wrong before it told us anything about the market.
5. Meetings. One third to two thirds of positive replies turn into a call. The variable is you, not them: speed of response and whether booking takes one click.
6. Customers. Depends entirely on your close rate. Use your real one.

Worked example: a staffing agency in Germany

You place industrial and warehouse workers around Hannover. You want new client companies, not candidates.

• Target list: 2,000 decision makers at manufacturers and logistics operators in Lower Saxony with open blue-collar vacancies.
• Sequence: four steps. Total sends about 5,600 over the month.
• Bounces at 2%: about 112. Delivered about 5,488, reaching roughly 1,960 of your 2,000 people.
• Reply rate at 2% of people reached, which is a fair DACH assumption: about 39 replies.
• Positive at 25%: about 10 interested companies.
• Meetings at half of those: 5 calls booked, 4 held after no-shows.
• Close rate at one in four: one new client company.

One client signs, places four workers, and the margin on those placements over a year pays for the campaign several times. Run the same arithmetic with your own placement margin before you spend anything.

Now reverse it. Three new client companies a month needs 6,000 contacts, roughly 17,000 sends, roughly 770 sends a working day. That is an infrastructure decision, and Chapter 2 prices it.

What goes wrong

• People quote reply rate on delivered mail and forget that platforms compute it differently. Salesforge, for one, computes its reply percentage on delivered mail rather than on people contacted, so the same campaign shows two different numbers depending where you read it. Pick one denominator and hold it.
• People count every reply as a lead. “Please remove me” is a reply.
• People plan for the average and get a sector at the bottom of the range. Plan for the low end. Be pleased by the high end.

Checklist

• You have written down your six numbers with your own assumptions.
• You know how many contacts a month the plan needs.
• You know what one new customer is worth over twelve months.
• You have agreed with yourself to run it for two months minimum before judging.

Chapter 2Domains and mailboxes

Why

Mailbox providers score reputation at the domain level. A campaign that collects spam complaints drags down everything sent from that domain, including the quote your sales manager sent an hour earlier. We have met owners who lost their main domain to one enthusiastic week of sending and had to change their address to get out of the spam folder.

Separate domains ring-fence the risk. If a sending domain burns, you retire it and spin up another. Nothing touches the company address.

The how

Buy variants that read as yours. If your company is kofitech.eu, buy kofitech.com, try-kofitech.com, kofitech-hq.com. Redirect each with a 301 to your real website, so a prospect who types the domain lands on a real company. Random words with no connection to your brand kill the reply.

How many. For 100 to 150 sends a day, buy 3 to 5 domains. For larger client runs we buy 5 to 10 and spread 20 to 30 mailboxes across them. Do not register twelve lookalike domains at one registrar inside an hour, because that pattern itself reads as spam infrastructure.

Mailboxes per domain. Two or three, with real human names, photos and a plain signature. Gediminas at hello-yourbrand.com, g.petraitis at yourbrand-hq.com. Never sales@ or info@, which go to an administrator who forwards nothing.

Daily caps. A warmed Google Workspace mailbox carries 20 to 50 cold sends a day, Microsoft 365 tenants sit lower, and new mailboxes start at 10 to 20. Run at roughly 70% of the safe limit so a busy day does not spike you. Three mailboxes per domain at 30 sends each gives one domain a working capacity of 40 to 150 a day. Plan with that number. These caps are practitioner rules of thumb rather than published limits, and they have held across our client runs.

SPF. A DNS TXT record listing which servers may send for the domain. Two hard limits: only one SPF record per domain is valid, and the record may trigger no more than 10 DNS lookups. Past ten lookups SPF silently stops working and nobody tells you.

DKIM. A cryptographic signature on every message, verified against a public key in DNS. Your mailbox provider generates the key. Publish it before the first send.

DMARC. The policy telling receivers what to do when SPF or DKIM fails, plus a reporting address. Publish at p=none, read the reports for two to four weeks, then tighten to quarantine, then reject. Since February 2024 Google and Yahoo require bulk senders to run all three, to offer one-click unsubscribe, and to keep spam complaints under 0.3%. Those are published requirements, not opinions.

BIMI. The logo that appears next to your name in Gmail. It needs DMARC at enforcement and a Verified Mark Certificate, which is an annual cost in the four-figure range. Put BIMI on your main brand domain where your real correspondence goes. Skip it on cold sending domains, where it buys nothing.

Tracking domain. A sending platform that tracks opens inserts a pixel on a shared tracking domain thousands of other senders also use. Point a subdomain of your own sending domain at it by CNAME, or better, turn open tracking off on cold campaigns. Apple Mail Privacy Protection made open data unreliable, and the pixel costs deliverability to collect it.

When to buy. At the moment you decide to run outbound, not the moment you want to send. Infrastructure plus warm-up takes four to six weeks. Our client campaigns go live within 14 days of kickoff for that reason: the first two weeks go to warming, then sending starts.

Cost table

Approximate 2026 list prices in EUR. Ranges, not quotes.

Item	Unit	Typical cost
Sending domain (.com/.eu)	per domain per year	10 to 20
Mailbox, Google Workspace	per mailbox per month	6 to 8
Mailbox, dedicated cold-email provider	per mailbox per month	2 to 4
Sending platform	per month	30 to 100
Warm-up tool	per month, often bundled	0 to 50
Email verification	per 1,000 addresses	3 to 10
Contact data and enrichment	per verified contact	0.05 to 0.30
Verified Mark Certificate for BIMI	per year	1,000 and up

A four-domain, twelve-mailbox setup with a platform, verification and a modest data budget lands around 200 to 400 EUR a month in tooling. The expensive part of outbound has never been the tools.

What goes wrong

• Two SPF records on one domain. Both become invalid. Authentication fails everywhere and the symptom looks like bad copy.
• DMARC published straight at reject before the reports are read, which silently kills legitimate mail from a tool nobody remembered.
• A sending domain with no website behind it. The prospect checks, finds a parked page, and files you under fraud.
• All mailboxes created on the same day on the same domain and pushed to 40 sends each in week one.
• Sending cold from the main domain “just for a small test”. The small test is how the domain gets burned.

Checklist

• Three to five sending domains registered, each redirecting to the real site.
• Two or three mailboxes per domain, human names, photos set.
• One SPF record per domain, under ten DNS lookups, verified with a checker.
• DKIM signing confirmed on a test send.
• DMARC published at p=none, reporting address monitored.
• One-click unsubscribe header enabled in the sending platform.
• Custom tracking domain configured, or open tracking switched off.
• BIMI on the brand domain only.

Chapter 3Warm-up

Why

A brand-new domain that sends 200 messages on its first day looks exactly like a spam operation, because that is what spam operations do. Providers have no history to judge you by, so they judge the pattern. Warm-up builds the history: small volumes, real conversations, replies going back and forth, gradually more of it.

Warm-up is also the only part of this that cannot be rushed with money. You can buy better data tomorrow. You cannot buy three weeks of sending history.

The how

A ramp that works, per mailbox. Treat the week numbers as a rule of thumb and slow down if the numbers in Chapter 10 move against you.

Period	Volume per mailbox per day	What you are doing
Domain age, weeks 0 to 2	0	Domain registered, DNS published, nothing sent
Week 1	10 to 20	Warm-up pool traffic only, replies enabled
Week 2	20 to 30	Warm-up pool, start a handful of real replies
Week 3	30 to 40	Blend in the first real campaign sends, small batches
Week 4 onward	20 to 50 real sends plus background warm-up	Full campaign, warm-up never switched off

Three details that matter more than the exact numbers.

Let the domain age. Register two to four weeks before the first send. Providers weigh domain age.

Keep warm-up running underneath the campaign. Most people switch it off the day real sending starts, which drops the reply-to-send ratio at the exact moment it matters.

Automated warm-up pools are a supplement. They exchange mail between accounts on a network, and providers increasingly recognise those patterns. Get real human replies in early: send your first batch to twenty friendly contacts who will actually answer.

When to stop and go back. If bounces climb past 3%, or spam complaints cross 0.1%, or reply rate halves against the previous four weeks with no change in list or copy, cut volume to a third and hold there for a week. Re-verify the list before you climb again. Recovery is measured in weeks, so the cheap move is to slow down early.

What goes wrong

• Warm-up for five days because the client is impatient, then 150 sends a day in week two. The whole setup is burned by week three and nobody knows why, because the copy looks fine.
• Warm-up switched off on launch day.
• Warm-up run on a domain with no DKIM, so the whole period builds reputation for mail that fails authentication.

Checklist

• Every mailbox has at least fourteen days of warm-up behind it.
• Authentication was in place before the first warm-up message.
• The ramp is written down and the sending platform enforces the daily cap.
• Warm-up remains on after launch.
• You have a stop rule and you know what number triggers it.

Chapter 4The ICP

Why

Most bad campaigns are correctly written letters sent to the wrong people. The prospect reads it, understands it, and has no reason to care. Uniform indifference across a thousand sends is an ICP failure, and no amount of subject-line testing fixes it.

The sentence forces the question that matters, which is the buying trigger. Once you have it, the list builds itself from evidence of that trigger.

The how

Start from five current customers. Not your best five, your most typical five. Pull their size, sector, country, what they were doing when they bought, and who signed. Then find what is the same across them. This is the first thing we ask a new client for: give us five companies you already serve and we will find the ones that look like them.

Describe a company, not a person. The ICP is firmographic: employee band, revenue band, country, sector, plus situational criteria like owning a fleet, holding a certification, or running a particular software platform. The person comes second, as a job title inside that company.

Write the exclusions down. They do more work than the inclusions. “Manufacturers, 50 to 500 staff, Germany and Austria, excluding automotive tier-1 and staffing firms.” Licence registers and directories are full of staffing firms filed as if they were the thing they supply, and a staffing client who sees three competitors on your shortlist stops reading.

Classify before you pick the signal. This rule costs the most when it breaks. Job adverts are a signal for companies that sell people: staffing, labour hire, recruitment, executive search. A manufacturer advertising for a machinist wants an employee, not a subcontractor, so a vacancy list is the wrong list for a contract manufacturer. One rejected a deck built on “635 DACH manufacturers are hiring machinists”, and the same mistake had already cost two other campaigns before we wrote it down.

For everyone else, build the signal from the buyer’s own situation:

• An IT or e-commerce consultancy: the platform the target runs today. A shop still on an end-of-life version, an old ERP, a legacy CRM.
• A producer or exporter: directory category, buyer-side sourcing patterns, part fits, certification requirements.
• A funding or grant consultancy: an open programme with a deadline, and companies eligible for it.
• Anyone selling into events-driven sectors: exhibitor lists from trade fairs, used before and after the fair.

Narrow with real filters. A register pull for one sector in one country can start at tens of thousands of companies. Four filters take it down fast. One worked example from Polish register data: all active registered companies, 734,565. Transport and logistics only, 39,156. With a filed revenue figure, 18,351. Turning over more than two million zloty, 3,666. Four filters, 734,000 down to 3,666, and the list is now worth writing to.

Read the websites. Registry activity codes lie. A Lithuanian accounting-services code once returned a meat producer, because the company files accounts. We run automated readers across every website on the list and keep only the companies whose own site says they do the thing. On one client run that cut 15,000 register matches down to 3,500 real fits, and the 3,500 went first.

Test with 200. Send to 200 matching companies and read the replies. Uniform irrelevance means the ICP is wrong. Polite brush-offs from right-fit companies mean the ICP is right and the message is wrong. Interested replies from outside the profile mean it is too narrow. Revisit quarterly, and only when something real changed.

What goes wrong

• The ICP is “B2B companies in Europe”. That is a market, not a profile.
• The signal proves the target wants to hire when the client does not sell people.
• The list is ranked by posting volume, so generalists float to the top and the niche brief is contradicted on page one.
• Nobody writes exclusions, so the client’s own competitors end up on the list.

Checklist

• The one-sentence ICP is written and the buying trigger is in it.
• Five current customers have been profiled.
• Exclusions are written down, including competitor types.
• The signal proves the buying trigger, not just that the company exists.
• Every company on the list has been checked against its own website.
• A 200-company test batch is planned before the full send.

Chapter 5Building the list

Why

Two reasons, and both of them bite. The first is deliverability: bad addresses bounce, bounces destroy domain reputation, and a bought list with an unknown vintage is a bounce generator. The second is legal: under GDPR you need to be able to say where a contact came from, and “we bought it” is not an answer that survives a complaint.

A third reason: the list is the product. A well-targeted 500 beats a sloppy 5,000. Send to a candle workshop and a car repair garage and no message saves you.

The how

Start with open registers. They give you the company universe for free or close to it.

• Germany: Handelsregister for companies, Bundesanzeiger for filings.
• Poland: KRS for registered companies, CEIDG for sole traders.
• Lithuania: the business register, plus commercial aggregators that resell it with far more fields. One of ours costs a fraction of the alternative and carries over 130 fields per company, including turnover, headcount changes and court records over ten years.
• Australia: ASIC and the ABN register, free, and the source behind 14,000 Australian recruitment and HR firms we pulled for one campaign.
• UK: Companies House, free, with a bulk download.
• Elsewhere in the Nordics, Benelux and France: national equivalents, each with its own openness and cost.

Add the signal layer. Job boards for staffing clients. Tender portals for anyone selling to the public sector. Exhibitor lists for trade-fair sectors. Technology fingerprints for software and e-commerce work. Funding programme eligibility lists for grant consultancies.

Two warnings on job boards. StepStone’s search pages embed the job list as JSON, which is usable. The German Bundesagentur returns 403 to automated requests. Role slugs on job boards are query labels, not classifications, so a slug called “engineer” contains everything from a site foreman to a DevOps hire. Filter on the actual job title, never on the slug.

Find the decision maker. Company data is the easy half. In Lithuania we find a named decision maker with a working address for roughly 70% to 80% of target companies, and that number falls in larger markets. Use LinkedIn to match a person to the role, then an enrichment provider for the address.

Enrich and verify, in that order. Enrichment tools such as Prospeo and Clay find and pattern-match addresses. Verification proves they exist: syntax check, MX record check, an SMTP handshake asking the receiving server whether the mailbox exists without sending anything, then risk flags for role accounts, disposable domains and catch-alls.

Then verify again through a different route, because an address that passed a static check three months ago may belong to someone who has left. Our bounce rate sits under 3% by policy and usually lands between 1% and 2%. Market tolerance stretches to 5%. Aim well below it.

Catch-all domains. A catch-all accepts every address, so verification cannot confirm anything. Send to them in small monitored batches, and if one address on a catch-all domain hard bounces, suppress the entire domain.

List decay. Re-verify anything older than three months. A meaningful share of any B2B list goes stale within a year as people change jobs. Verify at the point of sending, not at the point of buying.

DNC and suppression. Maintain a do-not-contact list from day one: every “remove me”, every unsubscribe, every client-forbidden account, every competitor. It is permanent and it survives platform changes. A prospect who asks to be removed and hears from you six months later because you changed tools has a complaint with a paper trail.

What goes wrong

• A bought list of 100,000 addresses with no provenance, mostly info@ and biuro@, which no administrator forwards.
• Role slugs used as classifications, so a staffing client gets a list of firms that do not hire the role at all.
• Licence and trade registers not filtered for staffing firms, so the client sees three of their own competitors on the shortlist.
• No source and date column, so when a complaint arrives nobody can prove where the contact came from.

Checklist

• Every contact carries a source name and a collection date.
• The list has passed a website-content check, not just a registry code filter.
• Every address has been verified, and the pass is under three months old.
• Role accounts and disposable domains are stripped.
• Catch-all domains are flagged and batched separately.
• The suppression list has been applied.
• Competitors and existing customers are excluded.

Chapter 6The message

Why

The reader gives you about three seconds. An A4 block about your company history loses in the first one. Nobody cares how many years you have been in business. They care what this costs them or earns them, and whether you understood their situation before you wrote.

We rewrote a client’s own message once and the difference was not style. Their version opened with “we have been operating since 2009”. Ours opened with a question about the machine they run. Same company, same offer, completely different reply rate.

The how

The six-line shape. This is the exact structure behind the strongest staffing campaign we hold, a Polish blue-collar sequence that pulled 7.1% replies.

1. Greeting with the first name, correctly inflected for the language.
2. One question, or one specific observation about their company.
3. One sentence on how they handle this today.
4. One sentence on what you do.
5. One proof: a number, a named customer, a certification, a guarantee.
6. One question as the ask, then a sign-off.

Step one runs about 90 words. Later steps run 70 or fewer. Cap every cold email at 99 words. Published benchmark work points the same way: six to eight sentences, under 200 words.

Bans that earn their place.

• No colon followed by an explanation. It reads as a slide, not a letter.
• No lists of three. They read as marketing rhythm.
• No explainer tails. Say the thing and stop.
• Positive words only. “Avoid losing” becomes “keep”.
• No ALL CAPS, no exclamation marks, no “free”, “urgent”, or “exclusive”.
• No links in email one. Links in cold mail cost you deliverability. Say so in the postscript and make replying the only action.
• No tracking pixel and no HTML on cold. Plain text only. Save the branded HTML for people who know you.

Spam-word pass. Before a sequence goes live, sweep the body for the words that trip filters and scoring tools. The ones that have cost us rewrites: marketing, win, money, off, get, “for you”, all, leave, now, catch, selling, crack, here, collect, phone. Rewrite the term, keep the meaning. Leave greetings and sign-offs the founder wrote himself exactly as they are.

Native language. A German reads German. A Pole reads Polish. Lithuanian buyers answer direct, practical Lithuanian and ignore the polished marketing kind, with the formal “Jūs” on first contact. In Australia and New Zealand use first names from line one, stay blunt, and do not boast. Both markets punish a seller who showcases their own success, so ask for the twenty minutes and offer the list rather than promising the outcome.

Spintax on greetings. Rotate three variants of every block so a thousand identical messages do not hit the same filter with the same fingerprint. Inflect names properly: in Lithuanian it is “Labą dieną, Rolandai”, not “Labą dieną, Rolandas”. We check a thousand-plus inflections per campaign with an automated pass, because getting it wrong marks the letter as bulk in the first two words.

Subject lines. Lowercase, under 50 characters, no hype, no spam words. Questions outperform statements. Company name in the subject lifts opens on cold. Write six per campaign and rotate.

Working examples:

• your {{role}} search
• {{company}} + faster delivery?
• one less compliance headache
• a question about your current supplier
• still looking for {{job_title}}?
• right person?

The sequence. Four to five touches over three to four weeks, with widening gaps.

Step	Day	Purpose
1	0	The angle, the signal, the ask
2	+4 or +5	Short bump on the same thread, “coming back to my previous message”
3	+9 or +10	A different angle, new value not mentioned before
4	+16 or +17	Low-effort ask, or a local hook
5	+25 to +30	Break-up, closing the loop

A large share of replies arrives at steps two and three, because the first one landed while the reader was driving or at a fair. We have seen 20% to 30% of conversions come after the first or second follow-up. The break-up works because a door closing pulls an answer out of people who meant to reply and forgot. If the whole sequence draws nothing, wait 45 days and restart with a different angle.

Angles. Write at least six per campaign, each aimed at a different buying motive, and match the angle to the job title. Speed goes to purchasing and production managers, cost to commercial directors and owners, documentation to project engineers. Six versions of the same idea in different words is one angle, not six.

Five example sequences

All adapted from real campaigns. Client names withheld.

Example 1. Staffing agency in Germany, hiring signal

Step 2, day 5, two sentences: “Ich komme auf meine letzte Nachricht zurück. Falls die Stelle noch offen ist, schicke ich Ihnen die Profile heute.” Step 3, day 10, adds the trial period or the replacement guarantee, which the opener did not mention. Step 4, day 17, closes the loop.

Example 2. Industrial exporter into the Nordics, speed angle

This is the angle that produced our Nordic campaign’s strongest segment: 718 precast manufacturers reached, 20 replies at 2.8%, 7 of them positive. Step 2: bump. Step 3: offer a sample Declaration of Performance, which is the documentation angle and a free, low-commitment thing to say yes to. Step 4: break-up.

Example 3. IT consultancy into DACH, platform signal, no hiring data

The point of this one: the signal is what the buyer runs, not who the buyer hires. A consultancy gets nothing from a list of companies advertising jobs.

Example 4. Trade fair, before and after

Run it twice. Three weeks before the fair, write to the exhibitor list and ask for a slot at the stand. Two weeks after, write to everyone you did not meet. The subject carries the fair name and the company name, which is why it gets opened.

Example 5. Break-up and right-person, any market

This one doubles as a short opener for someone who went quiet.

What goes wrong

• The email is about the seller. Years in business, team size, mission.
• The offer is vague, so the reader has nothing concrete to say yes or no to.
• A heavy CTA. “Book a 30-minute demo” from a stranger is a large ask. “Should I send the list?” is a small one.
• Fake personalisation. “I loved your recent post” when there is no recent post. One real researched line beats five generated ones.
• Follow-ups that repeat the opener with “just bumping this” and add nothing.

Checklist

• Every email is under 99 words.
• Six lines, one signal, one ask.
• No colon-plus-explanation, no lists of three, no negative framing.
• Written in the reader’s language, with names correctly inflected.
• Three spintax variants per block, six subject lines per campaign.
• Spam-word pass completed.
• No links and no tracking in email one.
• Angles mapped to job titles in a matrix, so purchasing does not receive the compliance letter.

Chapter 7Sending

Why

The bottleneck in outbound is answering capacity, not sending capacity. We have paused client campaigns because five interested replies a day were arriving and nobody was reading them.

The how

Start small and climb. Launch at 100 sends a day. If replies arrive faster than anyone can answer, hold. If the volume is comfortable, go to 200. Adjust weekly, not daily.

Respect the caps. 20 to 50 per mailbox per day, run at about 70% of that. A 200-a-day campaign needs seven to ten mailboxes across three or four domains.

Timing. Tuesday to Thursday mornings in the recipient’s timezone. Australians start around 08:00, so a European sending at a European hour arrives at the end of their day. Set the window per country. Avoid the last week of December, the first two weeks of August in Germany and France, and the Lithuanian and Polish holiday blocks.

Gaps between steps. Four to five days to step two, then widen. Tight gaps read as pressure and collect complaints.

A/B testing. Change one variable. Test in this order, strongest first: the offer, then the audience segment, then the subject line, then the opening line, then the call to action. Send timing and message length come last, because their effects are real and small.

Volume matters more than most people think. Several hundred emails per variant before you call a result. A forty-email test is a coin flip. One of our campaigns split 2,000 prospects evenly and the observation-led opener pulled 2.1% against 1.2%, which is 21 replies against 12. At that volume the gap is worth acting on. At 200 sends it would have been noise.

Do not call a test early. Replies keep arriving for a week or more after the last send goes out.

The spam-word pass, again. Run it on every variant, not just the first one.

Multichannel. Email plus LinkedIn lifts response in markets where LinkedIn penetration is high. In Germany it also sits in a lower-risk legal position than unsolicited email, which Chapter 9 covers. Keep the message consistent across both.

What goes wrong

• Volume raised before the reply-handling capacity is there.
• All the volume pushed through two mailboxes because the platform rotation was misconfigured.
• Three changes made at once, so nobody knows which one moved the number.
• Sending on a public holiday in the target country and reading the flat week as a copy failure.

Checklist

• Daily send volume matches reply-handling capacity.
• Mailbox rotation verified in the platform, no mailbox over its cap.
• Send window set per target country.
• One variable per test, several hundred sends per variant.
• Results read at least one week after the last send.

Chapter 8Replies and meetings

Why

Speed is the thing you control that moves the number most, after targeting. A prospect who replies at 10:00 is at their desk with the tab open. By 10:00 the next day they have googled the category and found a competitor who answered faster. We watched that happen to a client in real time, with interested replies sitting unopened for two days.

The how

One shared inbox. Every reply from every mailbox lands in one place. If two people can see it and neither owns it, nobody answers. Assign an owner per campaign.

Inside the hour during working hours. Set that as the standard and measure it. A short answer now beats a considered answer tomorrow.

Sort replies into four buckets.

• Interested. Answer with one line of acknowledgement and one calendar link. Do not send a brochure. Do not send pricing. Book the call.
• “Send me more info.” This is a soft no most of the time, and occasionally a real request. Answer both at once: send one concrete artefact, a sample list, a specification, a one-page case, and attach a booking link with a reason to use it. “Here is the sample. If it looks right, fifteen minutes on Thursday and I will walk you through how we built it.” Never send a deck and wait.
• Not now. Ask when, put a date in the CRM, and remove them from the sequence. A dated follow-up in four months converts better than three more emails this month.
• Remove me. Suppress permanently and confirm in one line. No argument, no last pitch.

Pricing in a written reply. Explain the shape and hold the figures for the call. A setup fee covers the build, a monthly retainer covers the running. Numbers in writing price the call away before you have shown anything.

Handing to sales. The handover needs four things: the original message that was sent, the reply verbatim, the signal the contact was selected on, and the company’s basic data. A salesperson walking into a call without knowing which angle triggered the reply will re-pitch the wrong thing.

CRM. Every reply becomes a record the same day. Log the source campaign, the angle, the date, the outcome. Without that you cannot answer the only question that matters at renewal: which angle produced revenue.

Automation limits. AI drafts work when the answer space is small and the product is simple. For anything needing a real conversation, keep the agent out of the middle: agents wander off the brief, and that is expensive in front of a serious buyer. Five replies a day is easy by hand.

What goes wrong

• The shared inbox nobody owns.
• Sending a 20-page deck to “send me more info” and hearing nothing back.
• Pricing in writing, then the meeting never happens.
• The break-up email going to someone who already replied, because the reply did not stop the sequence.
• No CRM record, so three months later nobody can say which campaign produced the customer.

Checklist

• One shared inbox, one named owner per campaign.
• A one-hour reply target during working hours.
• A calendar link ready and tested on mobile.
• A standard answer written for “send me more info” before the campaign launches.
• Replies stop the sequence automatically.
• Every reply logged in the CRM the same day.

Chapter 9Compliance in the EU

This chapter describes how we run it. It is general information rather than legal advice, and a German campaign in particular deserves a lawyer’s read before the first send.

Why

The risk in EU outbound is not usually a regulator. It is a competitor or an annoyed recipient with a lawyer, and in Germany that mechanism is fast and cheap for them to use.

The how

GDPR. Article 6(1)(f) allows processing for a legitimate interest, balanced against the data subject’s rights. Recital 47 names direct marketing as a possible legitimate interest. For B2B that balance usually holds when four things are true: the contact is approached in a professional capacity, the message is relevant to their job, the data came from a public or business source, and the impact on them is minimal. Document that reasoning once, in writing, before you start. A legitimate interest assessment that exists on paper is the difference between a defensible position and a guess.

ePrivacy. The ePrivacy Directive sits on top of GDPR for electronic marketing, and every member state transposed it differently. That is why there is no single EU answer.

Germany. The Act against Unfair Competition, UWG, Section 7, treats advertising by email without prior express consent as unreasonable harassment, and it does not carve out B2B the way it carves out telephone advertising. Legitimate interest under GDPR does not override it: German supervisory authority guidance takes the view that if a message breaches Section 7, the Article 6(1)(f) balancing test fails too.

The narrow exception inside Section 7 is the existing-customer rule: an address collected during a sale may be used to advertise similar goods and services, provided the customer has not objected and you tell them at collection and in every message that they can object at no cost.

Enforcement comes as an Abmahnung, a cease-and-desist letter from a recipient, a competitor or a competition association. The exposure is legal costs and an injunction rather than a headline fine, and signing the undertaking creates a contractual penalty for any repeat. The business benefiting from the advertising stays answerable alongside the agency that sent it, so hiring an agency does not move the risk.

The practical consequence for Germany: email carries the highest exposure, telephone is lower for business contacts where interest can be presumed on concrete grounds, LinkedIn and business networks are lower again, and postal mail is least restricted. Many German campaigns we would run as LinkedIn-first with email second.

Poland and Lithuania. Both follow the mainstream B2B legitimate-interest line, and both transposed ePrivacy into national electronic communications law with direct-marketing provisions that require care when the address belongs to a named individual rather than a company. The safe operating rule in both: contact people in a business role at a business address, identify yourself fully, keep the message relevant to their work, offer an opt-out in every message, and honour it permanently.

Nordics and Benelux. Same base rule, with national variation in how strictly the marketing provisions are read. The operating discipline above travels.

Australia. Outside GDPR. The Spam Act 2003 governs commercial electronic messages: you need consent, which can be inferred where the address is published on a business website without a statement against unsolicited messages, and every message must identify the sender and carry a functional unsubscribe.

Unsubscribe wording. One plain line at the bottom, no legal block. “If you would rather not hear from me, reply with stop and I will remove you.” Add the List-Unsubscribe header in the sending platform so the one-click unsubscribe required by Google and Yahoo since February 2024 is present. Honour every request permanently, across every tool you ever move to.

Data retention. Keep a source and a collection date on every contact. Re-verify anything older than three months before reuse. Delete enriched contact data you have not used within twelve months. Keep the suppression list forever, because that one exists to protect people who asked to be left alone.

What goes wrong

• A German campaign run exactly like a Lithuanian one, and an Abmahnung two weeks later.
• No legitimate interest assessment written down, so the defence is assembled after the complaint.
• An unsubscribe that goes to an unmonitored mailbox.
• A suppression list left behind when the sending platform changes.

Checklist

• Legitimate interest assessment written and dated.
• Source and date recorded per contact.
• Germany handled separately, with counsel consulted and channels ranked by risk.
• Full sender identity in every message: legal name, address, contact.
• Plain-language opt-out plus List-Unsubscribe header.
• Suppression list permanent, tool-independent, applied before every send.
• Retention rule written and actually enforced.

Chapter 10Measuring and fixing

Why

Without a baseline, every bad week looks like a copy problem, and people rewrite the email when the bounce rate doubled. The numbers tell you which layer broke, and each layer has a different fix.

The how

The weekly dashboard. Six lines. Fifteen minutes.

Metric	Read it as	Healthy	Act
Sent	Volume against plan	Within 10% of plan	Check mailbox rotation
Delivered	Sent minus bounces	97% or better	Below 97%, re-verify
Bounces	List quality	Under 2%	Over 3%, pause and re-verify
Replies	Targeting plus copy	1% to 5% by market	Under 1% sustained, rebuild list
Positive replies	Message fit	25% to 50% of replies	Under 15%, change the angle
Meetings	Your responsiveness	33% to 66% of positives	Under 25%, fix reply handling

Add two watchdogs that sit outside the funnel:

• Spam complaint rate. Over 0.1% is a warning. 0.3% is Google’s published limit for bulk senders and the point at which real damage starts.
• Blacklist checks. Any listing gets investigated the same day.

Thresholds that mean stop.

• Bounces climbing week on week, for example 1% then 2.5% then 4%: stop the campaign, re-verify the entire list, drop volume to a third, resume with the most engaged segment first. Recovery takes weeks.
• Spam complaints approaching 0.3%: stop. This is the one that ends domains.
• Reply rate halved against the four-week baseline with no change to list or copy: a placement problem, not a copywriting problem. Check authentication, check whether a mailbox went over cap, run an inbox placement test.
• Zero replies after 1,000 well-targeted sends: either you are in spam or the ICP is wrong. Check placement first, because it is cheaper to test.

Common fixes, in the order to try them.

1. Authentication. Confirm SPF, DKIM and DMARC still pass on a live test send. Somebody else’s DNS change breaks this surprisingly often.
2. Volume. Cut to a third for a week.
3. List. Re-verify, strip catch-alls and role accounts, drop anything older than three months.
4. Offer. Change what you are asking them to say yes to. This moves the number more than any subject line.
5. Segment. Run the same message to a narrower slice.
6. Copy. Last, one variable at a time.

Monthly, spend thirty minutes on an inbox placement test across Gmail, Outlook and one corporate domain, a blacklist check on every sending domain, and an audit of volume per mailbox against the warm-up plan.

What goes wrong

• Reading open rate as a health metric. Apple Mail Privacy Protection inflates it and a tracking pixel costs you deliverability to collect it.
• Comparing this week to last week instead of to a four-week baseline, so normal variance reads as a trend.
• Rewriting copy when the fault is authentication.
• Nobody owns the dashboard, so it gets read in month three when somebody asks about results.

Checklist

• The six metrics are recorded weekly in one place.
• A four-week rolling baseline exists.
• Google Postmaster Tools is connected for every sending domain.
• Stop thresholds are written down before launch, not decided during a crisis.
• One named person reads the dashboard every week.

Chapter 11The 30-day launch plan

Why

Every attempt to compress this fails in the same place. The domain is too young, the mailboxes are too fresh, and the campaign burns before the copy gets a fair test. A month of preparation costs a month. A burned domain costs the domain and the month.

The plan

Day	Do this	Done looks like
1	Write the one-sentence ICP, pick the signal, profile five current customers	“We sell ___ to ___, who buy when ___” on paper
2	Register 3 to 5 domains, publish SPF, DKIM and DMARC at p=none	Domains live with 301 redirects, checker passing
3	Create 2 to 3 mailboxes per domain, connect the sending platform	Human names and photos, rotation tested
4	Start warm-up, write the exclusion list	10 to 15 a day per mailbox, replies enabled
5 to 7	Pull the raw list from open registers and the signal source	3,000 to 10,000 raw companies
8 to 9	Qualify against company websites	Raw list cut by half or more
10	Find decision makers and enrich addresses	Named person plus address per company
11	Verify every address, strip role accounts, apply suppression	Bounce risk under 2%, source and date columns present
12 to 14	Write six angles, six subjects, three spintax variants per block	Every email under 99 words
15	Spam-word pass, warm-up to 20 to 25 a day	Zero flagged terms, ramp holding
16	Legitimate interest assessment, shared inbox, owner, calendar link	Dated document on file, link tested on a phone
17 to 18	Build the sequence, write the “send me more info” answer	Reply stops sequence, artefact ready
19 to 21	Warm-up to 30 a day, seed 20 real replies from friendly contacts	Conversation history in each mailbox
22	Send the 200-company test batch	First real sends live
23 to 25	Read the test replies, correct one thing	ICP confirmed or message fixed
26	Open the full campaign at 100 sends a day	Volume matched to reply capacity
27 to 28	First dashboard read, step two lands	Six metrics recorded, baseline started
29	Raise to 150 to 200 a day if replies are handled inside the hour	Mailboxes still under cap
30	Review bounces, complaints, replies, positives, meetings	Decision made for month two

Month two costs less and performs better. The infrastructure is built, the list exists, the message has been corrected once, and every mailbox is fully warm. Judge outbound on month two and three, not on month one.

What goes wrong

• Day 22 pulled forward to day 8 because a deadline moved.
• The list built before the ICP sentence is written, so the filters are guesses.
• Reply handling arranged in week four, after the replies started arriving.
• Six angles written but only one used, so nothing is learned about what the market responds to.

Checklist

• Nothing real sends before day 22.
• Every gate above has a date and a name against it.
• The test batch of 200 happens before the full list.
• Month two is budgeted before month one launches.

Chapter 12If you would rather we did this for you

Everything in this book is what we do for clients, and plenty of people run it themselves. What we bring is the part that takes longest to build: register access across the EU, Baltics, Nordics and Australia, decision-maker enrichment, double verification before anything sends, infrastructure already warm, native-language copy, and somebody reading the numbers every week so a bounce spike gets caught the day it happens.

We handle the machine. You handle the conversations that come out of it.

How we charge. The first month costs EUR 3750 and covers the build: domains, mailboxes, authentication, warm-up, the ICP work, the list, and the copy, with the first sends by day 14. From month two the fee is EUR 2850 a month and covers the running: sending, monitoring, reply routing, and the weekly numbers. You can cancel any month with 28 days’ notice. The full breakdown is on the pricing page.

We do not promise a fixed number of meetings. Nobody honest can. We tell you what the range looks like in your market before you sign, and we show you the real numbers every week once we start.

The free target list. Send us two things: your ideal customer in one sentence, and five companies you already work with. We will build you a shortlist of companies in your market that we could contact now, with the signal we would use on each. No charge, no call needed to receive it. If the list looks right, we will talk about running it.

Where to go from here

Run the 30-day plan in Chapter 11 yourself. Everything you need is in these pages, and plenty of companies do exactly that. If you would rather skip the four to six weeks of infrastructure work, we run the machine and hand you the conversations that come out of it.

• Start with a free target list. Send us your ideal customer in one sentence and five companies you already work with. We build a shortlist of companies in your market that we could contact now, with the signal we would use on each. No charge, and you do not need a call to receive it. Request the free target list.
• Or book a free pipeline audit. We walk through your market and the range a campaign in it realistically produces. Book a free pipeline audit.
• Or reply to a human. Questions about anything in this book, including the parts where your market works differently, reach a person who runs these campaigns every week: info@ripeleads.eu.

Get the PDF

The same thirteen chapters as a 44-page PDF, laid out for printing and sharing with whoever runs your outbound. One field. The PDF arrives in your inbox, and the link on the confirmation opens it straight away.