x-default still points at the .com.ph homepage while the UK folder and the US subdomain fight the same English query in Search Console country. That is the failure. The new landing pages are not the problem.
We sit in the Philippines. The brief is Britain, the States, sometimes both, and the person sending it is the international marketing lead who has to keep clusters from eating each other. International seo from here only works if the cluster, the hreflang, and the market English are decided before anyone opens a CMS.
This is the briefing, not a tour of the office.
The cluster is wrong before the copy is wrong
English is not a market. en-GB and en-US are markets. A PH blog post written in fluent English will still rank in the US property if that is the only strong URL Google has for the query.
We see it in Search Console country every week. United States selected, and the top page is a Manila thought-leadership URL with a peso phone number in the footer. The title is fine. The intent is British or American. The hostname is not.
Fix the cluster first. Hostname, folder, or ccTLD. Then hreflang. Then copy. Reverse that order and we spend a quarter rewriting pages that should never have been eligible in that country.
If the UK and US URLs share a title template, they are one page as far as the index is concerned. Market English will not save them.
Hreflang seo that survives the next crawl
Hreflang seo is reciprocal or it is noise. Every en-GB URL must point at its en-US twin and at x-default, and those twins must point back. A one-way annotation is how the PH homepage keeps leaking into United Kingdom.
We implement in this order: sitemap, then HTML head if the CMS will not fight us, HTTP headers only when the HTML is a PDF or a file the template cannot touch. Mixing all three on the same URL is how annotations drift after the next theme update.
Codes are the boring part that still breaks launches. en-UK is wrong. en-GB is right. EN-us is wrong. zh-CN is not a substitute for a proper regional pair. Return tags on redirects, noindex, or 404s are treated as missing.
Google localized versions and hreflang is the rule set we actually follow for return tags, x-default, and the HTML, HTTP and sitemap methods. We do not invent a fourth method because a plugin offered a checkbox.
x-default is a choice, not a dumping ground. If there is no language picker, x-default is the primary commercial market, not the .ph home because that is where the CMS lives.
After ship, we fetch the live HTML, grep the annotations, then wait for Search Console to stop listing the pair as invalid. Invalid hreflang in the report is not a warning to ignore. It is the cluster still competing.
Market English is a brief, not a dialect plugin
Market English is spelling, units, proof, legal, and the SERP the person in that country actually sees. Colour and color are the easy tell. They are not the brief.
UK pages need GBP, VAT language, postcodes, Companies House tone where it matters, and service pages that match how British queries are written. US pages need USD, state-level reality where it exists, zip codes, and claims that would survive an American scan of the same query.
PH English is a third thing. It is often excellent and still wrong for both markets. Sentence rhythm, hedging, and which noun takes the heading all leak. A British-led edit pass is not optional on commercial templates. It is the difference between a page that looks translated and a page that looks like it was written for that Search Console country.
We keep a term list per market. Not a glossary of 400 words. The 40 that appear in titles, H1s, and FAQs. Aluminium, solicitor, lift, boot, aubergine stay off US templates. Elevator, attorney, trunk stay off UK ones. Currency symbols in titles are a ranking and a trust problem when they are wrong.
Do not brief “make it more international.” Brief the market, the query set, and the forbidden PH tells.
Search Console country is the only scoreboard that matters
Global performance is a vanity roll-up. We work in Search Console country.
Separate properties where the hostname is separate. On a folder setup, the same property, then country and page filters until the UK folder and the US folder stop sharing queries. If they share, hreflang is not doing its job or the copy is still one template.
Look at query × country × page. That three-way cut is where cannibalisation shows. A US query landing on /gb/ is the briefing failure, not a content gap.
International targeting in Search Console is gone as a setting people still argue about. Country in the performance report is not gone. Use it. Compare United Kingdom against United States against Philippines. If Philippines is winning English head terms you sell in London, the site architecture is still PH-first.
We screenshot that view in every monthly. Not because it is pretty. Because it is the only shared artefact that stops a debate about “the English site.”
ccTLD, subdomain or folder for the next market
ccTLD is the cleanest country signal and the heaviest ops load. Separate Search Console, separate CMS risk, separate certificates, separate everything. We use it when the business already trades as a local entity and the domain is not a trophy.
Subdomain can work. Google has crawled them as separate hosts for years. They still inherit link equity poorly if the root is a .ph and nobody linked the subdomain on purpose.
Folder is what most briefs actually need. One host, one CMS, hreflang between /gb/ and /us/, x-default decided in writing. Folders fail when the template still injects a PH phone number into every locale and when the XML sitemap is one blob with no locale split.
Pick one pattern and live with it for a year. A folder for UK and a subdomain for US is how annotations rot.
Whatever you pick, the live URL in hreflang must be the canonical, indexable, 200 URL. Staging, HTTP, and www mismatch are still the top three ways a Philippines-based ship fails a UK crawl.
UK and US English are not interchangeable
Britain and the States are the two live markets we actually run as separate commercial work, not as a spelling toggle.
When the brief is Britain, it is seo services uk with en-GB, GBP, and a Search Console country view that is not the .ph property. That cluster has its own title patterns, its own proof, and its own internal links. It does not borrow US case studies and swap the currency.
When the brief is the States, it is seo services usa with en-US, USD, and a hostname Google already associates with United States. US pages do not cite British regulators as if they bind a Texas buyer. They do not use “optimisation” in the H1 and then wonder why the snippet looks off.
Both can live on one technical stack. They cannot share a content ticket. If the same writer is asked to “do the English version,” we will get one page and two hreflang lies.
Demand is not symmetric. A query that converts in Manchester can be a research query in Ohio. We check Search Console country before we clone a page. Cloning is how folders fill with thin locale copies that never earn a click.
What we need in the briefing pack
Send the markets, not a brand deck.
We need: live hostnames, the chosen pattern (ccTLD, subdomain, folder), current hreflang method, Search Console access per property, a list of URLs that must win in each country, and the pages that should never rank there. We need the commercial difference between UK and US in one paragraph each, written by the person who owns pipeline, not by the CMS vendor.
We need crawl access from outside the office network. A site that only looks right on Philippine fibre is not ready for Googlebot fetching from the US.
We need the forbidden claims. If legal will not allow a US page to say a thing the UK page says, that goes in the brief before draft one. Hreflang will still pair them. The lawyer will still hate the US URL.
Timezone is not a briefing item. Overlap is. British-led review on commercial templates happens on UK hours. US legal review does not. Plan the ship date around the slower of the two, not around Manila finishing the build.
If white-label is the arrangement, say who the public author is. EEAT for a US medical or legal adjacent page will not be carried by a PH byline the market has never heard of. That is a content architecture decision, not a staff-page afterthought.
Hosting, bots and who actually crawls Manila
Googlebot does not live in our office. Latency to a UK user is a ranking and a conversion problem, but the first failure is still crawl.
A PH origin with no CDN, no geo routing, and a certificate that mis-matches the locale host will waste the hreflang. We put the locale hosts on a CDN that serves from the market, then confirm the HTML Googlebot receives is the locale HTML, not a cookie-based guess.
Cookie-based language is hostile to international seo. The URL must change. Hreflang cannot annotate a single URL that paints different English depending on IP. IP detection as the only switch is how US Googlebot gets the GB page and the annotations look “fine” in a Manila browser.
We test with a US and a UK fetch, not with a VPN screenshot in Slack. View-source, annotations, canonical, hreflang, currency in the JSON-LD. If the JSON-LD still says PHP, the template is still PH.
AI search work sits on the same cluster logic. A citation that quotes the PH English page for a US query is the same failure as a blue-link ranking. Clean locale URLs and market English reduce that leak. They do not require a separate product name.
When not to add another locale
Another locale is not progress if the first two still share queries.
Do not add en-AU, en-CA, or a second European English until Search Console country shows the UK and US URLs owning their own query sets. A third folder of thin copies will not impress a crawler that already ignored the second.
Do not add a locale because sales opened a conversation in that country. Indexable pages without demand, without ops, and without a person who can approve market English, are how international seo turns into a graveyard of 200s.
Do not migrate and localise in the same window. Migrations need a map, 301s that preserve locale, and hreflang rewritten to the new live URLs on day one. We have watched a UK folder 301 into a US folder because someone used a global find-and-replace on “en.” That is a week of recovery, not a lesson.
If the CMS cannot output reciprocal annotations without a developer, stop the content plan and fix the CMS. Copy will not compensate.
The next useful step, then a scoped audit
Open Search Console country. United Kingdom. United States. Philippines. Export query × page for each. If the same URL appears in the top twenty for both English markets, the next job is cluster and hreflang, not more articles.
Then fetch three live commercial URLs per market and read the annotations. Reciprocal, valid codes, x-default that is a real URL. If that fails, we do not brief writers.
If the clusters are clean and the English is still PH, the next job is a market English pass on the money pages only. Not the blog. The pages that already attract the wrong country.
When that cut is honest and the fit is a Philippines-based team under British lead, send the briefing pack and ask for a scoped audit on the hosts you actually want to win. The audit should name the pattern, the annotation method, the Search Console properties, and the first ten URLs per market. Anything wider than that is a tour. We do the tour after the ten URLs are decided.
