Check this page with an assistantOpens a chat asking it to summarise this article and name the evidence behind each claim.
Claude opens with the prompt on your clipboard: Anthropic does not support prefilled prompts on the web, and we would rather copy it than ship a button that drops it.
Disclosure: SearchHandled is a product in the fourth category discussed below. We have tried to describe where the other three win, including where they beat us, because a comparison that only ever concludes in the author's favour is not worth reading.
Ask the Implementation Question First
Before comparing models, answer this: when a change to the website is needed, who makes it, and how long does it take? If the answer is "nobody reliably" or "weeks", that is your binding constraint, and buying more analysis will not relieve it. Most failed SEO engagements are implementation failures wearing a strategy costume.
This reorders the entire decision. A brilliant consultant delivering into an organisation that cannot ship changes produces an expensive document. A mediocre agency with a developer who acts on things produces results. When people say SEO did not work for them, the post-mortem very often finds a list of sensible recommendations that were never implemented, and nobody in the chain considered that their job.
So write down, honestly, three things: who owns the website, how quickly they can publish a page or change a template, and who decides what is worth doing. Those answers eliminate at least one of the four models below before you have looked at a single price.
The Four Models Side by Side
Each model is a different answer to the same three questions: who supplies judgement, who supplies labour, and who carries continuity. Price differences mostly follow from those, which is why comparing monthly fees alone tends to produce the wrong choice.
| Model | Best when | Fails when |
|---|---|---|
| Freelancer or consultant | One clear problem, and someone internal who can decide and ship | The work spans disciplines, or they become unavailable |
| Agency | Broad work needing coordination nobody internal can provide | Your account is small relative to their book, so attention drifts |
| In-house | Search is a primary channel, with writers and developers around them | One isolated person with no support and no second opinion |
| Software or managed product | You need to know what to do next, and can already execute | Nobody actions the output, or the calls are genuinely ambiguous |
Published price ranges put freelancers around $1,000 to $3,500 a month, small-business agency retainers around $1,500 to $5,000 or more, and in-house specialists between roughly $100,000 and $170,000 a year once benefits and tools are counted. Those figures come from vendor surveys without disclosed methodology and vary hugely by market, so use them to size the decision, not to budget it. Our own breakdown of what drives the number is in how much SEO costs for a small business.
Where Each One Genuinely Wins
Every model has a situation where it is clearly the right answer. Recognising yours is more useful than a scored comparison, because the scoring depends entirely on circumstances the scorer does not know.
- A freelancer wins on depth and directness.
You get the person who does the work, not an account manager relaying to a junior. For a specific technical problem, a migration, or a diagnosis, an experienced independent is frequently better than an agency and costs less. The risk is concentration: one person, one calendar, no cover.
- An agency wins on breadth and continuity.
Technical, content, and outreach coordinated by someone whose job is coordinating them, with the work continuing when somebody is on holiday. If nobody internally has time to run a programme, this is what you are actually buying, and it is worth paying for.
- In-house wins on context and speed.
Someone who knows the product, sits near the developers, and can act on Tuesday rather than at the next monthly call. When search is a primary acquisition channel, that proximity beats any external arrangement. Below that threshold the fixed cost is hard to justify.
- Software wins on consistency and cost.
It runs the same analysis every week without getting bored, does not have a bad quarter, and costs a fraction of a retainer. What it cannot do is judge a genuinely ambiguous trade-off or make anyone act, and vendors in this category, including us, should say so plainly.
Month Seven Is the Real Test
Almost every arrangement works in month one, when attention is high and the obvious wins are available. The differences appear around month six or seven, when the easy work is done, the initial enthusiasm has faded, and somebody has to decide what is worth doing next.
This is where each model fails in its characteristic way. Freelancers get busy with a larger client. Agency accounts get reassigned to a more junior team as the novelty passes. In-house specialists run out of things they can implement without developer time and start producing reports instead. Software keeps producing recommendations that nobody has actioned since March.
Ask about month seven explicitly before signing anything. Who will be doing this work then, what happens when the obvious wins are exhausted, and how will we decide what is next? The answers are more predictive than any case study, and the questions worth asking any provider, including us, are collected in questions to ask an SEO agency.
Combinations Usually Beat Single Choices
The framing as four exclusive options is mostly an artefact of who writes these comparisons. In practice the arrangements that hold up for small businesses tend to mix: something that decides what to do next, and someone who can do it.
The common working pattern is a system producing a prioritised queue, a person inside the business who owns publishing and can ship changes, and occasional access to an experienced independent for the genuinely ambiguous calls. That is cheaper than a retainer, faster than in-house hiring, and avoids the failure mode where analysis accumulates and nothing changes. The operating rhythm for it is in SEO content operations.
If you are an agency reading this rather than a business, the calculation is different and mostly about which parts of delivery should be systematised so your people spend their time on judgement rather than on assembling the same audit for the eleventh time. That is a separate conversation, and we keep it on the agencies page.
Questions People Ask About Choosing an SEO Model
- Is an SEO agency or a freelancer better for a small business?
A freelancer is usually better for one clearly defined capability; an agency is usually better for a broad, coordinated program. The freelancer still needs someone internally who can make decisions and ship changes. The agency fits when work spans technical, content, and outreach at once and nobody internally has time to coordinate it.
- How much does each option cost?
Published ranges put freelancers around $1,000 to $3,500 a month and agencies serving small businesses around $1,500 to $5,000 or more. An in-house specialist is commonly cited between $40,000 and $100,000 in salary, or roughly $100,000 to $170,000 fully loaded with benefits and tools. Treat all of these as directional, since they come from vendor surveys rather than disclosed datasets, and vary enormously by market.
- Is in-house SEO worth it for a small company?
Rarely, and not for the reason people expect. The problem is not the salary but the isolation: a single in-house specialist has no one to check their thinking, no exposure to how other sites behave, and no cover when they leave. In-house works when SEO is a primary acquisition channel and the person is supported by writers and developers who can actually ship what they recommend.
- What is the most common reason SEO engagements fail?
Nothing gets implemented. The audit arrives, the recommendations are reasonable, and then they sit in a document because whoever would have to change the site has other priorities. Before choosing any model, decide who will make changes to the site and how quickly, because that constraint determines which model can work far more than price does.
- Should I use software instead of hiring someone?
Software is a good fit when your bottleneck is knowing what to do next and you have someone who can execute, and a poor fit when you need judgement about ambiguous trade-offs or when nobody will action the output. Tools that produce a prioritised queue replace the analysis, not the implementation, and being honest about which half you are missing is the whole decision.
