Most outreach tools send from a shared pool of servers you have never seen. MRK Outreach does not. Every email leaves from your own mailbox, through your own provider, using an app password you create and can revoke. That one decision shapes everything else in this guide: your sending limits are your provider's limits, your reputation is your own, and replies land in your inbox exactly as they would if you had typed the email yourself.
This walkthrough covers the whole path, from connecting a mailbox to reading the replies a campaign brings back. If you are on the 14-day trial, everything here works except automatic reply handling, which is on Pro and above.
Before you start
You need three things: a mailbox you are willing to send cold email from, a list of sites you want links from, and a first email worth reading. The tool supplies the machinery for the second and third. It cannot supply judgement about who to contact, and a precise campaign sent to the wrong 500 people is still the wrong campaign. The link building strategy guide covers how to decide who belongs on the list.
A word on the mailbox. Do not use the address your invoices go to. Cold outreach carries a small but real risk of spam complaints, and you want that risk sitting on an address you could abandon without consequence. A dedicated address on your own domain is ideal; a separate free Gmail account is acceptable.
Step 1 — Connect your mailbox
Open Settings and add a mailbox. You will be asked for SMTP details, IMAP details, and a password. For Gmail, that password is not your normal one. Turn on 2-Step Verification, then create an app password at myaccount.google.com/apppasswords and paste that instead.
When you save, the tool tests both halves of the connection separately and gives the mailbox a status. The difference matters more than it looks:
| Status | What works | What does not |
|---|---|---|
| Ready | Sending and reading. Replies, bounces and opt-outs are detected automatically. | Nothing. |
| Send only | Sending. Campaigns run and follow-ups go out. | Reply detection. Follow-ups will keep going to people who already answered. |
| Unverified | Nothing yet. | The credentials did not pass. Check the app password. |
Do not settle for Send only. A campaign that cannot see replies is a campaign that emails someone a third time after they said yes. If IMAP fails, the error now tells you exactly why rather than guessing at your Gmail settings.
Step 2 — Import your prospects
Go to Prospects and upload a CSV or XLSX. Column headers are matched by intent rather than position, so Domain, Website and URL all resolve to the same field and you do not have to reshape an export before uploading it. The mapping screen shows you what it worked out and lets you correct anything it read wrongly.
The only column that genuinely matters is the domain. Everything else — contact name, site name, notes — improves personalisation if present and is quietly skipped if not.
Step 3 — Let it find the contacts
After import, the tool visits each site looking for a contact address, scores how confident it is, and checks that the domain can actually receive mail. Prospects move through a small set of states as this happens:
| Status | Meaning |
|---|---|
| New | Imported, not yet looked at. |
| Ready | A contact address was found and passed the MX check. Eligible for a campaign. |
| No contact | Nothing usable was found. These are excluded rather than guessed at. |
| Contacted | A first email has gone out. |
| Responded | They wrote back. Follow-ups stop immediately. |
Confidence is worth understanding before you use it as a filter. A published editor@ address on a contact page scores high. A pattern guessed from a name scores low. Filtering to 70% and above gives you a smaller, cleaner list; dropping to 40% gives you volume with more bounces. Neither is wrong, but bounces cost sender reputation, so start high on a new mailbox. More on how contacts are found and scored is on the link prospecting page.
Step 4 — Write your templates
Under Templates, write a first-touch email and at least one follow-up. Merge tokens pull in what the import gave you, so a token with no value degrades to nothing rather than printing a bare placeholder into someone's inbox.
Most replies come from the second or third email, not the first. A campaign with no follow-up throws away most of its own results, which is why the campaign builder nudges you towards adding one.
If you want a starting structure, nine outreach email templates covers the common asks, and the templates feature page shows how they are managed in the app.
Step 5 — Build the campaign
Open Campaigns and create one. The settings you choose here decide how much the campaign is allowed to do without you:
| Setting | What it decides |
|---|---|
| Send from | Which connected mailbox the emails leave from. |
| First email | The opening template. Rendered per prospect when the queue is built. |
| Follow-ups | Which templates follow, and how many days after the previous message. |
| How many prospects | Size of this batch. Anyone already queued, already contacted, or on your do-not-contact list is excluded automatically. |
| Minimum confidence | The quality floor for contact addresses. |
| How it runs | Review each batch holds everything until you approve it. Send automatically starts as soon as the queue is built. |
| What counts as a win | Free links only, paid placements, or either. This shapes how paid requests are handled later. |
| Email progress to | Where the periodic progress report is sent. |
Choose Review each batch for your first campaign. You have not read your own first email in the voice of a stranger yet, and the review screen is where that happens.
Step 6 — Read the queue before anything sends
A campaign is built as a draft with the whole send list rendered and waiting. Open the queue and read the first few messages exactly as they will arrive: the real subject line, the real merge values, the real signature. This is the single highest-value minute in the whole process. Almost every embarrassing outreach email that has ever gone out would have been caught by someone reading three of them before pressing start.
The batch plan on the same screen shows how the send will spread across days, which is usually longer than people expect.
Step 7 — How pacing works
Nothing sends in a burst. A new mailbox warms up, starting at a handful of emails a day and climbing, and every mailbox has a daily cap with randomised gaps between sends. Roughly fifty a day is the safe ceiling for cold email on a free Gmail account; a Workspace address on your own domain supports more.
This is why a thousand-prospect list is a multi-week campaign rather than an afternoon. The alternative — sending fast — gets the mailbox flagged, at which point nothing you send arrives at all.
Step 8 — Reading the replies
Once a mailbox is Ready, the inbox is checked regularly and every reply is classified. This is what turns the Conversations screen (the outreach CRM) into something you can triage rather than a second inbox:
| Classification | What it means | What happens |
|---|---|---|
| Interested | A positive human reply with no fee attached. | Follow-ups stop. Yours to answer. |
| Wants paying | They quoted a price to publish. | Follow-ups stop. Handled per your campaign goal. |
| Opt out | They asked not to be contacted. | Added to the suppression list permanently. |
| Bounce | The address is dead. | Excluded from further sending. |
Alongside this, each prospect carries a deal type — free or paid — which is a different question from what any single reply said. A site can answer interested and quote a fee two emails later. The Free and Paid tabs and the matching counters on the dashboard track the terms you actually ended up on, which is the number that tells you what your link building is costing.
Automatic reply handling
On Pro and above you can let a campaign answer replies for you. It is worth being precise about what this does, because it sends real email to real people without you seeing it first.
The text is composed from templates rather than a model. That is deliberate: a template is something you can read in advance and predict, which matters when nobody is watching. Around it sit hard rules:
| Rule | Why |
|---|---|
| Never names or asks a figure | Money stays a human decision. Any draft containing a number is blocked before it sends, whatever produced it. |
| Paid requests get one fixed answer | A polite decline and a swap offer — or, if your goal allows paid, the thread is handed to you rather than negotiated. |
| Anything legal or unclear stops | Contracts, invoices, complaints, anything it does not understand. Failing closed costs hours; failing open costs a relationship. |
| Three replies per prospect, maximum | A loop with a real person on the other end is the worst available outcome. |
| Never replies to opt-outs or bounces | Self-explanatory, and worth stating. |
Leave it off until you have watched a few conversations go well by hand. It is a switch you turn on once you trust the campaign, not one you inherit.
Editing a campaign after it starts
Campaigns can be edited in place. Changing the name, the sending mailbox, the follow-up ladder, the run mode, the goal or the notification address takes effect immediately. Changing the first email rewrites the messages still queued and leaves everything already sent exactly as it was sent — your record of what actually left the mailbox is never rewritten.
Pausing leaves the queue intact. Stopping cancels everything not yet sent. Neither deletes the history of what went before.
What the rest of the screens are for
| Screen | Use it for |
|---|---|
| Dashboard | Sent, replies, positive replies, and the free-versus-paid split at a glance. |
| Outreach | Conversations, the sending queue, and the follow-up schedule. |
| Domains | The sites you are working, and what stage each is at. |
| Tasks | The manual jobs a campaign creates — a draft to write, a call to make. |
| Reports | Performance over a date range, plus CSV export. |
| Integrations | Webhooks, so a reply can trigger something in another tool. |
When something is not working
| Symptom | Usual cause |
|---|---|
| Replies show as zero although people have written back | The mailbox is Send only. Reply detection needs working IMAP. Re-verify it in Settings. |
| Nothing is sending | The campaign is a draft awaiting review, or the daily cap for today is already used. |
| Fewer prospects queued than expected | People already contacted, already queued, or on the suppression list are excluded, and anyone below your confidence floor is skipped. |
| A mailbox stopped sending mid-campaign | The app password was revoked or regenerated. Creating a new one and re-verifying restores it. |
A sensible first week
Connect one mailbox and confirm it reads Ready. Import fifty prospects rather than a thousand. Write one first email and two follow-ups. Build the campaign in review mode, read three queued messages end to end, then start it. Let it run for a week before judging anything — the first email is not where most replies come from, and a campaign judged on day two is being judged on a quarter of its own results.
Once that cycle has worked once, scale the list rather than the number of things you are changing at the same time.
Plans and limits for each tier are on the pricing page. For the strategy side, the email outreach guide is the best next read, and the rest of the series is on the MRK Outreach blog.