How link building outreach works on Pressmerit
Eight stages, from the first read of your site to a link that is still being re-checked months later. The platform runs six of them. Two need a person: hitting send, and agreeing the terms.
This is the long version: what each stage does, what it costs, and where it stops. The places the platform cannot see are written down here too.
01 / 08
The platform reads your site first
Add a domain and Pressmerit reads it before it proposes anything. Up to 25 pages: the homepage, the about page, pricing and product pages, whatever sits in the top navigation, and the most recent posts on the blog. It works from what you publish, not from what you type into a form.
Out of those pages come three decisions it needs later: which of 26 industry categories you belong to, which region you sell into, and which language you publish in. Every prospect the platform proposes downstream inherits them, so they are shown to you on the analysis card rather than kept internal.
It then drafts keywords in three groups (commercial, branded, informational) and pairs them with the pages on your site worth pointing links at. All of it is editable. Wrong category, or a keyword you would never bid on? Change it, or swap in your own. Adding and analyzing a site costs 10 ⚡; re-running the whole analysis afterward costs 5 ⚡.
domain score
41/ 100
Analysis card — example data for a fictional site.
02 / 08
Domain Score, and what feeds it
Your own domain carries a Domain Score from 0 to 100, and so does every prospect you are shown. It is our own number, computed from four signals: link authority, organic traffic, profile breadth, and naturalness.
Link authority reads how many independent domains point at it and how much weight those domains carry. Organic traffic estimates the search visits the domain earns. Profile breadth asks whether links arrive from many different sources or from the same handful repeatedly. Naturalness looks at whether the pattern reads as earned rather than assembled.
The score never travels alone: its components are always visible wherever it appears, on your dashboard and on every prospect card. Trust in a number requires seeing what feeds it. What we will not do is re-badge another vendor’s metric under our own name and present it as ours.
The weighting behind the four signals is not published, and that is a deliberate limit rather than an oversight: a score whose formula is public is a score that gets farmed. The inputs are open, the direction each one pushes is open, the arithmetic is not.
The four signals behind every Domain Score.
03 / 08
Finding sites on the open web
Once you approve the keywords, the platform searches the open web for sites that already rank and write about them. There is no catalog to browse: candidates come out of live search results, not out of a list of sites that registered themselves as inventory.
Candidates are filtered before you ever see them. Your own domains come out. Near-identical competitors come out. So do domains carrying spam and link-farm signatures, and sites in niches we will not work with at all. That rejection happens upstream, which is why a prospect list is short and unglamorous rather than long and hopeful.
What survives gets a match score from 0 to 100, built from topical similarity, authority fit, industry, region, and language. The components sit behind the score on the card; hover it in the product and each one is spelled out, so a 72 that lost its points on region reads differently from a 72 that lost them on topic.
match breakdown · quietfold.press
87Prospect rows with the match breakdown that opens on hover.
04 / 08
The person who can actually say yes
A pitch into a generic inbox is a pitch to nobody. For each site the platform works down a role ladder: marketing first, then content and editorial, then SEO and growth, then partnerships, and only when none of those can be found, the generic inbox.
The address is verified before it is revealed to you. No pattern-guessing: a bounce burns your mailbox, not ours. The card shows who the person is, the role they were matched to, a confidence figure, and where on the site the address was found, so you can judge the contact before you spend a send on it.
When no contact can be found at all, the 10 ⚡ hold on your credits (⚡ are Pressmerit’s usage credits; the pricing page lists every action’s price) is released in full and the request is never charged. A request is captured only when a verified contact is revealed: the contact and the written pitch are what the 10 ⚡ buys. The prospect keeps that mark, so the same dead end is not paid for twice.
Marta Kowal
Content editor
confidence
86
m•••@quietfold.press
found on /about — verified
Revealed contact with confidence and provenance.
05 / 08
You send it, from your own mailbox
The platform prepares the whole message: a subject line, a body written against the specific page you want linked, a claim link the recipient can open, and a disclosure footer that says why they received it and how to opt out of everything.
Then it stops. You copy the message or open it in your mail app, and you send it from your own address. Your domain, your reputation, your relationship with the person on the other end. Pressmerit runs no sending infrastructure and never sends in your name.
Marking a message as sent starts the clock. At +4 days and again at +9 you get a reminder to follow up, each with a fresh copy variant rather than the same text resent. Those reminders go to you. Nothing leaves your mailbox unless you send it.
The limit that comes with that arrangement is worth stating plainly: because the send happens outside the platform, deliveries and opens are invisible to us. We cannot report an open rate we never saw, so we do not report one, and we do not charge for sends.
tom•••@quietfold.press
Data for your piece on small-press print runs
Hi Marta — your piece on small-press print runs is the clearest write-up I have found on the subject.
I pulled the numbers on 240 independent runs this year. Happy to share the dataset, or write two short paragraphs you can cite.
Not a fit? Reply once and you will not hear from me again.
Why you got this · opt out
Prepared draft — you send it, we never do.
06 / 08
What the other side sees
The claim link opens the full request without a signup wall: who is asking, which page they want linked, which anchor, on what terms. The recipient can accept, open a discussion, or decline. Declining takes one click and needs no account. The fastest path through a request is always the honest no.
Discussion happens in a chat, but terms do not live in the chat. They live in a versioned agreement covering placement type, anchor, and rel policy. One side puts a version on the table, the other accepts it or answers with another version. Chat never silently rewrites terms.
That is also why there is a record afterwards. Verification later checks the published page against the version both sides accepted, not against a sentence someone typed at midnight in a thread.
request
- from
- tapelessaudio.com
- target page
- /guides/tape-transfer
- received
- by email · claim link
Reading it, discussing it, and declining it all work without an account, and none of it costs the publisher anything.
agreement · v2
proposed by the publisher
- placement type
- guest article
- anchor
- independent print runs
- rel policy
- dofollow
Agreement version — the only place terms are binding.
07 / 08
What counts as a verified link
When the piece goes live, Pressmerit fetches the URL and checks four things: the link is present in the server-rendered HTML, it resolves to the target that was agreed, the agreed rel policy is honored, and the page itself is indexable.
Each of those fails on its own terms. A link injected by JavaScript after load is not present as far as a crawler is concerned. A page-level nofollow fails the rel check even when the link tag looks correct, because the directive covers the whole page. A page that is not indexable passes nothing on, however good the link on it looks.
Anchor text is the exception, and it is a deliberate one: a changed anchor is a warning, not a failure. Editors adjust anchors to fit their own house style, which is a legitimate edit rather than a broken deal. You are shown exactly what was published and you decide whether it still works; nothing is held against the publisher for it.
quietfold.press/print-runs
- Passed: link present in the server-rendered HTML
- Passed: resolves to the agreed target page
- Passed: rel policy honored — dofollow, no page-level nofollow
- Passed: page is indexable
- Warning: anchor text differs from the agreed version
result: verified · 1 warning raised to you, none against the publisher
Verification result — four checks, one warning.
08 / 08
The link is watched for as long as you stay
A verified link is not a finished link. Every placement is re-checked weekly through its first month, then every 30 days for as long as the subscription is active. Verification and monitoring cost zero credits, in the plan and out of it.
A clean check is silent; you are not notified about a link that is still where it should be. If a link disappears, both sides are told: you, and the site that published it. From that notice you can send a restoration request in one click, which arrives as a short note pointing at the agreement the two of you accepted.
Not every failed fetch is a lost link. Timeouts, rate limits, and a server that is down for ten minutes are transient, so a check that fails on the network is retried before anything is called a failure. We would rather tell you a day late than tell you a link is gone when it is not.
quietfold.press/print-runs
peakroast.blog/gear-guide
link not found · day 62
Weekly through month one, then every 30 days.
Questions about the pipeline
What happens if no contact is found?
The hold on your credits is released in full and the outreach request is never charged. The prospect is marked as having no reachable contact, so it does not come back into your list and nobody re-pays for it. If an address that was revealed later turns out to be provably dead, support credits it back on review.
Can I run the steps manually instead of on autopilot?
Yes. Autopilot is on by default because most of this work is mechanical, but every step has a manual mode: run the analysis and stop, pick prospects yourself, reject the drafted pitch and write your own, start outreach one site at a time. Autopilot also runs inside a credit budget you set, and you can pause it at any point without losing the state it has built.
Who decides the rel attribute on a published link?
The two parties do. Whether a link is dofollow, sponsored, or nofollow is a term of the agreement between you and the publisher, recorded explicitly in the version both sides accepted. Pressmerit sets no default and changes nothing afterwards. Verification then checks the published page against whatever that version says. The anchor travels in the same version, and if an editor rewrites it at publication time, that raises a warning, not a failure.
Which sites are excluded from discovery?
Your own domains, near-identical competitors, domains carrying spam signatures or sitting in toxic niches, and anyone on the opt-out list. Publishers can opt out of receiving requests entirely, and that list is applied before a prospect is shown to you, so you are not spending credits on a site you were never going to be able to contact.
The pipeline is the product
Add a site, approve what the analysis drafts, and let autopilot work down the list. You send from your own mailbox; the platform verifies what gets published and keeps checking it every 30 days.