Backlinks used to have one job. They told Google your page was worth trusting, and rankings followed.
That job has split in two. Links still drive rankings, but a second system now decides whether ChatGPT, Perplexity, and Google AI Overviews name you in an answer. The signals overlap, and they are not identical.
How Backlinks Still Drive Google Rankings
Despite regular claims that links are finished, the evidence has not moved much.
- Referring domain count remains the strongest measurable correlator with rankings across large studies
- Analysis of 11.8 million results found the top ranking page had roughly 3.8 times more backlinks than positions two to ten
- Depending on the dataset, between 66 and 96 percent of pages have no external backlinks at all
That last figure is the useful one. The bar for standing out is far lower than most teams assume, because most pages never earn a single link.
What Links Actually Signal
Understanding the mechanism helps you choose better placements.
Authority
- Links from trusted domains pass credibility to yours
- A first time referring domain counts for more than another link from an existing one
Relevance
- Links from sites in your category help define what you are about
- Surrounding text gives search engines context beyond the anchor itself
Discovery
- Links are still how crawlers find and revisit pages
- Orphaned pages with no inbound links get crawled rarely
How AI Search Changes the Calculation
This is where strategy has genuinely shifted, and where most SaaS teams are still behind.
Mentions Outweigh Raw Link Counts
- Analysis of 75,000 brands found branded web mentions correlated with AI visibility at roughly 0.66
- Backlink counts correlated at about 0.22 over the same dataset
- An unlinked mention on a trusted source can carry weight a raw link does not
Nofollow Is Not Worthless Any More
- A study of 1,000 domains found nofollow and follow links showed similar correlation with AI mentions
- Rejecting a placement purely because of the link attribute is now a mistake
Authority Still Gates Entry
- Analysis of 129,000 domains found referring domains were the strongest predictor of ChatGPT citation
- Domains under 300 referring domains averaged about 2.5 citations
- Those above 24,000 averaged 6.8
The practical synthesis is simple. Links get you into consideration. Mentions decide whether you get cited.
Why Digital PR Sits at the Centre
One tactic satisfies both systems from a single placement.
- A press mention produces a link, a brand mention, and often a quote
- Coverage frequently gets syndicated, multiplying all three
- Trusted publications are weighted heavily as sources by AI systems
- Paid placements deliver one signal, editorial coverage delivers several
That is why the effort gap between editorial and paid link building is narrower than it first appears.
Google Rankings vs AI Visibility Side by Side
| Signal | Weight for Google Rankings | Weight for AI Visibility |
|---|---|---|
| Referring domain count | Very high | Moderate, gates entry |
| Branded mentions | Moderate | Very high |
| Unlinked mentions | Low | High |
| Nofollow links | Low | Comparable to follow |
| Anchor text | Moderate | Low |
| Source trust | High | Very high |
Note where the two columns disagree. Anchor text matters for one and barely registers for the other, while unlinked mentions are nearly worthless for one and valuable for the other.
What to Do Differently
- Stop rejecting nofollow placements on quality publications
- Chase mentions, not just links, and count both in reporting
- Prioritise trusted sources over high metric sites with no audience
- Keep building referring domains, since authority still decides whether you qualify to be cited
- Use branded and partial match anchors, which help entity recognition as well as avoiding spam signals
How to Measure Both
Most teams track one system and assume the other follows. Track them separately.
- For rankings: referring domain growth and position movement on the specific pages linked
- For AI visibility: run your top 20 buyer questions through ChatGPT, Perplexity, and Google AI Overviews monthly, logging which sources get named
- For both: branded search volume, which rises when the mention layer is working
Only around 19 percent of SEOs have changed how they build links, despite 74 percent believing links influence AI visibility. That gap is the opportunity.
Final Thoughts
Backlinks are not obsolete. They have become one of two signals rather than the only one that matters.
Build referring domains so you qualify for consideration, and earn branded mentions so you get named in the answer. Digital PR is the one tactic that reliably does both at once, which is why it now sits at the centre of serious SaaS link building.
Which Pages Benefit Most
Where a link points determines whether it produces revenue or just authority.
- Linkable assets such as studies and free tools, which attract links naturally and pass equity onward
- Comparison and alternatives pages, which capture buyers already evaluating vendors
- Category landing pages targeting your highest intent terms
- Product pages, reached mostly through internal links rather than direct outreach
Pointing every link at the homepage is the most common way SaaS teams dilute a campaign. Authority arrives in one place and never reaches the pages that convert.
How Long Before Any of This Shows Up
Expectations cause more campaign cancellations than results do.
- Weeks 1 to 4: links go live, crawling begins, nothing visible changes
- Months 2 to 3: early movement on lower competition pages
- Months 4 to 6: meaningful ranking change on comparison and category terms
- Months 6 to 12: compounding effects as assets accumulate their own links
Independent analysis has put B2B SaaS link building ROI at around 702 percent with a seven month break even point. That is a strong return, but it is not a first quarter channel, and budgeting it as one guarantees disappointment.
A Note on Risk
- Google has been explicit that when spammy links are devalued, the ranking benefit they generated cannot be recovered
- There is no undo, which is why cheap bulk placements are a poor trade
- Editorial coverage carries almost no policy risk by comparison
