How to Measure Media Coverage and the Results of a Digital PR Campaign

Measure what you can verify

If you want to measure PR coverage, begin with an uncomfortable fact: much of what makes a campaign valuable is hard to measure exactly. People read an article, remember a name and search for it a month later. No dashboard captures that journey from end to end. What you can do is collect a small set of indicators that are honest, repeatable and tied to things you can verify.

This article sets out such a set for a digital PR campaign, meaning the publication of press releases and sponsored articles on third-party sites, the kind of work our press release publishing service supports. It is deliberately modest. It does not use invented benchmarks, and it does not claim that any activity produces a given number of visits. It also assumes that you need to report to a manager or a client and want every figure to stand up to questions.

The indicators are arranged in four levels, from the most certain to the least. The higher the level, the less you control it: output, visibility, effect on your site and effect on your brand. If you are still deciding which kind of text to publish, our article on the choice between a press release or an advertorial explains how the objective changes what you measure.

Level 1: a publication log

The foundation is a log. Every publication gets a row, filled in on the day it goes live. Our own reports list the URLs that went live, and it helps to keep your own version too, because it lets you compare publications across months and agencies.

A log is not a result, but it is the denominator for every ratio you will calculate later. “We were published fifteen times” means little. “Fifteen publications across eight categories, thirteen indexed within two weeks” means something.

Level 2: is it live and can people find it?

Check each URL a few days after publication, and again after a month. You are looking for three things: the page opens, the text is as agreed and the link works. Then check whether the page is indexed. You do not have Search Console access to a publisher’s site, so use public methods: search for the exact title in quotation marks, search for a distinctive sentence, or search for the publication name and a key phrase. These checks are imperfect, but together they give a reasonable picture. The site: operator can help, though its results are not a complete count of what is indexed.

If the page is not indexed after a reasonable time, do not assume a fault. New pages on large sites are often found quickly, while pages on small or rarely updated sites can take longer. Our article on getting a new article indexed explains what helps and what is out of your hands.

Also note whether the page is linked from the publisher’s own navigation, a category page or the home page. A piece that sits several clicks deep and is linked from nowhere is harder to discover.

Level 3: what happens on your site?

Here Google Search Console becomes useful, because it covers your site, which you can verify. The tool is available at search.google.com/search-console. Review these items monthly:

• Links report. It shows the sites that link to you, the most linked pages and the common anchor text. Check that your new publications appear, bearing in mind that the report is not instant and may not be complete.
• Performance report for the target page. Look at impressions and clicks for the page that the article links to, compared with the months before. Changes can have many causes, so record them without claiming credit.
• Referral traffic in your analytics. Look at visits arriving from the publisher’s domain. Even a small number of qualified visits can be more valuable than a large number of casual ones, if you also look at what visitors did next.
• Conversions, if tracked. Enquiries, calls or sign-ups that can be traced to the referring page. Many will not be traceable, and that is normal.

Be realistic about attribution. A reader may see your article, close the tab and search for your name a week later. That visit will appear as organic or direct, not as a referral. Treat the referral number as a lower bound.

Level 4: signs of brand awareness

The softest level is also where PR is most likely to have an effect, even though it is the hardest to prove. Look for:

• Branded searches. In the Performance report, filter queries containing your brand name and watch impressions and clicks over several months.
• New mentions. Set up a simple alert for your brand name and note mentions you did not commission, linked or unlinked.
• Direct enquiries that mention an article. Ask new contacts how they heard about you, and record the answers.
• Appearances in AI answers. Test a fixed list of questions at intervals and note whether your brand is described correctly. The article on brand mentions and backlinks in AI answers shows how.

None of these is precise, and several move for reasons unrelated to PR. The value lies in the direction over time, read next to the log.

Metrics to treat with caution

Some figures are easy to obtain and look impressive. Use them only as context.

Domain authority of the publisher

It helps you choose sites, but it is a third-party estimate for the site, not the outcome of your campaign. If you need a refresher on what it means, see what DA and PA mean.

Claimed readership or reach

Estimates of visitors for a site are modelled, and they say nothing about how many people read your piece. Unless a publisher provides page-level data, do not report a reach figure for your article.

Number of articles

Volume is an output, not a result. Ten pieces in the wrong places can do less than three in the right ones.

Advertising value equivalent

Converting coverage into the cost of buying the same space has been widely criticised in the PR industry as misleading, because it equates editorial attention with advertising space. Avoid it.

A simple monthly report

A one-page report is enough. A sensible order is as follows:

1. What we published: the log extract for the month.
2. Are the pieces live and indexed: counts and exceptions.
3. What happened on our site: referral visits, target page performance, links in the report.
4. Brand signals: branded searches, new mentions, assistant test results.
5. What we learned: two or three sentences, including what did not work.
6. What we will change next month.

Compare quarters, not weeks. Seasonality, news events and algorithm changes will all shift your numbers, and a single week is noise. When results look good, say what you did, without promising that it will repeat. When they look bad, say what you will check.

Reading the numbers: a hypothetical month

Numbers only help if you know how to read them. Imagine a fictional bicycle workshop in Oradea that publishes one press release and two advertorials in a month. All figures below are invented for illustration.

• Log: three publications on three different sites, all live on the planned day, all with the agreed link.
• Indexed: two pages found by an exact-title search within a week; the third not found after three weeks.
• Site: the Links report lists two of the three new referring pages. Referral visits are small, but two visitors ask for a quote.
• Brand: branded searches rise slightly, but the workshop also ran a local event that month.

What can the workshop conclude? First, the output was delivered. Second, one page needs follow-up: check whether it is blocked, whether it is linked from anywhere on its own site, and wait a little longer before drawing a conclusion. Third, the enquiries are a good sign, but two quotes are too few to compare publication types. Fourth, the rise in branded searches cannot be credited to PR alone, because the event is a competing explanation.

That is a perfectly good report. It claims what it can support, it flags what it cannot, and it lists an action. A report that said “PR increased brand awareness by a large percentage” would sound better and be worth less. If the workshop repeats the same routine for several months, patterns will emerge, such as which kind of site tends to send visitors who ask for quotes. Those patterns, not a single month, should guide the next budget decision.

Looking at the shape of the campaign

Besides outcomes, review the composition of the campaign itself. Are the sites varied by category? Is the mix between national news, local news and niche publications sensible? Is one domain extension or one group of related domains overused? Our campaign X-ray tool looks at a campaign you assemble from the portfolio and highlights its balance, which makes this check quicker. It describes the campaign; it cannot predict its results.

For a store, the measurement logic is similar, but the target pages are categories and the seasonal pattern is stronger; see advertorials for online stores for the details of that case. And if your campaign is based on press releases, the general process is described in our press release guide.

Next step

Create the log first, even if it is only a spreadsheet with seven columns. Then pick three indicators from levels 2 to 4 that you can genuinely follow every month, and resist the urge to track twenty. A small honest report beats a large impressive one.

If you would like to publish a first press release and build the log from the start, our press release publishing service delivers a report with the URLs of the published pieces, which can serve as the first rows of your log.

Frequently asked questions