Google describes link analysis systems as part of its ranking systems. That supports taking links seriously. It does not support treating backlink count as a complete explanation for why one page outranks another.

A page also needs to answer the search well, be accessible, and give people a reason to trust it. We would fix a missing or confusing product page before paying to promote it.

A useful link can help someone discover a page, connect related information, and send interested visitors. Look beyond whether a tool labels it dofollow or assigns the domain a large number.

Consider two hypothetical placements: a feature comparison used by your intended customers, and a page that exists mostly to sell unrelated links. The former gives you a defensible reason to seek coverage even if its authority metric is lower.

What changes when AI search enters the discussion?

A source cited in an answer is worth investigating because it reveals part of that answer's evidence. It does not establish a universal recipe for AI recommendations. Track those observations separately from ordinary search performance.

A practical order of work

  1. Confirm the destination page is accessible and useful.
  2. Improve its product facts, examples, or original material.
  3. Identify relevant places where a link helps the reader.
  4. Measure coverage and business outcomes over time.

We would not abandon link earning because search interfaces changed. We would abandon a campaign whose only success measure is the number of links it can report.