Your Homepage Is in Japanese.
The Page They Have to Forward Isn't.

▶ 60-Second Slide Summary Prefer the 8-slide version of this article? Tap to read in 60 seconds. View slides →

Quick Answers

Why does a Japanese evaluation go quiet after a good demo?

Because in most Japanese companies the person running the evaluation is not the person who approves the purchase. They have to build the case internally — for procurement, legal, IT security and their own manager — out of the material you gave them. If the pages that case depends on are in English, they cannot be forwarded, and your champion is left doing unpaid translation work under their own name. Most people decline that. They wait instead, and the deal stops without ever being declined.

Do we have to localize our entire help centre and legal documentation?

No, and attempting that is how this gets postponed indefinitely. Only a small set of pages actually travels inside a Japanese buyer's company: the few help articles covering the workflow being evaluated, the terms, the data-handling and security page, and whatever your support team last replied with. Find that set and make it genuinely native. The long tail can wait.

TL;DR

A Japanese evaluation goes well, then goes silent. Nobody objects, nobody asks for a discount, and nobody says no. What has usually happened is procedural rather than commercial: the evaluator is not the approver. To move, they have to carry your product into procurement, legal, IT and their manager's inbox — and they can only carry what they can forward. Your homepage is in Japanese. The help article answering the actual integration question, the terms, and the security and data-handling page are not. So the internal case never gets assembled, and you never hear why. The fix is not translating everything. It is identifying the eight or ten pages that get forwarded, and making those read as though they were written in Japanese to begin with.

Key Takeaways

• The evaluator is rarely the approver. In most Japanese B2B purchases, the person testing your product has to convince procurement, legal, IT and a manager before anything is signed.
• They can only advocate with what they can forward. An English page is not a document that survives being pasted into an internal request.
• Your localized surface stops where the real questions start. Homepage and pricing get translated first. Help centre, terms and security documentation almost never do.
• The outcome is silence, not rejection. A stalled evaluation returns no signal at all, which is why this stays invisible for years.
• The forwarded set is small. Eight to twelve pages in most cases. Localize those properly before anything else.

The Evaluation That Never Said No

Here is a shape that repeats. The demo goes well. The Japanese team asks sharper questions than your other prospects do — about data residency, about what happens to logs, about an edge case in your permissions model. You give them a trial account and they actually use it, properly, for two weeks. Then someone says they would like to take it internally, and the conversation ends.

Your follow-ups get answered, courteously, with some version of "we are still checking internally." Three months later the champion has moved to another team, or the budget went somewhere else. There is no loss reason to record, because nothing was lost. The evaluation simply stopped.

That sequence is a composite of a pattern I see repeatedly in Japan-entry work — it is not a description of a specific client.

In Japan, the Evaluator Is Rarely the Approver

In a lot of Western pipelines, the person running the evaluation either has authority to buy or sits one conversation away from someone who does. In Japanese companies of almost any size, that is usually not the arrangement. The evaluation is one input. The decision gets assembled — written up, circulated, approved in sequence through the ringi (稟議) process, and typically preceded by informal groundwork with each stakeholder before anything formal appears, which is why disagreement in Japan tends to happen where you are not in the room.

That changes what your content is for. In a self-serve market, the job of your documentation is to answer the reader's question. In Japan, the same documentation has a second job that nobody briefed you on: it has to be forwardable. It becomes an attachment, a pasted paragraph, a link inside an internal request that four people who have never visited your website will read and have to feel comfortable signing off on.

Almost none of the localization work that gets done first is aimed at that second job.

The Forwarding Test

There is a single question that sorts your Japanese pages into the ones that matter and the ones that are decoration: if my champion had to send this page to their legal department this afternoon, could they send it as it is?

Run that across a typical "localized" site and the answer changes abruptly at the point of conversion. Everything up to the demo form passes. Everything after it fails. And everything after it is what an approver reads.

In practice the forwarded set is fairly consistent:

• The two or three help-centre articles covering the specific workflow being evaluated — not your getting-started page, but the awkward one about SSO, data export, or an API rate limit.
• The terms of service and the privacy policy: the pages legal actually reads line by line.
• The security and data-handling page — where data is stored, which subprocessors you use, what you are willing to sign. This is the one IT forwards, and it is often the first thing requested.
• The last reply your support team sent them. Frequently the single most-forwarded artefact in the whole deal, and almost always in English.
• Anything containing a number that will end up inside a budget request, alongside the corporate information a vendor registration form demands.

An English Page Doesn't Get Rejected. It Gets Postponed.

Picture your champion holding an English security page and facing a Japanese approval chain. They have three options.

Forward it as it is. Four people who did not ask to be involved now have homework. The request arrives looking unfinished, and — this is the part that is easy to miss from headquarters — it makes the person circulating it look as though they have not done their own work.

Translate it themselves. Unpaid, in the evening, and at the end of it they are personally vouching for the accuracy of your security and contractual commitments in wording you did not write and cannot check. Real exposure, no upside.

Wait. This costs them nothing at all. So this is what happens.

Notice that nobody in that chain ever frames it as a language problem. From the inside, "we would need Japanese documentation for this" is not an objection worth raising with a vendor. It is simply an internal reason to leave the request unwritten a little longer, and then a little longer than that.

It is not really a question of English ability either. In CSA Research's 2020 survey "Can't Read, Won't Buy" — 8,709 verified respondents across 29 countries — 90% of Japanese respondents wanted product information in their own language, against 76% across all 29 countries. That is a consumer study, so it should not be dressed up as a procurement statistic; what it does establish is that the preference runs stronger in Japan than the global average. Inside an approval chain, that same preference has more force rather than less, because the reader is not deciding whether to buy something. They are deciding whether to put their name on a document.

You Don't Have to Translate Everything

The reason this problem survives for years is that the honest version of the fix sounds enormous. Four hundred help articles. Every legal page. Support in Japanese. So it gets scheduled for after Japan has proved itself, which requires solving the thing that was postponed.

But the forwarded set is not four hundred pages. It is usually eight to twelve, because a Japanese pipeline concentrates hard: the same three use cases, the same two integrations, the same security questionnaire, the same clauses that legal wants moved. Once you can name the set, the work is a fortnight, not a fiscal year.

A workable order of operations:

1. Security and data handling, plus whatever you are prepared to sign. IT and legal are the hardest gates and the earliest ones to appear.
2. Terms of service and privacy policy. Not a literal rendering — the register Japanese legal readers expect from 規約 is its own thing.
3. The five or six help articles your Japanese users actually open. Your analytics already know which ones. Ranked by traffic, not by where they sit in the navigation.
4. Japanese support replies, at least during evaluations. The reply is a document. It gets forwarded like one.

This is where the difference between translation and localization stops being a marketing distinction and starts costing money. Machine translation of a security page takes about ten seconds and will be broadly correct. It will also read unmistakably as machine output — and a page that reads that way is a page your champion will not attach to a request with their own name on the cover, because it reflects on their judgement rather than yours. Japanese readers withdraw trust from a page when nuance, grammar and line breaks have not been cared for. An internal approver is the least forgiving reader you will ever have, and the only one whose opinion decides the deal.

How to Find Your Own Forwarded Set

Three checks, none of which need new tooling.

Ask a friendly champion what they sent. Which pages did you forward internally, and did you have to explain any of them? People answer this readily, because it is not a negotiation question. The answer is frequently one page you had never thought of and one you assumed nobody read.

Read your Japanese support tickets for the requests you filed away. "Is there a Japanese version of this document?" tends to get logged as a low-priority content request. Much of the time it is a stalled evaluation announcing itself in the politest available form.

Look at where your Japan deals stop, not why they were marked lost. If a cluster consistently goes quiet between technical validation and contracting, that is the approval chain, not your product and not your pricing.

What the Silence Was Telling You

An evaluation that stops without a reason is the most expensive feedback you can receive, because it arrives looking like nothing at all. There is no objection to answer and no competitor to displace. The opportunity ages out of the forecast, someone writes "no decision," and the same thing happens to the next one.

Go back through last year's Japan opportunities and pull only the ones with no recorded loss reason. Then ask what each of those buyers would have had to forward, and in what language they would have found it. If the answer is English every time, you have not been losing on product. You have been asking a series of careful, well-disposed people to translate your documentation on your behalf, and they declined without telling you.

A Japan Readiness Check reads your Japanese site the way an internal approver reads it — following the path from first click to the security page and the terms — and reports where a buyer would have to stop advocating for you.

Frequently Asked Questions