An Ahrefs export is a good starting list and a bad outreach list, and the gap between them is about an afternoon. This is the sequence that closes it without wasting requests.
You already know Ahrefs, so this does not explain what a backlink is or what the Site Explorer does. It covers the export, the four cleaning passes worth doing before anything opens a page, and the point where domain data stops being able to answer the question.
1. Pull the right report
For comment-link prospecting the useful starting points are the reports that give you pages rather than domains, because comment forms live on pages:
- Site Explorer, a competitor’s Backlinks report. The referring pages that already link to someone like you. This is the highest-fit source because a page that has linked once has demonstrated it can.
- Content Explorer, filtered to your subject and to pages published recently. Recency matters more here than authority: an active post has an active thread.
- Link Intersect across two or three competitors, when you want pages that link to several of them and not to you.
Whichever you use, export at the page level. A domain list means someone has to find the right page on each domain later, and that someone will be you.
2. Keep five columns, drop the rest
Exports are wide and most of it is noise for this job. What earns its place:
| Column | Why keep it |
|---|---|
| Referring page URL | The whole input. Everything downstream is a fact about this exact URL. |
| Domain Rating | Useful as a tie-breaker at the end. Not as a sort key at the start, for reasons below. |
| Referring page title | The cheapest relevance signal you get for free, and it survives the page being unreachable later. |
| Page traffic | The closest thing in the export to “does anyone read this”. More predictive of a useful placement than DR. |
| First seen or last seen date | A page last crawled two years ago is a different prospect from one crawled last week. |
Ahrefs renames columns and moves reports around. This was written on 7 September 2026; if a label has changed since, the column you want is the one carrying that meaning. Their own export documentation is the current authority.
3. Four cleaning passes that cost nothing
Do all four before a single page is fetched. Each is a spreadsheet operation and together they usually take a third or more off the row count.
Group by registered domain
Not by URL. On the purchased list we published, 69.9% of rows sat on a host that already appeared. Twelve pages on one domain is one conversation with one person, not twelve prospects. Keep the two or three strongest rows per domain and park the rest in a second sheet.
Drop the page types that cannot take a comment
Filter the URL column for paths that are structurally dead ends: /tag/, /category/, /author/, /forum/, /topic/, viewtopic.php, /sermons/, /podcast/, /guestbook/. On our list, audio and video players, forum threads and guestbooks alone were 243 rows, 24.3%.
Drop the wrong language and market
If the campaign is English, a Cyrillic or Polish slug on a matching ccTLD is a free decision. Make it now rather than after you have spent a request on it.
Strip tracking parameters and session ids
A URL with a session id is not a stable address, and it will produce a different page tomorrow or none at all. Trim query strings unless the parameter is load-bearing, which on a blog post it almost never is.
4. The point where the export runs out
Now you have a smaller, cleaner file, and nothing in it answers the question you actually have: can a link be placed on this page, and what would it carry.
Everything up to here was stage 1 of the qualification sequence: rejections cheap enough to make in a spreadsheet. What follows is stage 2, and the reason it is worth running is that it is the only stage that reads the page itself.
The export cannot answer it, because both facts are properties of the page and the export describes the domain. Comments close when a post is archived. A theme update removes the form. A plugin turns every comment link nofollow without changing a word of the copy. None of that moves Domain Rating.
There is a second reason not to lean on DR here, and it is the one that costs people real lists. On our 1,001 rows, not one of the 285 at DR 50 or above could carry a followed comment link, and 34 of the 37 that could sat at DR 10 or below. The high scores were hosted subdomains inheriting their platform’s rating. A DR 30 floor on that list removes every usable row. The full explanation is here.
5. Check the pages
Four facts per URL, all of which have to be read from the page:
- Whether the page can be read at all. On a bought list this alone was 213 of 1,001 rows.
- Whether a comment form is open on that exact page today.
- What the page already gives its commenters, read from links already in the thread rather than assumed from the platform.
- What the existing comments look like, which is the cheapest quality signal on the page.
By hand this is two to four minutes a page. The Backlink Prospect Checker does it in bulk: paste or upload the URL column, run a Quick Check, and every row comes back with those four fields, a verdict, and how much evidence the verdict came from. Five checks a URL is one credit, so 100 URLs is 100 credits, and the free allowance is 500 on first use.
It reads only. Nothing is posted, no account is signed into, no CAPTCHA is attempted. Where a page has an open form but nothing in the thread to read, the link type comes back Unknown rather than a guess.
6. Sort the report, and sort it in this order
The temptation is to sort by DR. Do not, for the third time. The order that reflects what you can act on:
- Comment opportunity, open first. Everything else is irrelevant on a page you cannot post to.
- Link type, followed first, then Unknown, then nofollow and ugc. Unknown belongs above nofollow: it might be worth something and it costs one page-load to find out.
- Spam risk, low first. A followed link in a comment dump is a followed link in a comment dump.
- Then Domain Rating, as the tie-breaker it is good at being.
7. Hand it over
Copy the qualified rows into your tracker or export them. Keep the verdict column and the confidence column together: a junior working the list needs to know that followed, 9 links read and followed, 2 links read are different levels of certainty, and a bare verdict hides that.
Add owner and date columns, and record the scan date on the sheet. Everything in the report was true on the day it was read, and in three months some of it will not be.
What you are left with
On a bought list the survival rate is brutal: 1,001 rows in, ten worth an email. On an Ahrefs export built from competitor backlinks it is usually much better, because every page in it has already linked to somebody. That is the strongest argument for starting from an export rather than from a purchased list, and it is worth knowing before you spend anything.
If you want to see the columns before running anything, there is a full sample report with no sign-in.