Offshore, local, or freelance: the honest tradeoffs

The debate usually gets argued as if location predicted skill. It does not. I have seen careless work produced fifteen minutes from my office and careful work produced eight time zones away.

What location does predict is the shape of the relationship. That is worth choosing deliberately, because the wrong shape fails even when the people are good.

What actually differs between the three?

Read the last row twice. Each model has a failure mode that is structural rather than personal, and you can usually predict which one will bite you.

What does the model actually cost you in your own hours?

This is the arithmetic that decides the question, and almost nobody runs it before signing. Do it with your numbers.

Start with production volume. Say the offshore proposal produces eight pages a month and twelve profile posts. You read a page properly in twelve minutes and a post in two. Eight pages is 96 minutes. Twelve posts is 24 minutes. Total, two hours a month of your attention, before any question, revision, or approval.

If you have thirty minutes a month, you can check 25%derived of what you are paying for. The other three quarters publishes on your domain, under your business name, unread by anyone who knows your business.

Then price the question latency. Take a team eight hours ahead of you. A question raised at 10am waits until their next working morning. Call that one lost working day per blocking question. Over a thirteen week engagement with two blocking questions a week, that is 26 questions and 26 lost days, which is 29%derived of the calendar you are being measured against.

Neither number disqualifies offshore work. They tell you what has to be true for it to work: a specification good enough to remove most blocking questions, and a reviewer on your side with real time. Buy the model that matches the hours you actually have, not the hours you intend to find.

Rerun it for local. Same eight pages, but the review happens live in a thirty minute call and the blocking questions resolve the same afternoon. Your two hours become one, and the 26 lost days become close to zero. That difference is what the local premium is buying. If it does not buy that, you are paying for proximity and getting nothing.

Is offshore work lower quality?

No, and treating it that way will cost you money.

What genuinely varies is context. Writing a page about neighborhoods in the East Bay, or knowing that a holiday matters in your market, requires knowledge of the place. That is a briefing problem, not a competence problem, and it is solvable if someone on your side supplies the local knowledge.

The details that trip up remote work are usually small and local. Something as ordinary as holiday hours on your profile can cost you ranking and customers if nobody is tracking your actual calendar, and a team working from a generic list of holidays will not catch it. Whitespark's 2026 expert survey put "business is open at the time of search" straight in at number five among local pack factors, which is a strong reason to treat the hours calendar as local knowledge rather than data entry.

The second variable is review capacity, which the arithmetic above prices exactly. Offshore models work when there is a clear spec and someone checks the output. If nobody on your side reads the pages before they publish, you will get volume, and volume of the wrong thing is worse than nothing.

There is one production risk that has become explicitly dangerous, and it is not about geography. The same 2026 survey added new negative and suspension risk factors, including the presence of AI generated text on the profile and AI generated photos or videos on the profile. Cheap volume in any time zone is now frequently machine written. Ask what is generated, ask what is reviewed, and keep generated material off the Business Profile specifically.

Which claim should make you walk, in any time zone?

An offer to build mentions of your brand across the web.

It arrives in several costumes: brand mention packages, digital PR at volume, community seeding, AI visibility placements. The mechanism sold is that language models notice repetition, so repetition can be manufactured.

The uncomfortable part is that the mechanism is real, which is exactly why the tactic is unsafe. Ahrefs ran the cleanest test of it available. They invented a luxury brand, Xarumei, confirmed the name returned nothing on Google, generated 56 adversarial questions with a model deliberately different from the ones being tested, and queried eight systems including GPT-5 Thinking, Claude, Gemini, Perplexity, Copilot, Grok and Google AI Mode. With only the official site live, the strong models answered nearly everything correctly and Perplexity failed a large share, at one point insisting the brand made smartphones.

Then they seeded three mutually contradictory fake sources: a blog, a Reddit AMA, and a Medium investigation. They also published an official FAQ on the brand's own site denying the claims. Perplexity and Grok became fully manipulated. Gemini and AI Mode flipped from skeptical to believing. The brand's own first party denial lost to the third party consensus.

That is the finding to carry into a sales call. A seeding program does not just fail to help, it establishes facts about your business that you then cannot correct from your own website. And it runs into three other walls: Google names seeking inauthentic mentions as ineffective and spam policed in its own AI documentation; a burned account gets a domain banned from the exact local subreddit you sell into; and for anything review shaped it crosses into 16 CFR Part 465, which since October 2024 carries civil penalties for fabricated reviewers and fabricated experiences.

Where the honest version lives: earned mentions on real local sites. Whitespark's testing of AI Mode local answers in May 2026 found reviews and unstructured citations, meaning blogs, news and community pages, among the heaviest inputs. That is the same asset, obtained slowly. Mentions on niche local sites and community blogs do work that directory listings do not, and they are hard to build from a distance, which is a genuine argument for local that does not require claiming technical superiority.

What does a freelancer risk that an agency does not?

Continuity. That is the whole difference, and it is a big one.

A good freelancer is often the best value available. You get a senior person doing the work directly, with no account management layer and no overhead priced in. For a defined specialty, that is hard to beat.

Then they take a full time job, or get a large client, or have a family emergency in the month you were launching. There is no bench. Everything stops.

Manage it rather than avoid it. Ask how many clients they carry and what happens when they are unavailable. Get access held in your accounts, not theirs. Keep the work documented so a replacement can pick it up. If you do those three things, the model is sound.

What does local actually buy?

Four things, and they are narrower than local agencies like to claim.

Physical presence. Photography of your real space, training your staff in person, watching how customers move through the room. If those matter to your business, hire someone who can drive to you.

Same hours availability. Not faster work, just aligned hours, which is the 26 days above.

Local knowledge that is hard to brief. Which streets people mean when they say downtown, which neighborhood publication actually gets read.

Accountability by proximity. Vendors who might run into you at a chamber event behave differently. That is not a noble mechanism, but it is a real one.

What local does not buy is technical superiority. Anyone claiming that is making a claim you should test, and most claims agencies make can be traced back to something checkable.

How should I decide?

Start with your own review capacity, because that constraint is real and it is yours.

If you have an hour a week to check work and give direction, offshore or freelance production can work well and will cost you less. If you have no time at all, buy coordination, and a local agency or a managed team is worth the premium.

Then check the work type. Repeatable and specifiable goes offshore comfortably. One deep specialty goes freelance. Coordinating five things at once goes to an agency of some kind.

Then check whether anything requires a body in the building. If yes, that part is local regardless of where the rest sits.

Most engagements end up mixed anyway. A local agency subcontracting production is a normal arrangement, not a secret failing. What matters is disclosure.

What to do this week

Write down two numbers. Hours per week you can spend reviewing work, and hours per month you need someone available inside your business day.

Then run the two calculations above with your own volumes. If the review hours you have are smaller than the review hours the scope demands, you have found your model before you have read a single proposal.

Then ask every candidate the question that a small local agency should find awkward, and that includes us. Which parts of my work leave your building, who does them, and who reads the output before I see it? SEOD is a small shop that uses contractors for some production, so we have to answer that too. An agency that will not name the parts and the reviewer is telling you the review step does not exist.

Finally, ask any candidate how they handle approvals when you do not fully follow what is being proposed. There is a short way to approve work you do not understand without pretending you do, and how a vendor responds to that question tells you how the next year will feel.

Questions about which model fits your business?

Email me at eric@seod.com with two things: how many hours a week you can realistically spend reviewing work, and whether anything you need requires someone physically at your location. I will tell you which model your capacity can support, including when the answer is an offshore team or a freelancer rather than an agency. No pitch and no followup.

There is more on hiring and working with agencies if you are still deciding.