Link Building Metrics: How to Measure Links, Mentions and ROI

Link building metrics tell you whether the links and mentions you pay for change rankings, AI answers and pipeline. Counting new backlinks only tells you that a vendor did some outreach. The useful metrics connect each placement to a page you want to rank, a prompt you want to win and a deal you want to close.

Flying Cat runs link building for B2B SaaS companies as part of their SEO and AI search programs. Every month the team reports on the same set of numbers, from referring domains to AI citations to the cost of each placement against the value of a deal. Below are those metrics, the way the team calculates link building ROI and the reporting template it sends clients.

The link building metrics Flying Cat tracks

Link building now has two jobs. Classic links help commercial pages rank in Google. Brand mentions on the roundups and review pages that AI assistants cite help a brand get named in ChatGPT, Perplexity, Gemini, Copilot and Google's AI Mode and AI Overviews. Each job needs its own metrics, so Flying Cat tracks them on separate lines.

Referring domains count the unique websites linking to you. They track authority better than raw backlink totals, because fifty links from one site carry far less weight than one link each from fifty relevant sites. Semrush's own guide to link building metrics makes the same call and treats referring domains as the more meaningful of the two.

Flying Cat reports referring domains at the page level, for the pages the team chose to support. Sitewide totals hide whether outreach reached the pricing page, the comparison pages and the product pages that drive pipeline. For Leapsome, the team moved part of its content budget into link building for a short list of priority pages. The team built around 50 targeted links to those pages. Leapsome's referring domains grew about 30% in twelve months.

How Flying Cat judges the quality of a referring domain

Third-party scores like Domain Rating and Authority Score give a quick filter. They are vendor estimates. Google does not use them. Before a domain counts as a good link, it has to pass these checks.

• Topical relevance. The site covers the client's category or the buyer's job. An HR software client gains more from an HR publication than from a general business blog.
• Organic traffic to the site and the linking page. A site that ranks for nothing passes little value. The team looks at the traffic trend as well, since a site that lost most of its traffic in a core update is a warning sign.
• Indexing. The linking page is indexed in Google. An unindexed page passes nothing.
• Outbound link pattern. The site links out to a reasonable number of editorially chosen pages. Sites that sell links to anyone show long lists of unrelated outbound links.
• AI citation footprint. For mention targets, the team checks whether AI assistants cite the page for the client's buying prompts. This check decides which mention placements Flying Cat pursues.

Anchor text and link attributes

Flying Cat tracks anchor text distribution across priority pages so no page ends up with a run of exact-match anchors. The team also records whether each link is followed, nofollow or sponsored. A paid placement should carry rel="sponsored" or rel="nofollow" under Google's spam policies. Nofollow mentions still count for AI visibility, since assistants read the page text regardless of the link attribute.

Rankings of target pages

Ranking movement on the linked page is the first outcome metric. Flying Cat tracks the target keyword set for each page that received links, then compares positions in the weeks before and after the links went live. A cluster of pages that climbs together after a link push is a strong signal. A single page that moves while nothing else changed on the site is a cleaner one.

Flying Cat's own research shows why this is the metric to watch. In a study of 488 product pages from 99 Forbes Cloud 100 companies, Maeva Cifuentes, Flying Cat's founder, found that referring domains correlated with AI citations at 0.68 on their own, but only 0.26 once rankings were held equal. Ranking breadth correlated at about 0.81. Links mainly help through the rankings they produce, so the team judges an SEO link by the ranking impact on its target page.

Timing affects the read as well. With Trevor.io, Flying Cat built external links only once a page had started ranking. New pages reached page one in about a month. Links to a page that already shows traction move faster and are easier to attribute than links to a page Google has not yet placed.

Organic traffic and conversions on linked pages

Rankings only pay off when they turn into visits and designated actions. For each linked page, Flying Cat tracks organic clicks in Google Search Console and conversions in GA4 and the CRM. Conversions here mean demo requests, free trial signups and other actions the client designated, never clicks.

The team follows these rules to keep the read honest.

• Compare the page against itself and against unlinked peers. The team looks at the linked page's trend before and after, and at similar pages on the same site that received no links over the same period. That comparison separates the link effect from a sitewide update or a seasonal swing.
• Report on the client's own definitions. If the sales team counts a lead as qualified only after a discovery call, Flying Cat uses that definition. For an AI customer service platform for ecommerce brands, every metric the team reports carries its CRM field name and exact filter, and the client must be able to reproduce it in two minutes or it stays out of the report.

Referral traffic from the linking site is a bonus metric. Some placements on busy industry roundups send buyers directly, which shows up in GA4 under Traffic acquisition filtered by session source. Links on smaller sites often send little referral traffic. Their job is rankings, so Flying Cat does not judge them on referrals.

A caveat on AI referral traffic

AI referral sessions have become a noisy signal. In October 2026, after ChatGPT rolled out a new interface, measured ChatGPT sessions fell by roughly half in a single week across several of Flying Cat's B2B SaaS clients. Those visits now land in GA4 as Direct traffic. ChatGPT still drove demo requests at the same pace as the previous two months. The team now treats demos and engaged sessions as the steadier signal, and it adds a "Where did you hear about us?" field to demo and signup forms so buyers can name the assistant themselves.

AI citations and brand mentions from placements

Brand mentions on third-party pages became their own link building outcome once AI assistants started building answers from those pages. Ahrefs studied 75,000 brands and found that branded web mentions correlated with AI Overview visibility at 0.664, against 0.218 for backlinks and 0.295 for referring domains. Mentions on the right pages now carry weight a link count cannot show.

Flying Cat measures mention placements with these numbers.

• Mention rate on tracked prompts. This is the share of AI answers that name the client across an agreed set of buying prompts. Each prompt runs many times a month across several assistants, so a 30% mention rate means the brand appeared in 30% of those answers. Position inside a single answer changes from run to run, so the team reports the percentage across many runs.
• Citations of the placed page. This is how often assistants cite the page that carries the placement. When a cited roundup adds the client, the client's mention rate on the prompts that cite that roundup should rise.

D2L shows the effect. Its outreach shifted from general link building to the third-party roundups that appear as sources in AI answers. D2L went from appearing in 12% to 33% of tracked AI answers, and from 11th to 7th among learning platforms. Monthly AI citations of d2l.com pages grew nearly 4x in a single quarter. At an enterprise procurement platform, placements on the roundups and industry publications that AI assistants cite went alongside a rise in ChatGPT answers citing the company's own site, from 25% to 41%.

Placements also take time to show up in AI answers, so Flying Cat reads mention rate over a rolling window of four to eight weeks. The AI search measurement guide covers prompt set design and tool setup in more detail.

Link building ROI and cost per placement

Link building ROI compares what a program costs with the value it produces. The standard formula is simple.

ROI = (attributed revenue minus link building cost) divided by link building cost

The hard part is the attributed revenue. Few deals start and end on a page that just got a link. AI-sourced visits often arrive as Direct traffic. Flying Cat combines the methods below.

Method 1. Cost per placement against deal value

Before a client pays for a placement, Flying Cat expresses its cost as a share of one average organic deal. This frames the decision for leadership in their own terms. For an AI customer service platform, one roundup the team evaluated was cited in 33 AI responses across six buying prompts. It appeared in 45% of Copilot answers to "best AI customer service platform" questions, where the client showed up in 7% of answers at the time. The placement cost came to about 1% of one average organic deal. One extra deal over the life of the placement would pay for it many times over.

The calculation uses these inputs.

1. The placement fee, plus any content or outreach time.
2. The client's average deal value for organic and AI-sourced customers, from the CRM.
3. The visibility the placement could add, estimated from how often assistants cite that page for the client's prompts.
4. The client's conversion rate from organic and AI visits to closed deals.

Method 2. Pipeline change on linked pages

After placements go live, Flying Cat tracks pipeline from the linked pages and from AI referrals. The team uses first-touch and multi-touch attribution in the CRM, plus the self-reported "Where did you hear about us?" answers. A hypothetical example shows the arithmetic.

Flying Cat reports the result as a range, since attribution is never exact. It also says plainly which part of the lift came from links and which came from content and technical work running in parallel. Links only pay off when the page they point to is worth ranking, so the team credits the program as a whole and use the link metrics to show where the spend went.

Leading indicators while ROI builds

Link ROI takes months to show in revenue. In the first quarter, Flying Cat reports leading indicators against the plan. Those indicators cover placements live against target, referring domains to priority pages, ranking movement on target keywords and mention rate on the prompts tied to each placement. The broader guide to SEO metrics explains how leading and lagging metrics fit together in a full search program.

Link building reporting template

A link building report should let a CMO see in one screen what was placed, what moved and what the spend returned. Flying Cat's monthly report follows the layout below.

Section 1. Placements this month

Section 2. Outcomes on target pages

Section 3. AI visibility from placements

Section 4. Spend and return

This section lists spend to date, cost per placement, the pipeline attributed to linked pages and AI referrals, and the ROI range.

Section 5. What did not work and what is changing

Every report carries this section. A placement that earned no citations or a page that did not move gets named, with the change the team will make next month. Clients trust a report more when it shows the misses as well as the wins.

The numbers in the template above are placeholders. Monthly reports compare month over month. Flying Cat keeps year-over-year comparisons for quarterly reviews, where the team also refreshes the backlink audit and set the next quarter's targets.

Metrics to stop reporting as outcomes

Some numbers look like progress but say little about results. Flying Cat still glances at them, but they never lead a report.

• Total backlinks. A sitewide count rewards volume, including low-value links from directories and scrapers.
• Domain Rating or Domain Authority alone. These scores are useful filters for prospecting. A rising score with flat rankings and flat demos means the links went to the wrong places.
• Emails sent and reply rates. Outreach activity belongs in the vendor's internal dashboard. Clients pay for placements and outcomes.
• Toxicity scores. Very few links a normal business receives harm it. Google's own disavow guidance reserves the tool for sites with many spammy or artificial links that have caused, or will likely cause, a manual action.

How Flying Cat measures link building

Flying Cat runs link building inside its full-service SEO and AI search programs and never sells it on its own. Each client's budget splits between classic SEO links to commercial pages and brand mentions on the sources AI assistants cite. Publisher fees pass through at cost. Link building inside a program starts from about $3,000 a month.

Every placement target gets scored before the client pays. Classic link targets need relevance, traffic and an indexed linking page. Mention targets come from the URLs AI assistants cite for the client's tracked prompts. Maeva approves each target before outreach starts. Freelance specialists who know the client's vertical run the work. The team reports SEO links and AI mentions on separate lines, each against its own outcome.

If you are comparing vendors, Flying Cat's list of SaaS link building agencies covers what each one offers. For tactics, the guide to link building strategies covers how to earn links in the first place. The off-page SEO guide covers reviews, communities and digital PR.

Frequently asked questions about link building metrics

What are the most important link building metrics?

The ones Flying Cat ranks highest are new referring domains to priority pages, ranking movement on those pages, conversions from those pages and AI mention rate on the prompts tied to each placement. Each one connects outreach to an outcome a CMO cares about.

How do you measure link building ROI?

Divide the attributed revenue minus the program cost by the program cost. Flying Cat attributes revenue from opportunities on linked pages and AI referrals, using CRM attribution plus a self-reported source field on demo forms. Before any spend, the team compares each placement's cost with the value of one average deal.

How long does it take to see results from link building?

Pages that already rank on page two or three can move within weeks of new links. New pages and competitive keywords usually take three to six months. AI mention rate on prompts tied to a cited roundup can change within a month or two of a placement going live, so Flying Cat reads it over a rolling window.

What should a link building report include?

A useful report lists each placement with its target page or prompt, the outcome on that page, AI visibility changes, spend against return and a section on what did not work. Activity numbers like emails sent belong in the vendor's own tracking.

Are Domain Authority and Domain Rating good link building KPIs?

They work as filters when choosing targets. Neither is a Google ranking factor. Both can rise while rankings and pipeline stay flat. Report rankings, conversions and AI mentions as the outcomes instead.