Workflows

Check a list you bought before you use it

A purchased prospect list arrives as a CSV with a row count and nothing else. The row count is the one number the seller controls completely and the one number that tells you least. This is how to find out what you actually bought, in about twenty minutes, and what to do about it either […]

A purchased prospect list arrives as a CSV with a row count and nothing else. The row count is the one number the seller controls completely and the one number that tells you least.

This is how to find out what you actually bought, in about twenty minutes, and what to do about it either way. We published a full check of one such list so there is something concrete to compare against: 1,001 rows, opened by hand.

Do this before you touch the pages

1. Count distinct hosts, not rows

This is the single most revealing number and it takes one formula. Strip each URL to its registered domain and count the uniques.

Our list: 1,001 rows, 467 distinct hosts, 999 distinct URLs. Two rows were literally the same URL twice. Nearly seven rows in ten sat on a host that appeared more than once, and one Polish forum platform contributed 30 rows by itself.

A list that is 1,000 rows and 200 hosts is a 200-prospect list with padding, because you write to a person, not to a URL. Work out your ratio before anything else.

2. Look for blocks of identical Domain Rating

Sort by DR and look for large runs of the same number. Every run is a hosted platform whose subdomains all inherit one rating.

On our list: 38 rows at DR 95, all *.blogspot.com. 76 rows at DR 61, all one forum platform. 48 rows at DR 68, all church media pages. That is 162 rows, 16.2%, whose impressive scores describe somebody else’s domain.

If this pattern is large in your file, the DR column is decoration and you should stop sorting by it. Why the metric inverts on lists like these goes through the inheritance in full, including where DR is still the right column to use.

3. Count the page types from the URLs

Before fetching anything, filter for structurally dead paths: /forum/, /topic/, /sermons/, /podcast/, /media/, /guestbook/, /tag/, /category/.

Ours came out at 243 rows, 24.3%, in page types that have never had a comment form. A quarter of the list could have been rejected by a text filter, which means the seller did not run one.

4. Check whether the pages resolve

Now spend the requests. This is the first check that costs anything and it invalidates everything downstream.

Ours: 213 of 1,001 rows, 21.3%, could not be read. Dead, blocked, password-protected, or in a redirect loop. A fifth of what was sold was not a page.

5. Check whether the placement is possible

The actual question. For a comment list: is there an open comment form on that exact page today, and what would a link in it carry.

Ours: 183 open forms, 18.3%. 37 would have carried a followed link, 3.7%. Ten rows were followed, readable and low spam risk at once.

The benchmark, so you have something to compare to

Measurement Our purchased list What it means for yours
Rows per distinct host 2.1 Above 3 and you are buying padding
Rows in identical-DR blocks 16.2% Large blocks mean the DR column is inherited, not earned
Structurally dead page types 24.3% Anything above 10% means no filtering was done
Pages that could not be read 21.3% Above 15% suggests the list was assembled a long time ago
Open comment form 18.3% This is the number the list was supposedly selected for
Would carry a followed link 3.7% Ask before buying what the seller claims here

One list, one vertical mix, checked on 7 September 2026. These are that list’s rates and not industry norms. Use them as a reference point, not as a pass mark: a different vertical would produce different numbers for reasons that have nothing to do with the seller.

What to say to the seller

An argument about quality goes nowhere. A measurement is harder to wave off, and it occasionally gets you a replacement file rather than a refund fight.

Send them three things:

  1. The host concentration. “1,001 rows resolve to 467 domains” is a factual statement about the file they sent.
  2. The unreachable rate, with the date. “21.3% of these URLs did not return a readable page on 7 September” is checkable by them.
  3. The share that cannot take the placement at all. If you bought a commenting list and a quarter of it is sermon players, that is the specific thing to name.

Keep the raw output. A seller who disputes a percentage usually stops when you offer the row-level file.

What to do with what survives

Whatever is left is a real list, usually much smaller than the invoice implies, and it is worth working properly. Sort it by what you can act on rather than by DR: open forms first, then link type with Unknown ranked above nofollow, then spam risk, and DR last as a tie-breaker.

Then apply the judgements no measurement can make for you: is this page about my subject, is this a site I want to be on, and is the effort proportionate to what I would get. Those three are covered in the vetting guide.

Running the checks

Steps 1 to 3 are spreadsheet work and worth doing yourself; they cost nothing and they tell you most of what you need before you spend a request. Steps 4 and 5 need every page opened, which is two to four minutes each by hand, or about 33 hours for a thousand rows.

The Backlink Prospect Checker does 4 and 5 in bulk and returns a sortable report with the evidence behind each verdict. The free allowance is 500 checks on first use, and at one credit a URL that is five scans of 100 rows, which is enough to grade a vendor list before you decide whether to buy from them again.

Share this article

Start

Same list. A fraction of the time.

Paste the list you already have and get back a shortlist you can run a campaign from. The free plan starts you with 500 credits, which is five full Quick Checks of 100 URLs before you decide anything.

Free plan · available now

welcome
500 credits
daily top-up
to 100, only if below it
per Quick Check
100 URLs · 50 with extras
above 12 checks
refused

Paid plans are not priced yet and allowances are undecided.