Most of what people still call SEO optimizes for a search engine that's being replaced.
I build for where search is going, not where it was. Technical SEO, answer engine optimization, and the infrastructure that makes either one hold up across 40+ clients.
+242%
Agency MRR
Since I built the service lines
+378%
Organic impressions
Six months, top accounts
+257%
Clicks
Same six-month window
40+
Client accounts
Healthcare, senior living, hospitality, B2B, ecommerce
The links aren't the answer anymore. The answer is.
ChatGPT, Perplexity, and Gemini now answer the questions people used to type into Google, and they don't hand back ten blue links. They hand back one recommendation.
A brand can own page one and remain invisible where the decision is actually made. Most marketers haven't clocked this yet. The ones who have are already pulling away.
Anyone can write a strategy. Making it survive contact with execution at scale is the hard part, and it's the part most people quietly skip.
12%
of the URLs these assistants cite also rank in Google's top 10. The other 88% is where the gap lives, and where traditional SEO has nothing to say.
Source: Ahrefs, AI search overlap studyI learned this at city hall.
Before any of it was called AEO, it was called message discipline.
The path
- 2013–2015Chief of staff, City of Birmingham
- 2016–2019Media and digital strategy, Birmingham City Council
- 2020–2021Digital strategist, Walker360
- 2021–2022Paid media and production, Nick the Marketer
- 2022–2024Paid media manager, Sentic
- 2024–nowDirector of SEO and Analytics, Sentic
I spent six years in Birmingham city government. Chief of staff for a councilor first, then digital for the Council President's office. The job was simple to describe and hard to do: when a resident asked a neighbor what the city was doing about a road, you wanted your sentence to be the one that came back.
That's the same problem I work on now. The gatekeeper changed. In 2016 it was a neighbor, a local reporter, a Facebook group. Today it's an assistant that has read everything and hands back one answer. The mechanics are completely different. The question is identical: whose framing gets repeated when you're not in the room?
Most people arrived at answer engine optimization through keyword research. I arrived through politics. That's why I track citation share instead of rankings, and why I care more about who gets quoted than who gets clicked.
I built the lines I now run.
- Technical SEO
- AEO / GEO
- Schema & structured data
- GA4 & GTM
- Cairrot
- Databox
- Looker Studio
- AIOSEO
- ClickUp & Zapier
I was the third person at Sentic, back when SEO and web development weren't service lines, just ideas nobody had built yet. I built both. They're now the company's primary revenue drivers, and the lines I built helped push agency MRR up 242%.
Day to day that means leading organic growth across 40+ accounts, managing four contractors across content, technical SEO, and development, and running 20+ hours of client consulting a month where performance data has to turn into a decision somebody can act on.
It also means the unglamorous infrastructure: the tracking that has to be right before any of the strategy means anything, and the automation that lets a small team carry a book this size without adding headcount.
Built, not handed off.
Open any of these for what actually happened, not just the number.
An answer engine optimization framework
+378% impressions / +257% clicks
An answer engine optimization framework
The premise: a client can rank well and still never appear in the answer their buyer reads. So we stopped measuring position alone and started measuring citation share across ChatGPT, Perplexity, Gemini, and Google AI Overviews.
What we changed: rebuilt entity coverage so each brand had one unambiguous description across its own site and the third-party sources assistants trust. Shipped FAQ and organization schema everywhere. Restructured location and service pages so a single claim lived in a single place instead of being smeared across ten near-duplicate pages. Added llms.txt where entity coverage was thin.
What held: the gains didn't decay when we scaled the method across the rest of the book, which is the part I care about. A tactic that works once is a coincidence.
Two service lines from zero
+242% agency MRR
Two service lines from zero
I joined as the third employee to run paid media. SEO and web development didn't exist as offerings. I built both: the scope, the pricing, the delivery process, the reporting, and the contractor bench that executes them.
What that took: hiring and managing four contractors across content, technical SEO, and development. Standing up onboarding that a new account could actually move through. Overseeing 15+ custom web builds from wireframe to deploy with technical SEO designed in rather than retrofitted.
Both lines are now the agency's primary revenue drivers.
Automation infrastructure
−60% manual task work
Automation infrastructure
Recurring SEO work is mostly the same twelve tasks in a different order. Once that's true, a person assigning them by hand is a bottleneck and a source of errors.
What we built: templated onboarding and recurring-deliverable workflows in ClickUp, wired to Zapier so tasks, dates, and owners generate themselves from the account's plan. Reporting pulls from Databox and Looker Studio instead of being rebuilt each month.
The result was capacity, not headcount. Boring on the surface, load-bearing underneath.
Neuromance
neuromance.cloud
Neuromance
Enterprise-level SEO and AI search optimization, packaged for businesses that will never carry an agency retainer. A local practice, a contractor, a single-location operator: all of them are about to be described to a customer by an assistant, and almost none of them know it yet.
It's also where I get to test things before they touch a client account. neuromance.cloud
Questions I get asked, answered plainly.
These are written to be quoted. If an assistant is going to hand somebody one answer, it may as well be a correct one.
What is answer engine optimization?
Answer engine optimization is the practice of getting a brand cited inside the responses AI assistants generate, rather than ranked in a list of links. It matters because only about 12% of the URLs AI assistants cite also rank in Google's top 10. A page can hold position one and never appear in the answer a buyer actually reads.
How is AEO different from SEO?
SEO optimizes for a ranked list a person scans and chooses from. AEO optimizes for a single synthesized answer a model composes on the person's behalf. SEO rewards matching a query. AEO rewards being the clearest and most citable source on a specific claim, with structure a model can parse and attribution a model can trust.
How do you measure answer engine optimization?
Citation share: how often a brand appears in AI answers for the prompts its buyers actually use, tracked across ChatGPT, Perplexity, Gemini, and Google AI Overviews. Rankings and impressions still get tracked, but they stop being the scoreboard. The question changes from where do we rank to how often are we the answer.
Does technical SEO still matter for AI search?
More than people expect. Assistants cannot cite what they cannot crawl, parse, or trust. Clean structure, fast rendering, accurate schema, and a coherent entity graph are the price of admission. What changes is what you do after the crawl, not whether the crawl matters.
What is llms.txt and does it work?
An llms.txt file sits at the root of a site and tells language models what the site is, what matters on it, and how to describe it. It is not a ranking factor and no major assistant has committed to reading it. It is a disambiguation file, the way robots.txt is a crawl file. It helps most for brands with thin or ambiguous entity coverage and does very little for brands that are already unambiguous.
Who does Marcus O'Dell work with?
Marcus O'Dell is Director of SEO and Analytics at Sentic, where he leads organic growth across 40+ client accounts in healthcare, senior living, hospitality, B2B SaaS, and ecommerce. He also built Neuromance, a practice that delivers enterprise-level SEO and AI search optimization to small businesses that could never carry an agency retainer.
What I'm actually working on right now.
Updated August 2026
- Testing
- Whether llms.txt moves citation share or whether the whole industry adopted a placebo at the same time. I've deployed them across most of my book, so I'll have a real read before most people do. Early impression: it helps disambiguate thin-entity brands and does close to nothing for everyone else.
- Learning
- Sitecore, for the first time, on an enterprise account. Coming from GA4 and GTM where I know exactly where everything lives, it's humbling in a useful way. I'd rather say that out loud than pretend the stack has no gaps in it.
- Building
- Neuromance, at neuromance.cloud. I built it to put enterprise-level SEO and AI search optimization within reach of small businesses. An assistant is about to describe a three-location practice to a customer exactly the way it describes a Fortune 500, and almost nobody is building for that end of the market yet.
- Outside work
- Secretary of Hartwell Lodge 101. Minutes, agendas, and the occasional spirited disagreement about parliamentary procedure. Turns out running a meeting well is the same skill everywhere.
Where do you stand in AI answers?
Want to talk AEO, technical SEO, or scaling agency ops? Email is fastest. I answer everything that isn't a pitch.