Earned links and AI citations can come from the same content asset. Kevin Rowe, CEO of PureLinq, presented two-year data at a Search Engine Journal webinar showing how a single piece of original research can generate press placements, organic rankings, and AI Overview citations simultaneously. One client campaign produced more than 1,000 earned citations, with coverage in The Wall Street Journal, Fortune, Axios, Fox Business, and Reuters, according to the session recap published by Search Engine Journal.
What broke traditional link building?
Around 2020, PureLinq nearly closed. Kevin says the agency lost approximately half its client base overnight because link building stopped producing results. He identified three forces behind the collapse. The third, he argued, is the one most SEO teams encounter last: watching click-through rate (the share of searchers who actually click your result) fall in Google Search Console while rankings stay unchanged. Revenue drops without an obvious signal to diagnose.
His response was to stop treating link building as a standalone tactic and rebuild the agency’s model around original research as the central asset. “Deserving is the biggest thing you could ever do,” he said, returning to that line throughout the session.
What data story do journalists actually cover?
Kevin says five specific angles make a dataset worth a journalist’s time, calling that slide the most valuable in the deck. Two of those angles underpinned a student loan study his team built using free federal government data on student debt. The study earned full feature placements in Fox News and Fox Business. “It’s really low hanging fruit and I think almost anybody can do it,” he said, according to the Search Engine Journal recap.
The ceiling on this approach is data quality. “AI slop is the biggest problem in research right now,” Kevin said. “If you send out slop to journalists and they catch you, and your data is wrong, they will probably stop working with you.” A language model can produce a study draft quickly; it cannot independently verify the underlying data or catch methodological errors before a journalist does.
How does a research hub build earned links over time?
One B2B client with national reach and no physical locations was losing search visibility to local competitors. Its existing blog ran on commodity content: how to collect debt, top financing tips. The team deleted it and rebuilt it as a research hub, publishing studies on a recurring schedule.
Month one produced one earned link. The client considered leaving. Kevin persuaded them to stay.
Placements climbed month over month from there. At some point during that stretch, the dynamic inverted: journalists began emailing the team to ask when the next dataset would be ready, rather than the team pitching outward. The Wall Street Journal eventually requested a dataset update directly. “That’s the inbound side of this,” Kevin said.
A second client, an online university operating in a regulated category, took a lighter approach. It kept its existing blog, added a handful of studies alongside it, and placed its own professors in front of journalists and podcast hosts. That generated earned links through direct one-to-one pitching without a full hub rebuild.
Can original research rank when AI Overviews answer the query first?
“We are not in the zero click world. 100% not,” Kevin said at the session. “You can still get a ton of traffic from search organically.” His team’s data supports that: individual studies draw traffic from AI Overviews and several hold position one for commercial keywords, not just informational ones.
One example from the recap: a study that ranked US financial districts by foreclosure risk earned coverage in Axios Seattle, Barron’s, and The Wall Street Journal while also ranking in organic search. A second study, written by a plant science professor at Utah State University, ranks first for its target terms and is cited as the primary data source inside AI Overviews.
On traffic from AI platforms directly: clicks from ChatGPT convert at above-average rates, including on ecommerce. Kevin’s team tracks it as the largest source of direct AI-platform clicks they can measure. Volume, however, “is not even close to Google search with AI Overviews,” he noted.
What this means for AI-search visibility
The core insight from Kevin’s data is straightforward: a press mention and an AI citation are often the same outcome from the same asset. They are not separate distribution problems that require separate budgets.
Here is why that matters practically. When a journalist at a high-authority outlet cites a dataset, the resulting article becomes part of the public web. AI systems trained on that web, and retrieval systems (the mechanisms that pull live web content to answer queries) that index it, treat editorial citations as credibility signals. A study covered in The Wall Street Journal is not just a link. It enters training data, search indexes, and the AI Overview ranking calculation at the same time.
Kevin’s timeline data also shows that this compounding is real and slow: month one produced one link. The inbound journalist relationship took months to develop. The plant science professor’s study is worth examining closely because academic authorship is one of the clearest credibility shortcuts currently available. Editors and AI retrieval systems both weight it heavily. Most brands cannot produce that on demand, which means the cadence and expert-sourcing strategy matter as much as the research angle itself.
One gap the source does not fully address: the difference between AI citation volume and AI-driven traffic. Kevin’s team found ChatGPT converts well but sends low click volumes compared to AI Overviews inside Google. That distinction matters for how brands should prioritize. Being cited inside AI-generated answers and receiving measurable referral traffic from AI are currently different optimization targets. Conflating them leads to misread results and early exits, which is exactly what Kevin’s B2B client nearly did after month one.
Hingewise’s read: the research-led model Kevin describes is increasingly the baseline expectation for brands that want to appear consistently inside AI-generated answers. But the compounding timeline he documented, and the role of academic authorship in the fastest-moving case, point to a 6-to-12-month strategy rather than a sprint. The risk is not that the approach fails. The risk is that brands measure it on a one-month window and conclude it failed before the compounding has started.
Before launching a research campaign: a short checklist
- Audit your last quarter of published content and flag anything a competitor could publish word for word without changing a line.
- Identify at least one free, publicly available dataset relevant to your industry (federal databases, municipal records, industry association data).
- Map which data story angles apply to your sector before commissioning a study.
- Define a publishing cadence you can realistically sustain, not just launch once.
- Identify a credentialed expert (academic, licensed professional) who could co-author or formally review the methodology before pitching.
- Build a journalist target list before the study is finished, not after.
- Establish an AI citation baseline for your brand before the first study goes live so you have a reference point to measure against.
The full session, including all five data story angles, placement and traffic numbers from both case studies, and the image and headline tactics Kevin’s team found lift click-through rate inside AI Overviews, is available on demand via Search Engine Journal.
Sources
This article was drafted with AI assistance from the sources cited above and checked against them before publishing. The "What this means for AI-search visibility" section is Hingewise's editorial analysis, built on the frameworks and positions we use in our GEO work. How we work: Editorial policy.