Enterprise SEO skills, build the team or hire an agency

Enterprise SEO skills are not one job description. A program on a site that size runs on about ten separate competences, and the real question behind building a team or hiring an agency is which of those ten you can hold on your own payroll and which you have to buy from outside. Answer that first and the staffing decision mostly answers itself.

Getting it wrong stays invisible for about a year. Hire one generalist and you get somebody who can name every problem on your site and ship none of them. Sign a retainer with nobody internal to receive it and you pay for a document that gets read once. Both arrive in the board deck looking like an SEO problem. Neither one is.

You should know where we stand before any of the argument below. We sell enterprise SEO programs for large sites, so a reader who finishes this page and posts an internal job ad buys nothing from us, and a reader who books a call might. That is the bias to read this against.

What enterprise SEO skills actually cover at this size

Start with what the job stops being. Search Engine Journal’s guide to enterprise SEO skills, written by Jes Scholz and published in July 2021, opens on keyword research, link building, HTML and CSS and says “these are all skills that are not directly needed by an enterprise SEO.” That is structural rather than snobbery. At this size, “Keyword research is done by the content team, link building is done by PR, and development won’t let you deploy changes to HTML” and CSS.

So the craft a smaller search hire is bought for has already been distributed to other departments, and what is left is much harder to write a job ad for. The same article describes a role where “their job is not to do SEO, but rather to advocate for SEO”, and where “Yours is a role of continual cross-functional collaboration.”

That one shift decides the staffing question. If the job were execution you would buy whoever executes best and the org chart would be a detail. Because the job is diagnosis, translation and influence, where the person sits changes how much of their skill reaches production. If you are still fixing the boundary of the category, what counts as enterprise SEO covers it.

One thing worth knowing before you research further. Of the ten results Google returns for this phrase, only two are about the skills themselves. The rest define the category or sell a platform.

Where we stand, including the part that does not flatter us

Here is the test that saves you having to trust us. Read every recommendation below and ask whether it would still be the advice if we sold software instead, or sold nothing. Where the answer is yes, it is probably sound. Where a passage recommends outside help and you cannot see what saying the opposite would cost us, treat it as sales copy and skip it.

Now the part that fails our own test. Our enterprise page sells this work on three named client stories, “iSmile Dental Spa”, “LifeStance Health Inc.” and “Host Duplex”. A dental practice, a mental health group and a hosting company. Those are real engagements and none is at the scale that page is written for. If you run a hundred thousand URLs across four markets, our proof at your size is not on that page. Ask us for it directly instead of taking the page’s word, and hold whatever comes back to the same standard, because reading an enterprise SEO case study properly is a skill in itself.

Run that test on everything else you read on this question too, because the pull is the same everywhere. WebFX’s in-house against agency comparison is the most even-handed page on this search, and the named expert it quotes on whether to use an agency is captioned “Director of Sales, WebFX”, saying “Working with an SEO agency saves you time, and you get to work with a team of people who have already tackled whatever digital challenges you’re facing many times.” That is a sales leader answering a buying question, on a page that argues fairly everywhere else. The pattern is not dishonesty. It is gravity, and ours pulls the same way.

The ten competences an enterprise search program has to hold

Every argument about enterprise SEO skills we have been pulled into came down to this table. Write it out before you write either a job description or a request for proposal, because the decision is not in-house against agency. It is which of the ten you already hold, which you can hire, and which you have to rent.

Four sit inside your company by definition and cannot be bought at any price. One sits outside by definition and cannot be built. The remaining five are the actual decision.

The skills required for enterprise SEO that only an employee can carry

The skills required for enterprise SEO that no contract can supply have one thing in common. They are not knowledge. They are standing.

• Cross-team advocacy. You cannot be pulled into somebody else’s project if you are not on the invite list. An outside consultant gets told about a replatform. An employee gets asked while it is still a wireframe, which is the only point where the answer is cheap.
• Engineering priority. Conductor’s guide to building an enterprise SEO team calls IT “the gatekeeper to the technical side of the site” and says “Implementation is the most significant hurdle in SEO”. Any competent provider writes a clean ticket. Whether it gets pointed into a sprint is decided by somebody holding a company badge.
• Executive translation at budget time. The same Search Engine Journal guide warns that “If you report on performance by showing keywords that have moved up in rankings, you’re doing it wrong”, because you are talking to “people who have never logged into Google Search Console”. An agency presents monthly. An employee is in the session where next year’s number gets argued over.
• Governance and decision rights. Conductor puts the failure plainly. “The biggest enemy of enterprise SEO success isn’t an algorithm; it’s disorganization.” An outside party can draft a responsibility matrix. Only an internal owner enforces one, because enforcement means telling a peer no and still being there next quarter.

None of those four improves if you buy a better vendor. They improve if you hire a person with authority and give them a seat. Which model of authority your program assumes gets settled in an enterprise SEO strategy that survives a year.

What an outside team supplies that a payroll line cannot

One competence is genuinely unbuildable internally, and it is the least discussed. You see one site. An agency sees dozens in the same week. When your category template loses traffic, the only way to know whether that was your change or the whole vertical moving is to have watched other sites through the same days. No internal team can construct that view, whatever it spends, because it has one data set.

Below that sit two practical arguments rather than structural ones. The first is specialties you cannot keep busy. Hreflang across thirty markets, a replatform redirect map, log file analysis, digital PR outreach. Each is a real skill and none is forty hours a week forever, so you either underuse a salary or rent the hours. The second is capacity that arrives trained, which matters most in the quarter when the audit and the migration land together.

Botify’s guide to team structure says “Even the most successful in-house SEO teams still utilize agencies & consultants to help execute and validate their strategies.” Botify sells an enterprise crawl platform, so read that as a vendor describing its own customers rather than as a finding. What makes it worth repeating is the list underneath, which is narrow. Strategy validation, testing and experimentation, content creation, reporting and monitoring. Nothing on it is owning the program.

Ask any provider, us included, for an engagement shaped like yours, and read our case studies as skeptically as a competitor’s.

The five competences either side can hold, and what decides it

These five are where the argument actually lives, and none turns on who is better at search. They turn on three practical things. How often the work is needed, whether you can license the tooling, and whether the person doing it has to know your product.

• Crawl and log diagnosis. Frequency decides. A site changing weekly needs continuous crawling and the license underneath it. If the platform bill is your blocker, enterprise SEO platforms compared shows where published prices stop and quotes start.
• Template and platform judgment. Either side, but it needs experience of your specific stack. A provider who has never shipped on your CMS writes recommendations your engineers cannot action, and that damage lands on your search lead rather than on them. A large catalog narrows the field further, which is why enterprise ecommerce SEO is usually scoped as its own problem.
• Segmentation and measurement. Most often assumed to be outside work and most often better inside, because your segments have to match how the business already reports. A provider that invents its own segments hands you a report nobody in your finance meeting can reconcile against anything they already trust.
• Subject knowledge in content. Our own enterprise page states the position against our interest. “Your writers know the product better than any agency will.” It is worth quoting only because it sits on a page selling agency services.
• Earned authority. Usually outside, because it runs on media relationships a single company cannot build from inside its own category, and because the brand risk is best held by somebody your comms team signs off.

Notice what decided each of those five. Frequency, licensing and product knowledge, never talent. If you find yourself arguing about who is more skilled, you are having the wrong argument about this half of the table.

Where the enterprise SEO skills you hire will actually sit

The same hire produces different work depending on which organization they report into, and most staffing decisions skip this. Your enterprise SEO team structure is not an administrative detail that follows the hire. It changes which of the ten competences that hire can convert into shipped changes.

Botify puts the split directly. “SEOs that live with marketing tend to focus more on content optimization and have a more difficult time getting SEO fixes prioritized.” Against that, “SEOs that live with product/engineering tend to have an easier time getting large-scale, big-impact SEO fixes deployed.” It goes further and says “SEO teams that have a dedicated resource on the product/engineering team are by far the most successful from our experience.”

Conductor is blunter about the wrong answer. “IT is where SEO teams go to die.” Its reasoning is budget and scope rather than technical fit, since a team under an IT leader gets measured on tickets closed, not revenue found.

Hold that against your own bottleneck. If nothing on your site is worth ranking, a marketing seat is right and a technical hire is premature. If four hundred approved fixes sit in a backlog, a marketing seat will not clear them and neither will an agency. That is an engineering capacity problem in a search costume.

One hire or three, and what the first person has to be able to do

Conductor argues against matching team size to company size, saying “you need a few specialized folks with the right support to keep the program moving forward”. Botify comes down on specialists, saying “it’s preferable for enterprises to employ a strategy with specialists, given the vastly different skill sets required for different aspects of SEO”. Both are vendors describing their own customers, so read them as pattern, not proof.

The practical version is narrower. Your first hire is not the best technician available. It is the person who can hold the four internal-only competences, because those are the only ones nobody else can be hired to do. Technical depth is rentable from month one. Standing is not.

Test for it this way. A candidate who can explain a canonical tag to a finance director in one sentence, without using the word canonical, is worth more in year one than one who writes better regular expressions. The second skill is on the market every day. The first keeps the program funded, which is why reporting an executive will read decides whether there is a year two.

One figure worth having while you scope the role. WebFX’s in-house against agency comparison gives salary bands for a United States based search professional and puts a head of SEO or director at “$120,000 to $180,000+” a year, read on 18 September 2026. It says those come from ranges it “gathered” without naming a source, so treat it as an order of magnitude. That band buys standing and judgment, not technical execution.

How to test for the skills required for SEO enterprise programs before you sign

Interviews reward presentation, and so do sales calls. Run the same four tests on a candidate and on a shortlisted provider, with identical materials, and compare outputs instead of answers.

• The segmentation test. Hand over one month of Search Console data for a single template group and ask what they would change first, and what evidence would prove them wrong. Nobody who has done this at scale answers without first asking who ships the fix.
• The ticket test. Give them a real change request your team failed to get shipped and ask them to rewrite it. This separates people who write recommendations from people who write acceptance criteria, and only the second kind clears sprint planning.
• The wrongness test. Ask for the last forecast they got wrong and what changed afterward. Google’s own guidance for buyers is direct about the failure mode. “No one can guarantee a #1 ranking on Google.”
• The access test. Ask what access they need on day one. Google tells buyers to “only grant read access to Search Console (at this stage, don’t grant them write access)” while an audit runs. A provider asking for write access before a scope exists has told you how it treats other people’s property.

The same four work on an internal candidate, which is the entire point. If you cannot tell an applicant and an agency apart on the outputs, the decision in front of you is about continuity and cost rather than skill.

When you should not buy from anyone

Four situations where hiring us, or any agency, or any consultant, is the wrong purchase. Each one is checkable before you take a call.

• You have nobody internal to receive the work. Not a busy owner. None. Every finding needs an internal name to argue for it and a calendar to defend it. Hire that person first even if it means waiting two quarters to buy anything else.
• Your bottleneck is release capacity. If you already know what is wrong, the fixes are approved, and they are queued behind product work, more diagnosis only lengthens the queue. That budget should buy developer hours or a standing sprint allocation. It should not buy us.
• You want an outside voice to settle an internal argument. Used as ammunition, an agency creates a second roadmap and a loser with a reason to slow the work down. Settle it internally, then buy execution against the settled answer.
• Your problem is one narrow specialty. A single hreflang implementation, one migration, one schema rollout. A specialist contractor does that faster and cheaper than a program retainer, and you keep the relationship for the next one.

There is a fifth that is not about your team at all. If your budget runs one year and your category is held by a competitor whose position took considerably longer than that to build, no provider can honestly promise the outcome you want inside your window. The ones that do are the ones to cut first.

Who should actually run both, and who only thinks they should

Now the uncomfortable part. Our own enterprise page answers this exact question with “Most large teams need both”, and adds “We work alongside it. Your enterprise SEO expert stays the owner.” That is not a neutral verdict. It is our commercial position, it is where almost every agency on this search lands, and arriving at it by default would be the easiest thing this article could do.

So here it is with conditions attached. Running both is right for a specific reader, and four checks tell you whether that is you.

• All four internal-only competences have a named employee against them, and that employee has hours this quarter.
• Your unowned rows are crawl scale, digital PR or benchmarks, which are the rows an outside team genuinely fills.
• You can run one backlog, and the provider writes into your tracker rather than into a monthly deck.
• You and the provider agree one definition of shipped in week one, before anything is scoped.

Fail the first check and running both is not a hybrid. It is an agency with no client-side owner, which is the first situation above. Fail the third and you have two roadmaps competing for the same engineers, which is slower than either model alone. Either way the answer is to buy nothing yet and fix the internal side.

If you do run both, get the exit clause in writing. Our page commits that “Search Console, GA4, and Tag Manager stay in your name. Leave us and the footprint stays.” Demand the equivalent from whoever you sign, us included, because the split that traps you is the one where leaving costs you your own data.

If you get past this decision and start scoring providers, how to choose an SEO agency takes it from there. This article stops at your side of the table deliberately.

What we would do first with this decision on your desk

We would not start from a budget. We would print the ten competences and write a name against each row. A real person, an employee or a named individual at a vendor, never a team or a department.

Then count the empty rows, because that count is the whole decision. If the empty rows are advocacy, engineering priority, executive translation and governance, you have a hiring problem and no provider solves it. If they are crawl scale, digital PR and benchmarks, you have a buying problem and a single hire takes three quarters to half solve it. If both sets are empty, you are not ready to buy anything, and the first move is an internal owner.

Then ask one more question of every row that does have a name. Does that person have the hours this quarter. Most enterprise search programs do not fail on competence. They fail because six of the ten rows carry the same name, and that person is also running the replatform.

Do that before you take any sales call, ours included. It tells you which of the three answers you actually have rather than which one a vendor prefers. If you want a second read on the rows you cannot fill, tell us what your ten look like and we will tell you which ones we would not take.