Thrive Market leader on why earned media has a longer shelf life than ever 

Alyssa Kluge is director of communications at Thrive Market, where she leads corporate affairs, reputation management and earned media strategy. Kluge built the company’s communications function and oversees thought leadership, public policy, and communications across an evolving digital landscape shaped by AI. She also led Thrive Market’s successful effort to become the first online-only grocer to accept SNAP EBT nationwide. 

Kluge will speak at Ragan’s upcoming virtual AI Communications Conference, where she’ll share how senior communications leaders are measuring AI’s impact across content, channels and teams, and which metrics matter most as AI reshapes communications. 

You built Thrive Market’s communications function from the ground up. What was the biggest challenge that wasn’t obvious in doing that?

The less obvious challenge was learning to sit something out. I think about it as the Bean Soup Theory popularized on TikTok: not every cultural moment, trending topic, or news cycle is meant for your brand. The instinct in comms is to often insert yourself, or be the loudest voice; the discipline in recognizing that selective participation actively builds more credibility than constant noise, and having the confidence to let something pass without being in it.

Your session is focused on moving beyond AI outputs to actual business outcomes. What’s the biggest mistake you’re seeing communicators make when they try to measure AI’s impact today?

The reluctance in moving from ‘did we appear’ paradigm to ‘are we shaping how the category / are we being understood?’ That requires a different set of metrics entirely: LLM visibility across the specific queries your buyers or consumers are actually asking, citation frequency that tells you whether you’re earning the real estate that AI treats as authoritative, narrative share of voice that tracks whether the market is adopting your language, and high-salience mentions that measure the dominance of your placement.

These metrics require more comfort with data than most comms teams have historically used. The communicators who close it fastest will stop asking ‘how do we prove PR’s value’ and start asking ‘what’s our citation share on the three questions our buyers are taking to Chat right now?’ That’s the question that connects comms directly to business outcomes.

Many communications leaders feel pressure to “do something with AI.” How do you separate meaningful adoption from simply checking a box?

The test I use is simple: does Claude / Chat let me spend more time on the work only a human can do, or does it let me produce more stuff faster? Two very different things, and I’ve noticed that tool adoption may signal more output, and there is a confluence with volume for value.

I am a human, and I like being a human with human instinct and judgment. Meaningful adoption, for me, is tapping this never-tiring intern (Claude/Chat) for use on a first pass, catching what I missed (I have an imaginary “board” poke holes before presenting a true draft!), and routine, non-judgement-related tasks.

More importantly, they cannot catch nuance. There is a deeply human part of telling stories and there is no tool that has the judgment or the wherewithal for it. 

How do you think brands should approach visibility in an era where consumers increasingly get answers from AI instead of search engines? 

The metric has evolved. Brands spent years optimizing for reach, like impressions, clip counts, domain authority and none of those numbers tell you whether a model will cite you when a consumer asks a question in your category….citation is the job now.

Depending on who you cite, roughly 90% of what AI models surface comes from non-paid sources. At Thrive Market, most of our citations across major LLMs trace back to earned. That is not a coincidence, and is a direct reflection of treating earned coverage as infrastructure.

What changed our strategy most is how we think about the half-life of a placement. A great earned mention used to fade after 48 hours. Now it has a half-life of every model release. When an outlet writes about your brand, that sentence enters the training corpus of general-purpose models and rides inside answers your future customers are asking right now. The ROI window stopped closing.

One thing shifted immediately. Which outlets you prioritize and when: we rebuilt our tier list from citation data, not reach. For example: a trade outlet will outperform major publications in LLM citations for our category. Your intuition about prestige is potentially wrong, and the data will show you why.

For brands just starting: run five category prompts across ChatGPT, Claude, Gemini, and Perplexity monthly. Log what surfaces and which sources get cited. That data turns a budget conversation that has lived on instinct into one you can actually defend.

You’ve spent your career translating complex issues into stories people can understand. What’s something about AI measurement that you personally changed your mind about over the last year? 

For a long time, I was skeptical about measuring earned media (to a point!). The whole point of PR is being a protector of reputation, living in spaces that you couldn’t control. Slapping an arbitrary impressions number on that felt reductive, and even a little dishonest.

What changed my mind during the past 16 months is realizing that AI measurement doesn’t place a price tag on trust, but is a utility in tracing how a single message travels from an internal idea to an external narrative, and shows where it gains or loses fidelity along the way. In short: it’s a translation problem, which mirrors the work I’ve done throughout my career. The difference now is I can see the translation happening at scale, rather than relying on instinct or my own grading metrics.

The shift for me was from “measurement flattens the craft” to “measurement reveals the craft.” It certainly does not replace judgement about what’s worth saying… more than anything, it signals whether what we said made it through.

What’s something you’ve become unexpectedly passionate about in the last few years? 

Home interiors and design. 

If you could spend an hour talking with anyone, living or dead, who would it be and what would you ask them? 

Andrée de Jongh, also known as the Nightingale (Kristin Hannah’s inspiration for her famed book). While I have hundreds of questions, one I’d be most keen to ask is if her fear ever went away, or if she just learned to move through it. Her legacy reminds me that ordinary people hold extraordinary power and being a good neighbor has no borders.

Don’t miss Alyssa’s insights on measurement at Ragan’s virtual AI Communications Conference on July 22. Register now.

Isis Simpson-Mersha is a conference producer/ reporter for Ragan. Follow her on LinkedIn.