How publishers can stop optimizing for an algorithm they don’t control, and start owning the relationship they do.
For twenty years, publishers built their growth engines on a trade they didn’t design and couldn’t renegotiate: write for the algorithm, and it sends you readers. That trade is ending. Not gradually. Measurably, this year.
Search referrals to small publishers are down 60%. Midsize publishers are down 47%. Even the largest, most SEO-sophisticated newsrooms are down 22%.
Readers haven’t lost interest in the news. The path they take to find it has become unpredictable, and increasingly, it doesn’t run through you at all. AI answers the question before the click happens. The publisher that wrote the definitive piece may inform the answer, but never get the pageview.
This is being covered as a traffic story, but it isn’t one. It’s a relationship story, and most publishers are solving the wrong problem.
The instinct to fight the algorithm is the wrong instinct
The reflexive response to a referral decline is to optimize harder for the channel that’s declining – chase the next SEO update, format content for AI Overviews, hope for a licensing deal. Some of that is worth doing, but none of it is a strategy, because it leaves the fundamental exposure in place: a reader you reach through someone else’s platform is a reader you can lose to someone else’s platform. You don’t own that relationship. You’re renting attention from a landlord who just changed the terms without telling you.
The publishers who come out ahead of this shift are the ones who convert every reader they do reach – however they arrive – into someone they can reach again, on their own terms. Winning back search rank was never the play.
The evidence that this still works is not thin. Average open rates on regular newsletters sit at 37.7%. Nearly three-quarters of consumers check a new text message within five minutes. When discovery fragments, the direct channel doesn’t decay – it becomes the more reliable path back to a reader you’d otherwise have to rediscover from zero every time.
The real gap is data, not traffic
Here’s the harder truth underneath the traffic numbers: most publishers can’t act on this shift even when they see it coming, because the reader who lands on a story from a search result, a social share, or a syndication feed is, to their systems, a stranger. No unified profile. No record of what they came for. No way to tell the newsroom’s CMS what marketing knows, or tell marketing what the newsroom just published.
Reach without the ability to act on it is just a traffic report. The ability to act without reach is just a guess about who’s out there. Publishers need both at once.
A framework for owning the relationship: Recognize, Connect, Reach
Three moves close that gap, and none of them require ripping out the CMS or the editorial workflow your newsroom already trusts.
1. Recognize readers before they register.
A visitor who’s clicked through five stories on a single topic has told you something valuable, even if they’ve given you no personal information at all. Capturing that signal – topics viewed, visit frequency and recency – turns an anonymous session into a profile you can act on the moment that reader is ready to subscribe, register, or opt into a newsletter, instead of losing the signal the second they close the tab. This is what Iterable’s Unknown User Activation is built for: defining the criteria that matter to your business, and letting the platform build the profile automatically as visitors meet them – no engineering sprint required to change the rules later. Morning Brew used exactly this kind of unification – pulling custom events and profile data into a single reader view – and saved $100k in acquisition costs they’d previously spent re-targeting readers they already had
2. Connect content to audience data without adding newsroom work (TL;DR: Your CMS stays your CMS)
Editorial should keep publishing exactly where it publishes today. The distribution layer is where the leverage is: a publish event from WordPress, Arc XP, or any other CMS should be able to trigger a journey that reads the story’s metadata alongside what you know about the reader, then branches automatically into the channel and message most likely to land. Email for one reader, push for another, SMS for a reader who’s shown they respond fastest there. Built in Iterable’s Studio, a Journey does exactly this: one story, published once, personalized on delivery.
3. Reach readers when attention is a five-minute window, not a five-hour one.
Breaking news doesn’t wait for a send queue, and a spike in audience volume shouldn’t force a choice between speed and relevance. This is where AI-driven decisioning earns its place – not to write the story, but to decide, in real time, which channel a given reader is actually likely to open right now, and to hold that reasoning at the volume a real breaking-news moment demands. Iterable’s Nova Intelligence is built for exactly this: removing the guesswork from when, where, and how often a reader gets reached, so editorial and marketing both have visibility into what actually shipped.
What this looks like when it works
This isn’t theoretical.
The Athletic applied exactly the “connect content to audience data” move above, replacing generic sends with behavior-based personalization, and saw subscriber click rates grow 127%.
Morning Brew applied the “recognize readers before they register” move: using dynamic newsletter content tied to reader interest, they added 15,000 new subscribers in just six months.
Across the broader publishing customer base, the same discipline shows up at scale: affiliate revenue up 20% through cross-channel expansion, email creation time cut 75%, and time-to-inbox on high-volume sends reduced 96%.
As Ken Pecca, Director of CRM, put it: choosing a platform came down to whether sophisticated segmentation and journey capabilities could actually scale personalized campaigns without breaking” – precisely the test a fragmenting discovery landscape now puts to every publisher’s stack.
The bottom line
Content discovery isn’t broken. It’s being redistributed away from the platforms publishers used to depend on, and toward the direct relationships publishers actually own. The publishers who treat that as a traffic problem will spend the next two years chasing an algorithm that keeps moving. The ones who treat it as a data and relationship problem – recognizing readers earlier, connecting content to what they know, and reaching people when attention is real – will spend that same time building the asset no algorithm update can take away: readers who come back because you reach them, not because a search result did.