Every click, from nothing to a running campaign

Buy the sending domain

A domain bought for this purpose, kept separate from the one your invoices go out on. If it ever burns, you throw it away rather than repair it.

1. Pick a name close to your real one. If you are brightwave.co.uk, buy brightwavehq.co.uk, getbrightwave.co.uk or brightwave-group.com. The reader should recognise it as you.
   • Avoid numbers, and avoid more than one hyphen: both read as disposable
   • .com and .co.uk are equally fine. Avoid the cheap new endings
   • Do not buy a domain that has been registered before. Check its history first, because you inherit whatever the last owner did with it
2. Buy it from any registrar. It does not matter which. You will move the nameservers in part 03 anyway.
3. Put a real page on it. One page is enough: who you are, what you do, a contact address.
   • A redirect to your main site is not enough. A domain that only redirects looks exactly like a throwaway, because usually it is
4. Point the nameservers at Cloudflare. Add the domain in Cloudflare, take the two nameservers it gives you, and set them at the registrar. Propagation is usually minutes.

Create the mailboxes

Three of them, not one. Volume comes from more mailboxes sending a little, never from one mailbox sending a lot.

1. Sign up for Google Workspace using the new domain. Choose the entry business tier. The cheapest one is enough; you are not using Drive or Meet.
2. Create three users. Our own pattern, which works:
   • yourname@: a real person, and the one most replies come to
   • outreach@
   • team@
   Each user needs its own licence. Aliases will not do: an alias is not a separate sending identity and the platform cannot rotate across them.
3. Set a real name and signature on each. A mailbox with no display name is a flag.
4. Log into each one once through the web interface. A mailbox that has never been opened has no activity history at all.
5. Turn on two-factor authentication, then generate an app password for each mailbox if your sending platform asks for one rather than using sign-in.
   • App passwords will not appear as an option until two-factor is switched on

Publish the DNS records

Four kinds of record. All of them go in Cloudflare under the domain → DNS → Records → Add record.

1. MX, so the domain can receive mail

Google Workspace gives you these during setup. Modern setups use a single record:

Type	Name	Content	Priority
MX	@	smtp.google.com	1

Delete any MX records the registrar added by default. Leaving a stray one in place sends a share of your mail somewhere you are not reading.

2. SPF, which servers may send as you

Type	Name	Content
TXT	@	v=spf1 include:_spf.google.com ~all

• There is only ever one SPF record on a domain. If you have two, neither works. Merge them into one line rather than adding a second.

3. DKIM, the signature

1. Go to Admin → Apps → Google Workspace → Gmail → Authenticate email
2. Select the domain, choose 2048-bit key length, and click Generate new record
3. Copy the value it produces. It is long, so take all of it
4. In Cloudflare add a TXT record with name google._domainkey and that value as the content
5. Go back to the Admin console and click Start authentication. It will not confirm until the record has propagated. Wait a few minutes and try again rather than assuming it failed

4. DMARC, what to do when the others fail

Type	Name	Content
TXT	_dmarc	v=DMARC1; p=none; rua=mailto:you@yourdomain.co.uk

Start at p=none. That is monitoring only: it changes nothing about delivery and asks receivers to send you reports. After two weeks of clean reports, change p=none to p=quarantine. Do not start at quarantine; if something is misconfigured you will not find out until your mail has already been disappearing for a fortnight.

Verify the records actually work

Publishing a record and having it work are different things. This is the step almost everybody skips, and it is where the expensive mistakes are found.

Check each one directly

On a Mac or Linux machine, in Terminal. On Windows, use nslookup -type=TXT or any online DNS lookup tool.

dig +short TXT yourdomain.co.uk
dig +short TXT google._domainkey.yourdomain.co.uk
dig +short TXT _dmarc.yourdomain.co.uk
dig +short MX  yourdomain.co.uk

Each should return something. An empty answer means the record is not published, whatever the dashboard says.

Then the check nobody runs

Take every hostname named in your SPF record and look it up on its own. If your record says include:spf.theirplatform.com, then run:

dig +short TXT spf.theirplatform.com

Finally, a real message

1. Send an email from each of the three mailboxes to a personal address on a different provider
2. Open it and view the original or the message source
3. Look for spf=pass, dkim=pass and dmarc=pass
4. All three must say pass. A permerror or none on any line sends you back to part 03

Connect the sending platform

Three mailboxes in, warm-up on, and nothing sent to anybody real for four weeks.

1. Create the account and go to Email Accounts → Add new
2. Add each of the three mailboxes. Google Workspace accounts usually connect by signing in; if it asks for a password instead, use the app password from part 02, not the account password
3. Set the sending name on each to a real person's name. Not "Sales Team"
4. Turn warm-up on for all three. Leave every warm-up setting at its default. The defaults are sensible and the ramp is the point
5. Set the daily campaign limit per mailbox to 20. Not the maximum it offers. You will raise this later, slowly, in part 10
   • Usually at Email Accounts → the account → Settings → Campaign Settings
6. Run the platform's own domain test if it has one. Most have a "test domain setup" button on the accounts screen. It is a second opinion on part 04, from the people who watch deliverability all day

Fix the settings that are wrong by default

This is the shortest part and the one that has cost us the most. Every item below is a default we did not check, and each one cost real time before we found it.

Setting	Do this
Open and link tracking	TURN IT ON. It can be off by default. If it is, your open rate is not low. It does not exist, and every number you look at for weeks is meaningless. This caught us twice on two different campaigns.
Automatically pause on high bounce rate	Usually at Settings → Preferences → Outreach Preferences. Leave it on, but know it exists and know where it lives. When it fires, a stopped campaign looks exactly like a slow one. See part 10.
Unsubscribe link header	Turn it on. It adds the opt-out into the message headers where mail clients can offer it natively. Good for deliverability, and it helps satisfy the opt-out obligation that applies to every message.
Provider matching	Turn it on if your mailboxes are Google. It sends Google-to-Google where it can, which measurably helps.
Automatic subject line optimisation	Leave it off for now. It needs hundreds of sends per variant before it means anything, and switched on early it will pick a winner from noise. Revisit once you have volume, and set it to judge on replies rather than opens.
Stop sequence on reply	On. Nothing loses a warm reply faster than the follow-up arriving anyway.

Wait four weeks

Nothing to do. That is the hard part, and skipping it undoes parts 01 to 06.

Warm-up is building a sending history so the domain is not a stranger when your first real email arrives. There is no way to compress it and no setting that substitutes for it.

Check once a week, for two minutes

• Health score on each mailbox. It should climb and settle high. If one lags well behind the others, disconnect and reconnect that one rather than waiting it out.
• Warm-up emails actually sending. A mailbox that has quietly disconnected shows a flat count. Reconnect it and restart its four weeks.
• Nothing else. No campaign mail. Not one test to a prospect "just to see".

Use the time

Run your Winnowed count, build the list, and write the three emails from part 09. By the time the four weeks are up you want to be pressing go, not starting to think about copy.

Import your Winnowed list

Suppression check first, then dedupe, then upload. In that order, because the first two cannot be undone after the third.

1. Open the file Winnowed produced. It contains kept rows, refused rows with the reason, and withheld rows with a phone number.
2. Split it in two. Rows with an email address go to the campaign. Rows with a phone and no address go to a separate sheet. That is your call list, and part 05 of the six-step guide covers working it.
3. Check against your suppression list. Anyone who has ever opted out, or hard bounced, comes out now. If this is your first campaign you have no suppression list yet. Create the empty sheet anyway, because you will need it within a fortnight.
4. Dedupe against every previous campaign. Contacting the same company twice from two campaigns is the fastest way to look exactly like what you are trying not to look like.
5. Upload. Map the columns when it asks: company, email, website, and anything else you want available in the copy.
6. Read the import summary. It will report rows rejected as duplicates or invalid. Look at what it rejected rather than clicking through. That summary is where data problems surface.

Build the sequence

Three emails at day 1, day 3 and day 10. Subject line variants on the first one only.

1. Create the campaign and open Sequences
2. Write step 1. Under 120 words, one concrete ask, no attachments, no images, and ideally no links at all
   • "Reply, or call me on 07…", not "worth a quick conversation?"
   • No merge field you cannot fill on every single row. "Hi ," is worse than no greeting
3. Add two or three subject line variants to step 1 using + Add Variant. Change only the subject. Keep the body identical across all of them, or the test tells you nothing
4. Add step 2 at day 3. The reason a customer would care, put a different way. Not "just following up" and nothing else
5. Add step 3 at day 10. A clean close: acknowledge the timing may be wrong, leave the door open, stop
6. Put your opt-out and a postal address in every step, not only the first. The obligation applies to each message
7. Set the schedule to weekdays, roughly 09:00 to 16:00 in the recipient's time zone

Launch, and the first week

Four checks before you press go, then one number that matters more than the dashboard.

Before launching

1. Send yourself a test of every step, to an address on a different provider from your own
2. Open each one on a phone. That is where it will be read. Check the line breaks and that the signature is not three screens long
3. Click your own unsubscribe link and confirm it works. If it does not, you have a problem in every message you are about to send
4. Confirm the daily limit is still 20 per mailbox and that tracking is on. Check it now rather than trusting part 06

Then the first week

When	Check
Day 1, an hour after launch	Mail is actually going out. Any immediate bounces
Day 2	Bounce rate. Above 5% and something is wrong with the list. Pause and look before it pauses you
Day 4	Bounce categories, not just the total. See below
Day 7	First read on replies worth having. Write the week's four numbers down

Raising the volume

Not before a run of clean days with no reputation-based rejections. Then 20 → 25 → 30 per mailbox, a week apart, and one policy-blocked bounce resets the clock. There is no prize for getting there faster and the downside is weeks of repair.