The SEO Playbook Is Broken: How to Win in the Age of AI Search

Introducing Allyson Cochran and Silverback Strategies

Tim (01:03): Welcome to Clutch Conversations, where we unpack real decisions business leaders face on a daily basis. I'm Tim Condon, CRO at Clutch. Marketing is changing from discoverability to content creation. Marketing leaders are no longer building strategies the way they used to, and the shift isn't over. AI search has changed what success looks like, and the agencies that get it are reaping the benefits. Joining me to discuss how marketers should plan moving forward is Allyson Cochran, the CEO of Silverback Strategies. Allyson, how are you? Tell us a little more about Silverback Strategies and your mission.

Looking for an SEO agency? Compare our list of top SEO companies near you.

Allyson (01:37): Thanks for having me, Tim. We're a performance marketing agency here in the DC area, and we're really focused on one north star with our clients. Our core services are media, SEO, and AI search — but the single north star is driving profitable growth. On both sides of the fence, with media and SEO, we've gotten hung up over the years on focusing on what's measurable, looking at marketing metrics going up and to the right and saying "job well done." In reality, if the efforts driving those vanity metrics don't translate into business outcomes, they're not worth it. So we're relentlessly focused on achieving business targets and mapping our marketing programs toward them, supporting mid-market B2C brands that are spending at least six figures on performance marketing to hit those goals.

When the Old SEO Playbook Broke

Tim (02:34): We're going to talk about a meaty topic today: the shift from SEO to AI search. Talk to me about when you and Silverback realized the old SEO playbook wasn't going to be enough. What did you see in the data? How long did it take your clients to realize it?

Allyson (03:03): We can all probably remember it was spring 2023 when ChatGPT came out. AI Overviews — I like to call them the artist formerly known as Search Generative Experience, but they rebranded — came out around the same time. People were shocked at how good it was, and it's even better now at producing written content. That was the beginning. But when we first really started to feel it — particularly on upper-funnel informational queries, how-to content no longer driving meaningful clicks to our clients' websites — was around mid-to-late Q1 2024. That's when the traffic on those high-funnel, lower-intent queries, the ones that are very easy for AI to summarize and answer, really dropped off.

(04:00): So we saw the bifurcation — the decoupling of rankings and traffic — at that time. And it's only spiraled as AI has dramatically changed the search results page and the platforms we use since then.

The Attribution Problem: Why Clicks Don't Equal Business

Tim (04:15): Measurement has changed a lot. We see it here at Clutch — showing up in AI search doesn't translate to clicks the same way. When you work with your clients, how do you manage expectations on what they can expect, and how does that affect everything from client satisfaction to how you charge them?

Allyson (04:43): That's a great question. We have to take a step back, because as a performance marketing industry, we've really done clients a disservice over the last 20 years by getting everybody addicted to click-based attribution — traffic and clicks to a website. The same can be said on the paid side: platforms like Google Ads or Meta take credit for leads, revenue, and sales they were probably just last in line for, versus actually having an influence on driving that interest to your brand.

(05:24): This has been a long time coming, and AI search just put a spotlight on the fact that we've been focusing on rankings and traffic and saying, "We're doing a great job for you" — but what did that mean for your business? At Silverback, rankings and traffic have always been a signal. In the AI environment they're still a signal for certain types of intent, but they're only a signal, not the end-goal north star. We want to understand what SEO as a blanket term — which now means a lot more across platforms — has historically contributed to bottom-line performance for our clients. Then we back into target growth from there on an organic revenue basis.

(06:19): As LLMs become a natural part of everyday life, we're looking at not just organic-search-attributed traffic in Google Analytics, but increases in direct visits and AI referral traffic. ChatGPT, for example, is relatively good at tracking and using UTM parameters to understand when it's referring traffic. The others are a little further behind. But we're widening the net of what's considered organic, tying it to organic revenue, and using traffic, rankings, and prompt visibility only as leading indicators of the revenue that should follow.

Tim (07:01): That makes a lot of sense. For better or worse, you've made the job harder.

Allyson (07:06): Absolutely. And to be clear — no one has figured it out. It's changing daily as new platforms and new tracking capabilities come out.

How AI Changed Content Strategy

Tim (07:15): How does that affect your content strategy? In the old world it was: create something, get backlinks, game over. Now you can't rely on that exclusively as a measurement. What's changed?

Allyson (07:36): To speak plainly, a lot of people have strong opinions on using AI to generate content or creative. On the written-content side, it's our opinion that it's safe to use AI to generate content — but you have to have guardrails. You need a human intervening to ensure quality and accurate inputs, so the output isn't just crap. So we use AI tools alongside writers, editors, and subject matter experts who put together the briefs. In my opinion, the biggest moat in this AI environment is your data. As an agency, having performance data across clients in different industries that we can segment to understand performance trends — that's value AI doesn't inherently have.

(08:34): AI has to be built and made smart on top of your data. Having research studies, test and performance data, and a honed brand voice, and using that to influence how we produce content, is what wins. On the types of content we're producing, you have to look at what search even is now. It's no longer just find a keyword, look at search volume in a tool, decide it's high-volume and relevant, produce a blog or service page, follow best practices, and hope it ranks and drives traffic. That playbook is broken.

(09:13): There's an AI Overview on upwards of 50 to 60% of searches at this point — not just informational queries on Google. People are using vastly different channels depending on your ICP. A home services client targeting homeowners uses different channels than a B2B SaaS provider. So we invest deeply in understanding who your customer is first and where they spend their time. Then, in parallel, we research how they research your business, your products, or the challenges you solve. From there we reverse-engineer what those channels use as sources to derive authority. It's a fun reverse-engineering project for our SEO and AI search team, tailored to the unique audiences and businesses we work with.

Tim (10:16): Are you seeing any surprises in the type or source of content that's showing up in AI answers?

Allyson (10:28): Yes, and it's getting better. It's been a bit like the Wild West — and still kind of is. If you're optimizing for citations or mentions on a platform like ChatGPT or Claude, in my opinion the mention is all that matters. Citations can be easier to come by; the mention is different, and we're seeing a different playbook to get cited versus to get mentioned. Six or eight months ago, you could put up comparative content writing about yourself and your competitors as a listicle, and that would work really well and quickly on ChatGPT — not necessarily on Google. But we're finally at the end of that era of quick-win tactics, and getting back to an era where you have to invest in being an authority around a topic, and expand that authority beyond just your website.

(11:32): The surprising part is that those tactics lasted as long as they did. What's fascinating now is how much a channel like YouTube or Reddit is cited. YouTube is in about 30% of Google AI Overview citations. So you really need to expand the breadth of your content — four blogs a month and optimizing your service or product pages just isn't going to cut it.

Mentions vs. Citations: What Actually Matters in AI Search

Tim (11:57): You said you feel the mention is more important than the citation. Let's make sure the audience understands the difference — then I'd love to hear more about that opinion.

Allyson (12:09): Picture going to ChatGPT and typing any prompt — "What are the best HVAC repair companies in XYZ city?" It answers in the interface with a text-based response, which can take a few forms: a table, a couple of bullets, or "If you're looking for this type of provider, here are three; if you're looking for that type, here are another three." But there's also a sidebar showing where it's deriving those companies from. Those are the citations, the sources. They're not necessarily equivalent to what's mentioned in the answer. I'd be curious how often you click into the sources, Tim — it's far less frequent for most people. Practitioners click in to see them, but real consumers who don't live and breathe digital marketing live in the answer. Being part of that recommendation — the mention — is where the value is.

Tim (13:11): I'm a highly skeptical searcher, so I might not be typical — I click them all the time.

Allyson (13:11): Definitely atypical.

Tim (13:11): How much of winning AI search is a content problem versus a technical and structural one? How important are schema, entity clarity, extractability, and site architecture now — more important, less important?

Allyson (13:42): Entity clarity — you've got to love the SEO buzzwords. Google came out a couple months ago and said, essentially, schema doesn't matter, don't worry about it anymore. In reality, there are a lot of things Google says, and then what they do is very different. We feel strongly that schema — structured data that marks up important information on your website so it's easily extractable by a search engine, like your business name, hours, or menu if you're a restaurant — is genuinely useful. That easy parsability and extractability is really critical for Google, even if they say it's not, as well as for AI search platforms like ChatGPT and Claude.

(14:41): On the technical side, SEOs have historically tried to achieve perfection — every light green on the dashboard. To a degree that's true, but if you're mostly solid technically and you have an excellent content strategy, that will outdo an excellent technical foundation with a so-so content strategy. So we prioritize technical optimizations insofar as they help content visibility, and we put all our calories into executing content strategies on-site and off-page to make sure our brand is mentioned appropriately in AI search.

How to Measure Visibility When AI Answers Keep Changing

Tim (15:26): AI answers are non-deterministic — you and I put in the same prompt and get different answers, and we might come back in a week and get different answers again. How do you make sure your visibility measurements are fairly objective and not cherry-picked or too focused on one source of answers?

Allyson (16:00): You're hitting on a couple of nuances. One is that the citations used to answer the same prompt can vary day to day. The other is personalization — if you search something and I search something, we have very different, personalized experiences, and search across all channels is only getting more personalized. I'll come back to measurement, but as you think about optimizing for AI search: keywords gave us grounding, an anchor in volume. Google could say people searching "HVAC repair" number 27,500 a month, and that might be a good keyword to target.

(17:02): With prompts, the average prompt is 42 words long, so there's no search volume associated with that exact phrase. The key to AI search is getting away from volume as the indicator of where to prioritize, and retraining your leadership team to start from: who do we want to do more business with, who are our best customers? Then build a prompt and keyword strategy around them. On the measurement side, you need to look at the impact on qualified lead generation and organic revenue, rather than visibility metrics, rankings, and traffic based on keyword volume — those are dying metrics, in my opinion.

Tim (17:54): It's such an interesting topic. We've provided visibility tools to help, but there's a big gap between "am I visible?" and "is it translating into actual business?" How do you manage that expectation with clients?

Allyson (18:22): Great performance marketing is about knowing how to read the tea leaves between different metrics and connect them, because there won't be just one metric that says "we're great" — unless it's tied to your pipeline and CRM. Visibility metrics matter insofar as we're getting in front of the right audiences, but they're not the be-all and end-all. If visibility doesn't translate to your CRM, it doesn't count. For us, organic revenue is a combination of organic-search-attributed revenue plus direct and AI referrals — a newer attribution component. Looking at all of those and seeing growth over time is the goal; the visibility and referral metrics are the early indicators of AI search performance.

How to Win in AI Search

Tim (19:24): Say you're a CRO talking to clients and prospects. Some of them say, "Allyson, I don't know what to do. How do I win in this new AI search world?" What do you talk to them about? How do you get them to feel like they have some control?

Allyson (19:47): A lot of the marketers we talk to initially come in with the delusion that SEO and AEO — or GEO, or AI search, whatever you call it — are the same thing. "Good SEO is good AI search." That couldn't be more wrong. There's overlap — it's a Venn diagram — but to feel a sense of control, they need to understand the nuance in how these platforms index, rank, mention, and recommend brands. Even before that, the big lightbulb moment goes back to personalization. So many companies market to the general public or their broad category — "CRM software" — versus what their CRM is specifically for. Who are you for?

(20:55): Get crispy on who you serve and why you're different. That can be the unlock that fuels the right kind of AI visibility and ultimately leads to significant growth down-funnel.

Tim (21:08): That's great advice. It's almost back to Marketing 101 — who's your ICP and how do you talk to them directly.

Allyson (21:15): We've gone back 50 years, for good reason.

Rapid Fire: Metrics, Hype, and What to Change on Monday

Tim (21:18): Rapid fire. If you could only pick three metrics to report to a client over the next 12 months, what would they be?

Allyson (21:31): Organic revenue, number one, defined by a broader net of attributed sources. Then AI mention share, and AI referrals.

Tim (21:46): Most over-hyped claim in the GEO/AEO/AI search conversation right now?

Allyson (21:53): Am I allowed to say "SEO is dead"? People have been saying that for 20 years. But no — I'll go back to "SEO and AI search are the same." They're not.

Tim (22:02): What are you deliberately not chasing?

Allyson (22:06): Citation volume. There's a decoupling between citations and being part of the recommendation. Unless you just want people grabbing your content without mentioning or recommending your brand, citation volume is a vanity metric.

Tim (22:26): You're talking to someone running a small marketing team, no budget, growing. What's the first thing they should change on Monday?

Allyson (22:46): Flat budget, small team, they need to be scrappy. I'd use a tool like SparkToro to understand where your ICP spends their time. The first channel is always Google, but look at the top two or three. Pick one, go into it. If you don't have an AI search tool, do some manual searches and understand what the citation profiles look like. Pick one prompt, do a little competitive research to understand what type of content is winning, and produce a handful of those assets to start.

Tim (23:31): Last one. A buyer trying to hire an agency to help them navigate AI search — how do they figure out if that agency knows what they're talking about?

Allyson (23:41): Ask questions about the playbooks. If the answer is "SEO and AI search are the same, you don't need to do anything differently" — run for the exits. It's very different, and different approaches get you visibility on different channels. Get into the weeds on how they think about driving visibility here versus there. And go back to the metrics conversation: what's their focus on actual business growth and revenue versus visibility gains? It's very easy to drive visibility across all these channels, but as it always has been, that visibility doesn't always lead to what matters.

Where to Find Allyson & Silverback Strategies

Tim (24:25): Allyson Cochran, CEO of Silverback Strategies. Allyson, where can people find you?

Allyson (24:31): I'm active on LinkedIn — Allyson Cochran. And if you want to check out Silverback Strategies, it's silverbackstrategies.com. Just like the gorilla.

Tim (24:42): Awesome. Thanks for doing this — really great to have you, and hope to have you back sometime.

Allyson (24:46): Likewise. Great to see you. Thanks, Tim.
 

About the Author

Tim Condon Chief Revenue Officer at Clutch

Tim Condon is the Chief Revenue Officer at Clutch, the leading global marketplace of B2B service providers. Prior to Clutch, Tim served as the Chief Revenue Officer at Homesnap, the top-rated real estate app built for agents, which CoStar Group acquired in 2020. During his tenure, Homesnap grew its paying user base from 0 to over 80,000 clients. In addition, he previously served as the Director of New Ventures at The Washington Post. In this role, he built several new businesses, including The Capitol Deal, which became the third largest deal site in the DC metro area and was known for giving away 100,000 pizzas

See full profile