Digital PR is the work of getting cited by sources you don't control: online newspapers, trade magazines, blogs, podcasts, comparison sites, creators. The aim is for them to talk about you on their own initiative, with a link to your site when it goes well and with just the brand name when it goes less well.
Until a couple of years ago the value of a campaign was counted almost entirely in backlinks. Now there's a second sum to do, because ChatGPT, Gemini and Google's AI Overviews build their answers largely from third-party sources. Analysing more than a million links cited by AI platforms, Muck Rack found that 95% come from unpaid media and 27% from journalistic content (July 2025, figures not split by country).
The guides on Google UK's first page, checked on 10 October 2026, are mostly agency pages explaining what digital PR is. Here you'll find the international figures on what actually gets cited, a test we ran on 5 UK searches, what Google says about press releases and sponsored articles, and what a campaign costs.
What digital PR is, in practice
It's the online version of media relations, with a difference in method. Instead of sending out company news and hoping someone picks it up, you prepare something a journalist has a reason to use: new data, a survey, a ranking, an expert's comment on a story that broke that morning.
The recurring activities are these:
- research and proprietary data offered to newsrooms;
- press releases;
- expert comments in response to journalists' requests;
- newsjacking, that is, hooking onto the news of the day with a figure or an opinion;
- creative campaigns designed to be picked up (known as hero campaigns);
- relationships with creators, newsletters and trade podcasts.
The first two are by far the most common. In BuzzStream's State of Digital PR 2026, a survey of more than 150 practitioners, mostly in the US and the UK, 95.9% of respondents use data-led content and 79.1% use press releases (up from 68.8% the year before).
Digital PR, traditional PR and link building: the differences
The three are often sold together, but they serve different goals and carry different risks.
| Traditional PR | Digital PR | Link building | |
|---|---|---|---|
| Main goal | Reputation and coverage in print, radio and TV | Mentions and links from online editorial sources | Links to specific pages on the site |
| Who decides whether to publish | The journalist | The journalist or the publisher | Often whoever is selling the space |
| What is measured | Coverage, reach | Editorial links, mentions, citations in AI answers, brand searches | Number of links, anchor text, domain authority |
| Risk with Google | None | Low, if the links are editorial | High, if the links are paid and not qualified |
In the projects we run, the two online activities have separate rules. In link building the target page and the anchor text are chosen around the keywords with the most headroom. In digital PR we leave the brand name or the bare homepage URL and aim for the mention, because a press release with commercial anchors republished word for word on twenty sites is exactly the signal Google treats as manipulation. Both sit within off-page SEO, but they're judged on different metrics.
Why AI answers have changed the sums
The first sign came from Ahrefs in May 2025. Across 75,000 brands, visibility in AI Overviews correlates with brand mentions on the web (Spearman coefficient 0.664) far more than with linking domains (0.295) or the number of backlinks (0.218). A name that appears in an article counts even without a link.
In 2026 BuzzStream published two analyses of the same dataset: 3,600 prompts across 10 sectors, more than 4 million citations collected from ChatGPT, Gemini, AI Mode and AI Overviews in the week from 27 January. Earned sources, meaning third-party pages that talk about a brand, make up 47.7% of all citations and rise to 93.75% for evaluative questions, the "which is the best" kind that come just before a purchase. In the same analysis by prompt type no classic SEO metric correlates with citations: Ahrefs' Domain Rating has a coefficient of -0.111, Moz's Domain Authority -0.128.
The second analysis, on news publications, is the one that should give pause to anyone buying placements in bulk. News titles account for 14.09% of all citations, but syndicated press releases are just 0.32% of news citations and newswires 0.21% of the whole dataset.
| Study | Sample | Finding | Market |
|---|---|---|---|
| Ahrefs, May 2025 | 75,000 brands | Brand mentions: 0.664 correlation with AI Overview visibility; backlinks: 0.218 | Global, mostly English |
| Muck Rack, July 2025 | More than 1 million cited links | 95% from unpaid media, 27% from journalistic content (49% for questions about current events) | Not stated by country |
| BuzzStream, April 2026 | 3,600 prompts, more than 4 million citations | Earned sources: 47.7% of the total, 93.75% for evaluative questions | Prompts in English, market not stated |
| BuzzStream, March 2026 | Same dataset | News titles: 14.09% of citations; syndicated press releases: 0.32% of news citations | Prompts in English, market not stated |
There's an obvious limit: all of these studies use English-language prompts and don't say which market they measure. If you work on this, there's a wider picture in our guide to AI citations and mentions. For the UK specifically there's no comparable public dataset, so we ran a test.
The UK test: who AI Overviews cite on 5 commercial searches
On 10 October 2026 we queried Google UK (desktop, English, one reading per query, via DataForSEO) with 5 searches of the "best + category" kind: best current account, best car insurance, best energy supplier, best broadband deals and best CRM for small business. We ran the same test in Italy on 12 searches; for the UK we kept to five categories where comparison sites, consumer bodies and brands all compete.
The AI Overview appeared on all 5 searches. In 4 of them the sources were readable: current account, energy supplier, broadband and CRM (the car insurance one loaded separately and its sources weren't captured). In total, 32 citations, counted by domain with no duplicates within the same answer.
| Type of source cited | Citations | Share | Examples |
|---|---|---|---|
| Comparison and affiliate sites | 12 | 37.5% | moneysavingexpert.com (in all three consumer answers), moneysupermarket.com, uswitch.com, comparethemarket.com, clearscore.com, startups.co.uk |
| Company sites and blogs | 9 | 28.1% | barclays.co.uk, hsbc.co.uk, firstdirect.com, octopus.energy, vodafone.co.uk, slack.com |
| Video and user-generated content | 4 | 12.5% | youtube.com |
| Consumer and business organisations | 4 | 12.5% | which.co.uk (twice), citizensadvice.org.uk, uschamber.com |
| Publishers and magazines | 3 | 9.4% | forbes.com (Forbes Advisor), pcmag.com, uk.pcmag.com |
Source: Visilay check of 10 October 2026 on 5 queries, 4 AI Overviews with readable sources.
Sources outside the companies themselves are 23 out of 32, or 72%: a share in line with the American figures. The mix, though, is different. On searches that come just before a purchase, the third party that gets cited in the UK is the comparison site, not the newspaper. MoneySavingExpert appears in all three consumer answers, national newspapers in none. The Telegraph and This is Money ranked in the organic results for two of the queries, but the AI Overviews didn't use them. Which? and Citizens Advice, the UK's consumer bodies, were cited on energy and broadband. In Italy we found the same pattern, with Facile.it in the role of MoneySavingExpert.
The sample is small and AI Overviews change from one search to the next, so treat it as a pointer to repeat, not as a statistic. The pointer is useful all the same: in sectors with comparison sites and consumer bodies (banking, insurance, energy, broadband) a digital PR plan that aims only at national titles leaves out exactly the sources the AI uses to answer people who are about to buy. One more piece of context: according to the Reuters Institute Digital News Report 2026, 30% of people in the UK trust the news, five points fewer than a year earlier and below the global average of 37%.
Press releases and sponsored articles: what Google says
Google's spam policies, updated on 28 August 2026, list two practices among link schemes that are often sold as digital PR: paying for articles that contain links passing ranking credit, and putting links with optimised anchors into articles, guest posts or press releases distributed on other sites.
A sponsored article on a news site isn't banned. It's advertising, and as such the link has to be qualified: the documentation on outbound links gives rel="sponsored" as the preferred value for paid links, with nofollow as an alternative. Placements sold off a rate card on news sites and blogs are easy to find, and the problem starts when those links are passed off as editorial. We've set out what a site really risks in our guides to Google penalties and white hat SEO.
For press releases we follow two practical rules. In the version that gets picked up unchanged by several sites we put only the brand name or the bare URL, with no commercial anchors. If the content has value of its own, a piece of market research for example, we write a different, unpublished version and offer it as an exclusive to a single title. The BuzzStream figure of 0.32% says that a press release copied onto dozens of sites barely exists for AI: the article a journalist wrote starting from your data does.
Then there's a source of mentions that companies get anyway and hardly anyone makes use of: trade fairs. Exhibitor catalogues, trade press articles and organisers' announcements talk about the brand but rarely link to the right page. We've written about it in the case of an Italian manufacturer that went from position 88 to 2.
How to build a digital PR campaign
The starting point is knowing how newsrooms work. Cision's 2026 State of the Media, based on 1,899 journalists in 19 markets surveyed in January and February 2026, says that 66% use material supplied by press offices to find ideas, but 72% consider fewer than one pitch in four relevant and 53% object to pitches written with AI. What they ask for is original research, access to experts and exclusives.
- Start from the questions, not the press release. List the searches and the questions to AI assistants you want to appear in. How a model breaks a request down into several searches is something we tested in the article on query fan-out.
- Look at who gets cited today. For each question, note the sources in AI Overviews and in ChatGPT's answers. News titles, comparison sites, consumer bodies, specialist magazines: that's your list of targets, and it's different for every sector.
- Find data only you have. Aggregated sales or usage data, a customer survey, a price comparison collected by hand. The data has to stand up without your name attached.
- Publish the full version on your own site. A page with the method, tables and a date gives the journalist something to link to and gives AI a primary source to cite.
- Write the pitch for one person. A few lines, the figure first, the reason it matters to their readers now.
- Measure after months, not days. In the State of Digital PR 2026, 81% of practitioners get the first piece of coverage within a week of the pitch, but 51.4% say it takes 3 to 6 months to see measurable results.
How to measure digital PR
Among the practitioners surveyed by BuzzStream, the most used KPIs are quality links (85.1%), total mentions (72.3%) and total links (70.9%). Citations in AI answers are already at 55.4%, and 39.2% don't know their own cost per link. In our view there are five metrics worth keeping:
- editorial links earned, kept separate from paid ones;
- unlinked brand mentions, which count for AI even though they pass no SEO value;
- citation rate on a fixed sample of questions, as we explain in the guide to AI visibility;
- the trend in brand searches in Search Console, the most direct effect of good coverage (we cover it in the article on brand SEO);
- referral traffic from coverage, useful but almost always low.
On traffic it pays to be realistic. A Pew Research study of 900 US adults, with 68,879 searches tracked in March 2025, found a click on a link inside the AI summary in 1% of visits. Being cited gets you named in the answer; it does much less for visits.
How much digital PR costs
The most detailed reference is a BuzzStream survey of about 70 agencies and freelancers, 54% of them British and 23% American. The average retainer is $5,458 a month and the average cost per link $597, with big differences depending on the type of work.
| Type of work | Average cost per link | Average monthly contract |
|---|---|---|
| Guest posts | $150 | $2,500 |
| Responding to journalists' requests | $400 | $4,000 |
| Reactive and proactive PR | $625 | $6,000 |
| Creative campaigns (hero content) | $800 | $8,000 |
Source: BuzzStream, sample of agencies and freelancers mostly in the UK and the US, figures published in US dollars. In the UK the average cost per link is $818 (about £618 at the Bank of England rate of 8 October 2026, $1.3226 = £1), in the US $350. The State of Digital PR 2026 confirms the order of magnitude: 60% of budgets are under $10,000 a month and the most common cost-per-link band is $300 to $500.
The UK figure comes from the British part of a sample of about 70 respondents, so it's an indication rather than a market price. The practical advice is to ask whoever quotes you how they count results: a price list based on a guaranteed number of placements usually describes advertising space, not media relations.
Where people usually go wrong at the start
The most common mistake we see is buying digital PR as a package of placements on national titles, chosen for the prestige of the name. Our tests in the UK and in Italy say that on the questions that bring in customers, AI mostly cites comparison sites, consumer bodies and specialist sites, sources that no placement price list includes. If you plan from the queries, look at which sources actually get cited and build usable data for them, you spend less and end up in the right answers. It's the method we use in our SEO for AI and link building projects: first the map of sources, then the content.
Digital PR: frequently asked questions
It's the set of activities through which a company gets cited by online sources it doesn't control: news sites, trade magazines, blogs, comparison sites, podcasts and creators. The work is based on data, expert comments and stories journalists have a reason to use, and the result is mentions and editorial links.
Link building aims to get links to specific pages, with anchors chosen around keywords. Digital PR aims for mentions and editorial coverage: the link is a side effect, not the condition. That's why press releases should use the brand name or the bare URL rather than commercial anchors.
Yes. It brings editorial links from authoritative domains and increases brand searches. It also counts for AI answers: across 75,000 brands Ahrefs measured a 0.664 correlation between web mentions and visibility in AI Overviews, against 0.218 for the number of backlinks.
It depends how you use them. A release picked up word for word by dozens of sites is worth little: in BuzzStream's 2026 study syndicated press releases are 0.32% of news citations in AI answers. It works if it becomes the starting point for an article written by a journalist, ideally with an exclusive piece of data.
The first coverage often arrives within a week of the pitch, but the effect on rankings, brand searches and AI citations shows after months. In BuzzStream's State of Digital PR 2026, 51.4% of practitioners say between 3 and 6 months.
They identify the sources that matter in a sector, prepare data and stories newsrooms can use, write pitches, manage relationships with journalists and creators, and measure links, mentions and citations in AI answers. Compared with a traditional press officer, they work closer to SEO and data analysis.
