How to measure whether a press release was worth it

Four things can be measured honestly after a press release: how many published URLs exist and still resolve, what each of those pages does with your link, whether any referral traffic arrived, and whether branded search impressions moved. One thing cannot be measured at all, and it is the one every ROI template asks for: attribution to revenue. If a vendor hands you a return figure, ask what the denominator is. In this category it is usually a reach estimate multiplied by an advertising rate the buyer never paid.

The four measurable quantities

1. Live URLs, counted on a date

The base unit. Not "sent to 500+ outlets" — a count of pages that exist, that contain your company name in the body, and that returned a 200 on a date you record. This is the only figure in the whole exercise that is a direct observation rather than an inference.

Two cautions from our own measurement pass. A page can return 200 and still not be your placement: one host in our catalogue serves a real article for a nonexistent ID, and another serves any slug against a numeric key, so "the URL loads" is not the same as "the release is there". And a 403 is not an absence — Benzinga, Street Insider and Digital Journal return 403 to scripted requests whether the page exists or not, which means a deletion on those hosts is only ever discovered by a person opening it. How to run the check yourself takes about ten minutes per URL.

2. Link status, read off each live page

Open each page and read the rel attribute on the link back to your site. There are five outcomes and all five occur: a plain followed link, nofollow, rel="sponsored", a redirect through the wire's own tracker, or no link at all. Google's own documentation is explicit that links marked with these attributes "will generally not be followed", so nofollow and sponsored pass no ranking credit.

In our August 2026 sweep, the pattern was consistent enough to plan around: the better-known the outlet, the worse the link. AP News carried no client link at all on one wire and three followed links on another; openPR published the client's website as unclickable plain text; Digital Journal routed links through a redirect carrying nofollow. Any search-authority case for buying distribution has to survive that table first. Why almost every press link is nofollow explains why this is normal rather than a defect, and why the wire you buy decides it more than the outlet you land on.

3. Referral traffic, if you tagged the link before you sent it

You cannot add tracking after publication. The release body is copied verbatim onto domains you cannot edit, so the campaign parameters have to be in the URL you submit. Google's documentation on campaign parameters covers the mechanics: utm_source, utm_medium and utm_campaign appended to the link in your boilerplate.

Set expectations before you look at the number. Most syndicated copies strip or neutralise the link entirely, and press-release hub pages are not read by many humans. A near-zero referral figure is the normal outcome, and it is evidence about the format rather than about your announcement. What tagging genuinely buys you is separation: a tagged URL appearing on a page tells you that page is a syndicated copy of your paid distribution, and an untagged mention of the same news is independent coverage. That distinction is the difference between paid syndication and earned pickup, and nothing else establishes it cleanly.

4. Branded search impressions, with an honest caveat

Search Console's performance report gives clicks, impressions and click-through rate by query. Note the impressions for queries containing your company name in the fortnight before the release and the fortnight after.

The caveat is not optional: this is a before-and-after observation with no control group. Anything else you did that fortnight — a launch, a conference, direct outreach — sits inside the same window. Report it as a movement you observed, never as an effect the release caused. If the release accompanied a real event, the event is the likelier cause.

The quantity that cannot be measured

Attribution from a press release to revenue. The chain is broken in three separate places, any one of which would be enough.

The link is usually not followed or not present, so there is frequently no click path at all. The same text is published at many URLs the same day, and where duplicates exist Google selects one canonical version to show — the other copies are citations, not ranking assets. And the buyers who see a release rarely convert in the same session, so any last-click model attributes the sale to whatever they searched a week later.

This is why "media value" and advertising-equivalency figures persist: they fill a gap where no measurement exists. A reach estimate multiplied by a notional ad rate is not a return, because you did not buy advertising and nobody has shown the impressions occurred. When a vendor reports "$X in media value", the honest translation is "we counted the audience numbers of every site that republished you and priced them as though you had run ads there".

What about AI answers?

Worth measuring, worth not overselling. Meltwater's tracking found press releases' share of citations in AI search fell from 0.4% to 0.2% between April and May 2026, a halving in one month, and noted that AI systems "increasingly prefer third-party validation over self-published messaging". A press release is self-published messaging on someone else's domain. It can be cited; it is not a reliable route to being cited. What we can and cannot show about press releases in AI answers sets out the evidence in both directions.

The one honest unit: cost per confirmed live URL

Take the invoice, divide by the number of pages that exist and resolve, and you have a figure that is defensible in a budget meeting because every part of it was observed. On our own $599 package, with three placements bought by name, a fully confirmed order is about $200 a page. On the $899 package with seven named outlets it is about $128 a page — and roughly half of those seven cannot be confirmed by an automated check at all, so the honest denominator is the confirmed count, not the purchased count.

That figure is comparable across vendors and resistant to outlet-count inflation. It is also uncomfortable at the top of the market, which is why almost nobody publishes it.

The measurement plan, in order

1. Before you send: tag the link in your boilerplate, and record the current fortnight's branded-search impressions.
2. At 72 hours: collect every URL the distributor reports, open each one, confirm the release is on the page, and record the status code and the date.
3. At the same time: read the rel attribute on each live page and record it per URL.
4. At 14 days: pull referral sessions on the campaign tag, and the branded-search fortnight.
5. Compute cost per confirmed live URL. That is your number.
6. At 30, 90 and 365 days: re-open the URLs. Pages are removed at the publisher's discretion, and we have observed entire archive paths returning 410 Gone.

Step 6 is the one people skip, and it is where the badge on your website quietly starts linking to a dead page.

What we do, and what we will not claim

We are a reseller — we buy distribution from wire partners and mark it up. What we add is steps 2, 3 and 6, done for you: every placement bought by name is opened and checked, each row is labelled by who confirmed it (our own request, a person, or the network's pickup file with no independent confirmation), link status is read off the live page rather than copied from a rate card, and a row that could not be opened is marked rather than counted. What a send receipt confirms and what it does not is the same argument at length, and the sample report shows the format.

We will not give you a return figure. We do not know what your release earned, nobody selling distribution does, and a number invented to fill that gap is worse than the gap.