A $249 directory submission run, audited 99 days later
Done-for-you directory submission is sold by the row. We bought a run, waited, and then opened every site on the sheet to find out how many rows became a listing anybody could see.
On 17 May 2026 we bought a manual directory submission service for one of our products. Not SubmitMap: a small consumer site, so the run had nothing to prove either way. $249, one payment, no renewal. Four days later the delivery arrived as a public Airtable sheet with 229 rows on it.
That sheet is the entire product. It is what you are buying, and it looks like a lot. Two hundred and twenty-nine directories, one line each, delivered inside a week.
We left it alone for three months and then checked every row.
What the sheet actually contained
The first thing worth saying is that 229 rows are not 229 directories. Twenty nine of them were blogs, dead hosts, duplicates of a site already on the list, or pages that had never accepted a submission from anybody. That leaves 200 sites where a submission was at least possible.
The second thing is what the sheet does not contain. There are no listing URLs anywhere in it. Every link points at either a directory's homepage or its submission form, which means the sheet cannot tell you whether any given submission worked. The Status column is blank on 228 of the 229 rows.
So the delivery is a record of what was attempted. It is not a record of what happened, and those are different products.
How we checked
On 28 August 2026, 99 days after delivery, we went through all 200 real directories by hand. For each one:
- Crawl the sitemap, where the site publishes one.
- Run the directory's own search for the product name, plus a control query, so a site that returns its full index for every search does not count as a hit.
- Try sixteen listing URL patterns, the common shapes a directory uses for a product page.
- Open every hit and require the product's own domain in the body, so a category page that merely mentions the name does not count.
- Read the
relattribute out of the HTML on the ones that passed.
Search engines refuse a plain curl, so the search step ran in a real browser.
The whole thing took a day.
The result
| Stage | Count | Share |
|---|---|---|
| Rows delivered | 229 | |
| Really a directory | 200 | 87% of rows |
| Carried a live listing | 30 | 15% of the 200 |
| Passed a followed link | 9 | 30% of what went live |
Nine followed links for $249 is about $28 each. That is not a scandalous number on its own. Plenty of link building costs more.
The problem is the other one: the buyer cannot name which nine. Nothing in the delivery says which submissions landed, so the only way to find out is to do what we did, which took a day. A run you cannot audit is a run you cannot repeat, and it is definitely a run you cannot improve.
Why 170 submissions went nowhere
They were never eligible. That is the whole answer.
Directories publish their rules, and the rules are specific. A free subdomain disqualifies you from a large share of them. So does a waitlist page with no product behind it, a missing English version, a domain rating of zero, the wrong category, or a badge the directory requires on your homepage before it will publish. None of these are secret. All of them are on the directory's own submission page, and a volume run reads none of them, because reading them is the expensive part and the row is what is being sold.
This is also why the followed-link share is 30% rather than something higher.
Plenty of directories publish a listing and give it rel="nofollow", which is
their right and is fine, but it means a run that targets by count rather than
by what a platform returns collects a lot of listings that pass no authority at
all.
What we would want instead
The measurement that matters is not how many forms were sent. It is what share of the attempts became something you can point at afterwards.
That number can only go up by cutting attempts, not by adding them. Every platform that was going to reject you is a form filled, a month waited, and a row on a sheet that says nothing happened. Reading the eligibility rules first is unglamorous and it is the only lever there is.
We put our own launch through that process and counted it the same way. The comparison is on the automation page, including the row where our number is worse than theirs.
Both sides are unnamed on purpose. The vendor did what it advertised, and the point here is not that they are bad at it. The point is that the thing being sold, a count of forms, is not the thing anybody actually wants.
