Marketing Case Studies

New case study. We installed an entitymap on waikay.io on April 25. We changed nothing else. Here's what happened. AI Knowledge Scores on Gemini improved by up to 26 points in 48 hours. The entitymap.html file is now cited by Gemini 2.2x more often than our own About page. On Sonar (Perplexity), it's 3.0x. The most striking result: one topic had been declining for three months, dropping roughly 10 points per month with no obvious cause. Forty-eight hours after install, the entire decline reversed. Not partially. Entirely. New content takes weeks to index. Backlinks take months. A 48-hour reversal points directly at live retrieval as the mechanism, which is exactly what an entitymap is designed to influence. Where did it *not* work? ChatGPT, Copilot, and Claude never cited the file. Not because they rejected it, because Bing hasn't indexed it yet. Entirely our fault for not submitting it to Bing Webmaster Tools before launch. That's a 20-minute job we skipped. We'll publish updated data once it's done. If you want to run your own test, Genie Jones has written the whole methodology up in the case study: baseline first, deploy on one date, change nothing else, and watch your lowest-scoring topics. That's where the effect shows up fastest. The results are strong enough to act on. They're not strong enough to call settled science. One result on one surface. But it's a result worth replicating. https://lnkd.in/ekKT5RAY #EntitySEO #GEO #AIO #AISearch #SEO

We expected content quality to drive AI citations. After 90 days tracking 50 B2B SaaS companies across ChatGPT, Perplexity, and Claude Code, the data showed something embarrassing: Our best-performing blog post was almost invisible to AI search. A competitor's outdated pricing page was cited 23 times. That observation sent us down a rabbit hole. Because it suggested something uncomfortable: 𝐀𝐈 𝐬𝐞𝐚𝐫𝐜𝐡 𝐝𝐨𝐞𝐬𝐧’𝐭 𝐰𝐨𝐫𝐤 𝐥𝐢𝐤𝐞 𝐭𝐫𝐚𝐝𝐢𝐭𝐢𝐨𝐧𝐚𝐥 𝐬𝐞𝐚𝐫𝐜𝐡. Traditional SEO optimizes for ranking algorithms. GEO (Generative Engine Optimization) optimizes for citation algorithms. Different systems. Different incentives. Different winners. Over 90 days, we tracked citation frequency, entity recognition, structured data implementation, attribution patterns, and visibility consistency across three major AI platforms. The biggest finding? 𝐓𝐫𝐚𝐝𝐢𝐭𝐢𝐨𝐧𝐚𝐥 𝐒𝐄𝐎 𝐚𝐮𝐭𝐡𝐨𝐫𝐢𝐭𝐲 𝐡𝐚𝐝 𝐧𝐞𝐚𝐫𝐥𝐲 𝐳𝐞𝐫𝐨 𝐜𝐨𝐫𝐫𝐞𝐥𝐚𝐭𝐢𝐨𝐧 𝐰𝐢𝐭𝐡 𝐀𝐈 𝐜𝐢𝐭𝐚𝐭𝐢𝐨𝐧 𝐟𝐫𝐞𝐪𝐮𝐞𝐧𝐜𝐲. What mattered most was structured business context. Companies with comprehensive Organization + Product schema saw: 𝐮𝐩 𝐭𝐨 𝟑.𝟒𝐱 𝐡𝐢𝐠𝐡𝐞𝐫 𝐜𝐢𝐭𝐚𝐭𝐢𝐨𝐧 𝐫𝐚𝐭𝐞𝐬. We also found platform-specific behaviors: → Perplexity heavily weighted Knowledge Graph relationships and entity consistency. → Claude Code favored sources with strong attribution chains and verifiable claims. → ChatGPT responded best to coherent entity structures and consistent product definitions. The surprising part? The gap wasn't content quality. It was implementation completeness. Most companies had: • partial schema • weak entity relationships • inconsistent terminology • fragmented attribution In other words: 𝐀𝐈 𝐜𝐨𝐮𝐥𝐝 𝐫𝐞𝐚𝐝 𝐭𝐡𝐞𝐢𝐫 𝐜𝐨𝐧𝐭𝐞𝐧𝐭. 𝐁𝐮𝐭 𝐜𝐨𝐮𝐥𝐝𝐧’𝐭 𝐜𝐨𝐧𝐟𝐢𝐝𝐞𝐧𝐭𝐥𝐲 𝐮𝐧𝐝𝐞𝐫𝐬𝐭𝐚𝐧𝐝 𝐭𝐡𝐞𝐢𝐫 𝐛𝐮𝐬𝐢𝐧𝐞𝐬𝐬. The full report includes: → complete methodology → correlation datasets → platform-specific optimization frameworks → deployable schema templates → implementation roadmap Thoughts Are we witnessing the biggest search paradigm shift since Google's PageRank?

We audited an accounting software site with 3,000+ pages doing 10K monthly traffic. Deleted 2,000 pages. 13 months later: 340K monthly traffic. They were competing against themselves for "invoice template" and losing. We forced Google to pick winners. 𝗪𝗵𝗮𝘁 𝗪𝗲 𝗙𝗼𝘂𝗻𝗱 The site had 20+ variations of "invoice template" pages, 15+ "how to make an invoice" articles, 40+ duplicate templates with minor differences, and 200+ thin pages targeting industry-specific invoices. Google was paralyzed. 10K traffic across 3,000 pages = 3 visits per page average. 𝗧𝗵𝗲 𝗖𝗮𝗻𝗻𝗶𝗯𝗮𝗹𝗶𝘇𝗮𝘁𝗶𝗼𝗻 𝗣𝗿𝗼𝗯𝗹𝗲𝗺 They had separate pages for: - Invoice template - Free invoice template - Simple invoice template - Professional invoice template - Business invoice template - Invoice template Word - Invoice template Excel Plus 193 more variations. All ranking position 25-60. Zero on page 1. 𝗧𝗵𝗲 𝗦𝗼𝗹𝘂𝘁𝗶𝗼𝗻 Export all URLs with traffic data. Group by search intent. Keep the strongest page per group. 301 redirect others to the winner. Merge unique content from deleted pages. 2,000 redirects later, the transformation began. 𝗧𝗵𝗲 𝗧𝗶𝗺𝗲𝗹𝗶𝗻𝗲 Month 1-2: 10K traffic (Google processing) Month 3: 18K (first movement) Month 4: 35K (rankings improving) Month 6: 74K (hitting page 1) Month 9: 180K (featured snippets) Month 13: 340K (complete dominance) Same domain. Same backlinks. 34x traffic growth. 𝗦𝗽𝗲𝗰𝗶𝗳𝗶𝗰 𝗥𝗲𝘀𝘂𝗹𝘁𝘀 "Invoice template" - consolidated 47 pages into 1 mega-guide Result: Position 48 → Position 2 Traffic: 500/month → 65K/month "How to make an invoice" - merged 23 pages into 1 comprehensive resource Result: Position 31 → Position 3 Traffic: 100/month → 28K/month 𝗧𝗵𝗲 𝗖𝗼𝗺𝗽𝗼𝘂𝗻𝗱 𝗘𝗳𝗳𝗲𝗰𝘁 When you consolidate: - Link equity concentrates - User signals improve - Dwell time increases - Google trusts your domain more - Other keywords start ranking organically Unexpected keywords that exploded without targeting: "billing software" (22K searches), "invoicing for small business" (8K), "quotation template" (15K), "receipt template" (31K). Google recognized the site as THE invoice authority and rewarded related keywords. 𝗪𝗵𝗮𝘁 𝗪𝗲 𝗞𝗲𝗽𝘁 𝘃𝘀 𝗞𝗶𝗹𝗹𝗲𝗱 Kept: High-traffic performers (top 10%), comprehensive guides, interactive tools, unique angles Killed: Everything under 50 visits/month, duplicate intent pages, thin location/industry variations, outdated yearly content 𝗧𝗵𝗲 𝗖𝗹𝗶𝗲𝗻𝘁'𝘀 𝗝𝗼𝘂𝗿𝗻𝗲𝘆 Initial reaction: "You want to delete 67% of our content? We paid $50K for those pages!" 13 months later: "Can you audit our other sites?" Results silence objections. 𝗧𝗵𝗲 𝗕𝗹𝘂𝗲𝗽𝗿𝗶𝗻𝘁 Audit ruthlessly. One page per search intent maximum. Merge variations into comprehensive resources. Redirect aggressively. Build topical hubs, not page sprawl. Wait (this takes months). Scale exponentially. 34x traffic growth is possible. You just have to be brave enough to delete.

Just wrapped up one of the most interesting SEO audits I've done in a while. Client came to me with a 61% traffic decline over 12 months. The obvious culprit seemed clear: they'd built a ton of top-funnel content that was just a little too far away from their core product. It was the classic SaaS SEO playbook that has always worked but that we're now told doesn't work anymore. It wasn't AI cutting CTR, either, since it was major rankings that they had lost. I couldn't find the ONE thing that could have caused this drop - there were too many factors and variables. I went in expecting to confirm the hypothesis and recommend a content prune. Instead, the data told a completely different story. I found some technical issues. Toxic backlink spike. Slow Core Web Vitals. Crawl budget waste. But none of it explained a 61% decline. The timing didn't line up, the severity didn't match. I was genuinely stumped. Then I started digging into the content data. Their core product pages—the ones that should've been their strongest assets—had been quietly declining for 18 months before anyone noticed. The TOFU wasn't dragging the site down. It was masking the real problem. Even weirder: the TOFU content had better engagement metrics than their product pages. Lower bounce rates, higher engagement. Users actually liked it. So why did they decline 61% while a competitor with the same TOFU strategy only dropped 19%? Domain authority gap. Competitors had 2.5x more referring domains and sat 5-10 DR points higher. That authority acted as a buffer when the algorithm shifted. The sites with less margin for error got hit hardest. The other piece: topical depth. My client had 24x more keyword coverage around "Instagram" than around their actual product category. They were checking all the E-E-A-T boxes at the page level—first-person experience, screenshots, expert attribution—but they never built the foundational content depth on their core topics that would signal real authority to Google. Meanwhile their competitor publishes annual industry research that gets cited everywhere and earns links naturally. Same TOFU strategy, but with an authority-building engine underneath. A few things I'm taking away: - Surface-level content quality isn't enough anymore. You can do everything "right" at the page level and still lose if you're missing the structural pieces—topical depth, original research, domain authority. - TOFU content isn't automatically bad, even with AI on the loose. The engagement data didn't support pruning it. The problem was what they weren't building, not what they were. - Algorithm hits are rarely one thing. This was cumulative damage across multiple updates, compounded by an industry-wide shift. Every competitor declined—just not equally. - Sometimes the diagnosis is "you need to invest differently" rather than "fix this broken thing." That's a harder conversation but a more honest one.

iSkydive America had a hard rule going in: no budget increases. Monthly ROAS floor of 7 per location. Page-builder website with limited SEO levers, a third-party booking engine, and uneven seasonality across 9 cities from Miami to Detroit (some drop zones year-round, others spring through fall). We took the account in December 2024 for PPC, added SEO in January 2025. Here's what the work has done so far, without a single dollar of additional spend. PPC (Google Ads + Performance Max), peak season YoY: +36% conversions +13% conversion value −10% cost per conversion +35% online sales overall The biggest single win was New York. Most contested skydiving market in the country, premium auction prices, fierce bidding from national operators, locked daily budget. We treated it as a focused case inside the broader account: tight geo radius around the drop-zone airport, weekly negative-keyword pruning, restructured campaigns by intent, NYC-specific creative and ad extensions. Two months later: CPA down 60%. ROAS from 2 to 7+. Budget unchanged. SEO over 6 months told an even cleaner story. +119% organic traffic +73% revenue from organic +130 keywords in Google Top-3 +405 keywords in Google Top-10 ROAS 2,900%+. Every dollar spent on SEO returned $29 in revenue. That's roughly 29x the industry-success benchmark for an SEO program. Achieved through technical fixes (Core Web Vitals, 404s, 301 chains, crawl budget cleanup), content rewrites aligned to actual search intent, and a quality-first link-building program. No PBNs, no shortcuts. The lesson I keep coming back to with this account is that constraints made the strategy. A fixed daily budget forced us to win through structure, targeting precision, and bidding cadence. Not by spending more. A locked CMS forced us to fix what was actually fixable in technical SEO and double down on link equity. A rigid third-party booking engine meant we had to be ruthless about traffic quality, because we couldn't optimize the funnel after the click. In every account I've seen over 15+ years, the agencies that ask for "more budget" first usually haven't earned the right to ask. The ones that produce harder, more interesting results work the structure first. If your performance team's first answer to a slowdown is "let's increase spend by 30%," that's the moment to ask harder questions about what's actually broken. Want a structural audit of your account? DM me. #PPC #SEO #GoogleAds #PerformanceMax #DigitalMarketing #CaseStudy

ClickUp's blog went from 1.19 million monthly visits to 28,790 in 15 months. That's a 97.6% decline. And it's still falling. Kamila Olexa published one of the most detailed breakdowns I've read in a while — mapping every Google update, every content decision, every response ClickUp made during the crash. The kind of research that takes real effort. Go read it. Link in comments. It hit differently for me, because ClickUp's blog was one I've referenced more times than I can count. - For structure. - For how they approached topics. - For what a scaled content operation actually looks like. Seeing it reach this point is genuinely sad for a content marketer to watch. But it's also one of the most instructive case studies I've come across in a long time. Especially right now, when the instinct is to publish faster, scale harder, and chase every AI visibility trend going. Here's what I'm taking away: ✅ Scaling without a clear content quality bar compounds problems, not solves them. When things go wrong, publishing more of the same thing rarely helps. Volume is not a recovery strategy. ✅ Topic breadth alone doesn't explain a decline this steep. Covering topics outside your core product isn't the problem. How you cover them is. ✅ Ranking your own product #1 in every listicle is a short-term bet. When every page positions your product as the top recommendation, including where it genuinely isn't — that credibility eventually erodes. ✅ Content written for conversion and content written for readers pull in opposite directions. Too many CTAs, too much promotional weight per page — at some point, the page stops serving the person who landed on it. ✅ Your most important commercial keywords aren't protected just because you own the category. Core, high-intent queries can disappear from your rankings just as fast as peripheral ones. Nothing is safe by default. ✅ Authority metrics don't compensate for content quality gaps. Strong domain metrics matter, but they're not a ceiling or a floor. Quality still determines where you land. ✅ Content refreshes without pruning are only half the job. Adding new content while leaving underperforming pages untouched doesn't resolve a quality signal. Sometimes subtraction matters more than addition. The timing of this matters for me. There's a lot of noise right now about AI visibility and publishing faster to stay relevant. This case study is a reminder that the fundamentals haven't shifted much. Write for the reader. Be honest in your recommendations. Don't let your content become a funnel in disguise. Image source: contentlevers.xyz