← Back to blog

AI & Content · October 7, 2026 · 8 min read

How to Translate Your Blog Posts Without an Agency

Machine translation plus a paid native reviewer translates a 1,400-word post for $15–$40, against an agency's $140–$350 — here are the tools to buy and where an agency still wins.

By FluxWriter

Translating your blog posts without an agency is the right call for most small sites, because agency rates price a 1,400-word post like a legal contract. The owners who get burned swap the agency for a plugin and publish raw machine output unread. This guide covers what each model costs, the tools worth paying for, the review you cannot skip, and when an agency still earns its fee.

Why the Agency Quote Is Actually Priced for the Wrong Product

The obvious route is to send the posts to the agency that translated your website. That route fails on price before it fails on quality, because per-word translation was built for a different product.

Professional human translation runs $0.10–$0.25 per word, so a 1,400-word post costs $140–$350 per language. That rate makes sense for a contract or a manual, where a wrong word has a legal cost. A blog post is a speculative asset. It may earn nothing for six months and then get rewritten anyway.

The quote also leaves out the one check a blog post needs. A translator carries your answer across — nobody on the invoice checks whether the query exists on the other side. In French, the dictionary word for cheap is "bon marché", and almost nobody types it. Searchers type "pas cher". Same page, wrong door.

Some posts should not travel at all. A comparison of meal-kit services that only deliver inside the US has no search demand in Spain, however well it reads. In a US-centric library that bucket can easily hold a fifth of the posts — and an agency will translate every one of them at full rate.

What Each Model Costs Per Post

Price per post is the number everyone compares first. On its own it is a poor guide, because the cheapest row in the table below is the one that can cost you the traffic you already have.

Here is what each production model runs per 1,400-word post, per language, at 2026 market rates:

That last row is the trap. It looks free — and it is the only model that can lose you traffic, because a library of unreviewed pages is exactly what Google's scaled content policy targets.

The two middle rows are where most small sites belong. Machine translation with a native post-edit costs roughly a tenth of the agency rate and keeps a human in the chain. Drafting natively in the target language costs less again, because the reviewer is checking a post written for that market rather than repairing one bent to fit it.

Turnaround matters more than it looks. A 20-post batch typically comes back from an agency in 2–6 weeks. At post-edit pace the same batch ships in two weeks, which is the difference between measuring results this quarter and next.

The Three Tool Categories Worth Paying For

Three categories cover almost every small-site setup, and you need one from each.

Translation layers. Weglot, WPML, TranslatePress and Polylang add the language versions to WordPress and handle the hreflang tags that point Google at the right version for each market. Weglot is good — it is the fastest route to a working site. It is also metered by word count: roughly $15–$20 a month buys 10,000 words in one language, and 40 posts in three languages is about 170,000 words, nearer $80 a month. Menu and footer strings count toward that meter.

WPML runs roughly €39–€99 a year with automatic translation sold as credits on top. TranslatePress and Polylang have usable free tiers and paid tiers around €89–€199 a year. Budget an afternoon for setup rather than an hour. Cheaper every year after.

Translation engines. DeepL and Google Translate do the first pass. DeepL tends to read more naturally in the major European languages, and its paid plans start around $9–$11 a month. Google covers far more languages and bills around $20 per million characters, which prices a whole post in cents.

Native drafting. The newer option skips translation entirely. You hand an AI writer the brief and the target keyword, and it writes the post in Spanish or German. Big European languages need the 20–30 minute read-through. Arabic or Thai needs the full 60-minute post-edit, with whole paragraphs rewritten.

The Review Step You Cannot Skip

Machine output without a native reader is the one corner that costs more than it saves.

Google's spam guidance has listed automated translation published without human review as low-value generated content for years, and the 2024 scaled content abuse policy widened that net. The exposure is the whole site, not one page — 40 unreviewed translations can drag down rankings that were doing fine in English.

The review is cheaper than most owners expect. A native speaker on a freelance marketplace charges $15–$40 an hour. A read-through of a 1,400-word post takes 20–30 minutes, and a proper post-edit — fixing sentences instead of flagging them — takes 45–60 minutes. That is $5–$40 per post per language, as its own budget line.

Buy it as a brief with four jobs on it. Confirm the title uses the term people in that market actually search, not the dictionary equivalent. Swap units, currencies and source-country examples. Fix anything a local would find odd. Kill any post that should not exist in that language at all.

The fourth job saves the most money. A reviewer who kills five pointless translations a month has saved you $175 in production alone.

Which Languages Are Worth the Money

Let the evidence choose the language. Five languages with three posts each is a vanity project — one language with 20 posts is a batch you can actually measure.

Open the Countries tab of the Performance report in Search Console. If a country you do not write for is already sending 5–15% of your impressions, that is your first language, because demand is already proven. If nothing shows up, check the head term of your five best posts in a keyword tool set to the target country. Under 100 searches a month for the translated term is a skip.

Then run the arithmetic. Twenty posts in one language at an all-in $35 is $700, plus $15–$80 a month for the translation layer. The same 20 posts through an agency runs $2,800–$7,000. Both libraries take the same 6–12 months to rank — only one of them is cheap enough to be wrong about.

Start with the posts that already rank, however old they are. Ranking in English has proven the topic, which is more than a post published last week can claim.

When an Agency Is Still the Right Call

Agencies are good at what they were built for. Some of your pages are exactly that.

Anything with legal or medical weight belongs with a professional translator — but do not assume the invoice moves the risk. Before you brief one, ask what their terms say about errors and whether they carry professional indemnity cover; most contracts cap liability at the fee paid, and the published page stays your responsibility either way. So does anything a customer signs or pays on — the checkout and the terms page. A wrong word on a refund policy costs far more than a click.

Some markets punish tone errors harder than others. Japanese business register and the formal-versus-informal address in German are two places where machine output reads as foreign even when it is technically correct — the dialect question in Arabic is a third. If your brand depends on tone, that is where the agency money belongs.

Best for: regulated content, transactional pages, and any market where you have no native reviewer of your own. For the blog itself, an agency at $140–$350 a post is buying certainty you do not need on a page you will refresh in 18 months anyway.

FAQ

Will Google penalize me for using machine translation?

Not for the machine part — for the unreviewed part. Google's guidance names automated translation published without human review as spam, and a large batch of such pages counts as scaled content abuse. A machine first pass is fine once a native speaker reads every post before it goes live.

What happens to the translated pages if I cancel the plugin?

It depends on where the translations live. Weglot stores them on its own servers and serves them through the plugin, so cancelling takes the language versions offline unless you export them first. WPML, TranslatePress and Polylang write translations into your WordPress database as pages you own, and they stay up after the license lapses.

Can I translate my whole back catalogue at once?

You can, and it is where most multilingual budgets go to die. A 200-post library in three languages is 600 pages to review, and the ones nobody reads in English will not read better in Spanish. Translate the 20 posts that carry most of your traffic first, measure at 90 days, then decide.

The Practical Takeaway

Pick one language. Take the market already sending 5–15% of your impressions in the Search Console Countries tab, or require 100 or more monthly searches for your top five head terms in a keyword tool set to that country. Choose the 20 posts that already rank, budget $15–$40 per post per language all in, and book a native reviewer for 20–60 minutes per post before anything goes live. Measure clicks from that market at 90 days and expand only if the first batch earns them. Start with your five best-performing posts this month.

FluxWriter belongs in the native-drafting row: each connected site has its own output language, and posts are written in that language from the start rather than translated from English — which changes the draft, not the need for a native reviewer.

← All posts