Measuring digital PR: links, mentions and commercial outcomes

Measure digital PR in three layers, and keep them apart. First, what the campaign earned: unique linking domains and their quality. Second, the visibility it created: mentions, with or without links, and presence in AI answers. Third, what changed for the business: rankings, traffic and revenue on the pages the campaign was meant to help. A link count alone tells you about the first layer and nothing about the other two.

What practitioners measure now

BuzzStream’s State of Digital PR 2026 surveyed more than 150 practitioners on the metrics they use.

Metric	Share who track it
Quality links	85.1%
Total mentions	72.3%
Total links	70.9%
Organic traffic or rankings	63.5%, down from 78% in 2025
AI citation mentions	55.4%, new in 2026

Two things stand out. Mentions and AI citations have become standard measures. And fewer teams now tie their work to traffic and rankings than a year ago. Measuring impact was named as the top challenge by 14.9% of respondents, second only to getting coverage.

The commercial layer is thinner still. Digitaloft’s July 2026 compilation of industry statistics reports that only 28.4% measure by leads or conversions, and that only 28% track revenue from referral traffic. It is a compilation of several samples, so treat the figures as approximate.

Layer one: links

Count links in a way that cannot flatter the result. Our reporting rules are:

• Report unique referring domains, not raw links.
• Exclude syndicated copies from the total and list them separately. See syndicated links.
• Show nofollow links on their own line. Do not hide them and do not merge them.
• Never report an authority score alone. A high score on an irrelevant site means little.

For each linking domain, record:

• relevance of the site and the article to your subject. In BuzzStream’s 2026 report, 87.2% of practitioners used site relevance as a quality criterion
• an authority metric. The same report found 85.8% use authority metrics, with Ahrefs Domain Rating used by 93.2%. Authority metrics explained covers their limits
• the follow status and the page the link points to
• whether the article is indexed and the brand is named in the body

Cost per link belongs here. Divide the full campaign cost by unique linking domains. In BuzzStream’s 2026 report, 39.2% of practitioners did not know their cost per link. Those who did most often said $300 to $500, while BuzzStream’s own pricing guide, counting unique links without syndication, gives $1,250 to $1,500. Yield benchmarks differ for the same reason. BuzzStream’s guide counts 6 to 7 unique linking domains for a typical campaign. Digitaloft’s compilation cites an average of 42 referring domains per campaign at an average Domain Rating of about 61. The samples and definitions differ, so compare your results with your own past campaigns first. Link building cost and budget has the wider benchmarks.

Layer two: mentions and AI visibility

A campaign that earns coverage without links has still done something. This layer records it.

	PR for links	PR for mentions
Aim	Rankings through authority	AI visibility, branded search and reputation
Main KPI	Unique followed referring domains, relevance, authority	Mention count and quality, mentions beside category terms, AI share of voice
Measured with	Backlink tools	Media monitoring, AI visibility tools, branded search data

Useful measures:

• Mention volume and tier. How many articles named the brand, and in what kind of publication.
• Unlinked mentions. A list to work through with link requests. See unlinked mentions.
• AI mention rate. The share of answers to a fixed set of prompts that name your brand. Answers vary between runs, so only a rate over many runs means anything.
• AI share of voice. Your mentions as a share of all tracked brands for those prompts, which needs a defined competitor set.
• Citation rate. The share of answers that link to your domain. This is separate from mentions.
• Third-party source coverage. Of the pages AI answers cite for your prompts, how many mention you. This is the measure digital PR can act on directly.
• Branded search trend. Branded impressions in Search Console over time, as a proxy for attention that never produced a click.

The AI visibility tools comparison lists trackers, and brand mentions and citations explains the evidence behind these measures, most of which is correlational.

Layer three: commercial outcomes

At Digitaloft’s Digital PR Summit in May 2026, speakers including James Brockbank and Laura Joint argued for three habits: work backwards from commercial goals, isolate the pages PR supports and compare them with control groups, and stop reporting “brand awareness” with no way of measuring it.

In practice:

1. Start from the goal. Name the pages and the search terms the campaign should help, before the idea is chosen.
2. Define the supported group. These are the pages that receive links directly or through internal links from the campaign page.
3. Choose a control group. Similar pages on the same site that the campaign does not touch.
4. Compare over a quarter or more. Track rankings and organic clicks for both groups. A difference between them is the best estimate of the campaign’s effect.
5. Add the direct numbers. Referral visits, and conversions or assisted conversions from the coverage.

A control group is an estimate, not proof. Updates, seasonality and other work on the site all interfere. It is still far better than comparing a page with its own past and crediting PR with everything that changed.

Allow time. In BuzzStream’s 2026 report, 81% of practitioners said first coverage arrives within a week, and 85.2% see measurable results within six months. How long link building takes covers the lag between a link and a ranking change.

A reporting frame

The layers map onto a simple set of KPIs, adapted from the framework we use for link building in general.

Stage	What to report
Input	Budget spent, journalists pitched, campaigns launched
Process	Open and reply rates, coverage secured. A reply rate of about 1% is normal for cold PR pitches in BuzzStream’s platform data
Output	Unique linking domains by relevance and tier, follow share, syndicated copies, mentions, cost per link
Durability	Share of links still live and indexed at 90 and 365 days
Outcome	Supported pages against control, branded search, AI mention rate, referral and assisted conversions

Report output monthly and outcomes quarterly. Team roles and KPIs shows how this fits a wider link programme.

What not to report

• Raw link totals that include syndicated copies.
• Authority averages on their own, with no relevance.
• “Brand awareness” with no measure attached.
• Advertising value or reach estimates presented as results. They describe the outlet, not what your coverage did.
• Everything that improved after the campaign, with no control.

Agree it before the campaign

Most disputes about digital PR results are disputes about definitions. Before work starts, write down what counts as a link, whether nofollow and syndicated links count, which pages are the target, what the control group is and when results will be judged. If you use an agency, put it in the contract. The digital PR tactic page explains why a campaign can still earn nothing, and that risk should be stated too. Terms are defined in the glossary.