What automated link building actually builds
Data: RankMerit Authority Index, Common Crawl cc-main-2026-may-jun-jul
If you are shopping for automated link building software, there is a question worth asking before you pay for any of it: do the companies selling link automation have good links themselves?
We checked. The answer is uncomfortable.
The test
We run an open web graph — 236.7 million domains and 7.12 billion host-to-host links, rebuilt from Common Crawl. For 70 SEO, AI-SEO and answer-engine tools we counted referring domains and, more importantly, how many of those referring domains carry real authority (Domain Rating 50 or above).
What the numbers show
Established SEO suites have a median of 119 referring domains at DR 50+. The AI-native wave sits at 33. The newest category, answer-engine tools, sits at 17.
| Tool | Referring domains | DR 50+ | Share |
|---|---|---|---|
| trysoro.com | 1,357 | 8 | 0.6% |
| autoblogging.ai | 834 | 5 | 0.6% |
| machined.ai | 223 | 5 | 2.2% |
| byword.ai | 798 | 17 | 2.1% |
| koala.sh | 4,369 | 40 | 0.9% |
| ahrefs.com | 113,450 | 1,022 | 0.9% |
Look at trysoro.com: 1,357 referring domains sounds like a healthy profile until you see that eight of them clear DR 50. The rest is a long tail of dental clinics, a limousine hire company in Melbourne, a Greek hotel, a printing shop, and dozens of micro AI startups.
That pattern is not editorial coverage. It is a customer footprint — links that appear because the tool was installed somewhere, not because anyone chose to cite it. Automation produced volume. It did not produce authority.
Why volume without authority does nothing
Google's link spam policies treat links that were not editorially placed as things to ignore or devalue, not things to reward. A thousand ignored links and zero links have the same effect on ranking. The difference is that the thousand can also draw a manual review.
- Directory blasts and paid placements are cheap to detect at scale — the footprint is the giveaway, not any single link.
- Reciprocal networks show up clearly in a graph: everyone links to everyone, nobody links in from outside.
- Footer and widget links from a customer base repeat the same anchor across unrelated verticals.
None of this is a moral argument. It is an arithmetic one: the links these tools generate do not move the number they are sold to move.
What actually earns a DR 50+ link
Quality sites do not hand out body links to a guest post. Read their contributor guidelines and you find some version of the same line: article links should point to references and sources. They link to things worth citing.
So the work is not outreach volume. The work is being the source. Three formats do this reliably:
- Original measurement — a number nobody else can produce, with a stated method and date.
- A benchmark others want to position against — a ranking, an index, a threshold.
- A calculator or tool that answers a question people ask repeatedly.
The article you are reading is an example of the first. It exists because we can measure something most people cannot, and because anyone writing about link building automation now has a number to cite.
If you still want automation
Automate the parts that are safe to automate: finding which sites in your vertical already cite the kind of asset you can produce, tracking whether the assets you publish get cited, and measuring whether your authority actually moves. That is automation applied to research and measurement — not to link placement.
Want this measured on your own domain?
RankMerit runs the whole SEO and GEO operation — technical repair, daily rank tracking, content built from your Search Console, and the assets that earn citations. From $9.90/month.
Scan your site — freeRelated dataset: The SEO Tool Authority Index