Acquisition Board
13-agent research run · 72 live listings · 25 August 2026

Acquisition Board

Every buyable online business the fleet could actually load between USD 8,000 and 75,000, across fifteen Japanese and global marketplaces, scored against what you specifically can grow. Three findings below reorder the whole plan, and one of them is that the target needs resetting.

3-year median, full reinvestment$5,538/mo8,000 simulated runs. p25 $3,901, p90 $9,554.
Probability of hitting $20k/mo in 3yr0.0%On $106k of capital. Five to seven years is the honest horizon.
Entry multiple beats position count4×26.5x entry gives $3,914/mo at 3yr. 14.5x gives $14,067.
Flippa listings with verified revenue5.9%9% in the $8-20k band. Sellers type the profit in by hand.
The recommendation

Three assets at roughly $21,000 each, bought one at a time, deploying $63,000 and holding $43,000 back.

Not one large asset: a single $62,500 position has a 38% chance of ending below $1,000/month at month 24, against 16% for three. Not four or five: below about $21,000 the asset is too small for your repair work to be worth deploying. Sequence them four to six months apart, because reinvestment cadence sets the compounding rate, not the size of the first deployment. Target 14x to 18x monthly profit and treat entry multiple as the primary criterion, ahead of niche.

Binding condition: a $3,000/month personal draw cuts the three-year median from $5,538 to $2,031 and completed deals from 12 to 4. Living costs have to come from outside this portfolio for 36 months or it does not compound.

All candidates

72 live listings, filterable

Loaded from public listing pages on 25 August 2026. Prices and inventory move daily. Fit score is against your profile specifically: existing organic demand you could capture, transactional intent, Japanese-language assets where N1 is a moat, no audience needed, no staff, no support desk.

Region
Tier
Market
Min fit0
Category stance

What to hunt and what to refuse

The finding that contradicts the standard broker pitch: comparison and "best X" format carries a 95.4% AI Overview rate across 5.47 million measured queries, the highest of any format. Comparison affiliate is the most damaged category, not the safe middle ground. Local intent triggers AI Overviews on 15% of queries against 92% for informational.

huntLocal service lead generation and local directories (service plus city queries, phone or form conversion, ideally with recurring listing subscriptions)The most durable category on measured evidence, and the one where your citation builders, bulk content pipeline and index monitoring produce revenue directly rather than more unclicked informational pages. Whitespark measured AI Overviews on 15% of local-intent queries against 92% of informational, with local packs on 93% of local-intent queries; the relationship is inverse, and where the map pack is strong the AI Overview is weak. Seer measured 5% AIO on transactional intent and Semrush measured transactional AIO presence declining 5% over six months. Coverage scales mechanically by geography, which is your cheapest activity.
hunt一括査定 and 買取 bulk-quote lead generation in Japan (cars, property, cards, estate clearance done as pure lead capture)The highest-payout consumer verticals in Japanese affiliate marketing, with a conversion event that is a quote request rather than a purchase, so the intent barrier is low and the query set expands mechanically by product, model, year and prefecture. It sits on the transactional side of the intent ladder where AI Overview presence is lowest and falling, and it is Japanese-only so the bidder pool excludes almost everyone. Verified live examples in this harvest: SiteStock 118807 at ¥83,533/month on 43 pages, Site-Z 1409 at ¥135,000/month on 165 articles.
huntRecruitment and placement affiliate where the audience is already present and the payout layer is missing (driver, nurse, new-graduate, welfare-sector careers)Japanese 人材 pays among the highest per-conversion rates in the country, intent is late-funnel (a student searching a named company's selection process is days from converting), and the harvest repeatedly surfaced properties with real audiences monetised at almost nothing: 40,000 monthly page views earning ¥10,000, 300 articles on a 2010 domain running one display contract. Closing that gap is a monetisation rewrite that pays from month one rather than a growth project. Weigh the seasonality: 就活 peaks October to March, nurse and driver do not.
huntAssets whose seller is motivated, distressed or has stopped publishing, at 14x to 18x monthly profit, in any of the hunted categoriesThis is the highest-leverage instruction in the whole document and it outranks category choice. Empire Flippers' own scoreboard prices distressed at 14.5x against 26.3x typical, a 45% discount. Modelled three-year medians by entry multiple: 26.5x gives USD 3,914/month, 22x gives USD 5,584, 18x gives USD 8,344, 14.5x gives USD 14,067. Entry multiple moves the outcome roughly four times more than position count. A property that took an algorithm hit but still holds its links and topic authority is the same shape as the five engagements you have won on.
considerProgrammatic and database directories with a proprietary dataset and geo or entity segmentationStructurally attractive because the moat is the dataset rather than the prose and the pages scale at near-zero marginal cost, and index-and-rank gaps are squarely technical SEO. But the two live examples in this harvest both carry disqualifying detail: the 緊急避妊薬 pharmacy database has halved on both revenue and traffic in six months and sits in a regulated advertising category under 医療広告ガイドライン and 薬機法, and the AI tools directory at DR 71 on a fourteen-month-old site has a bought-link signature in the most saturated category of 2026. Hunt the shape, screen the specific asset hard.
considerAged authority domains bought for their link equity rather than their current earnings (androlib.com DR 63 at USD 10,709 is the live example)A genuine seventeen-year MOZ DA 63 domain with both Flippa verification badges for USD 10,709 is link equity you cannot otherwise buy, and 9,031 page views on that authority means you are buying a floor rather than a trend. It only works if the backlink profile is editorial rather than bought and the domain has never been dropped or repurposed. Run the full backlink and manual-action audit before treating the authority as real, because a seventeen-year-old tech site with DA 63 and near-zero traffic is the classic profile for links that were purchased.
refuseComparison and 'best X' affiliate content, including telecom, hosting and product comparisonThis is the finding that contradicts the standard broker pitch and it should change what you buy. Seer measured 95.4% AI Overview presence on comparison-format queries across 5.47M queries, the highest of any format, higher than question format at 85.9%. Aggregate commercial intent at 8% masks the fact that the comparison format specifically is where Google most wants to synthesise. Semrush separately measured AIO expanding 71% into commercial intent over six months, with finance commercial at 231%. Buying comparison affiliate today means buying the most-targeted format at a price that assumes it is protected. This downgrades Saitoma 30271, BATONZ 83045 and the Swiss and Danish telecom sites in the harvest.
refusePure informational and question-format content (recipes, how-to, symbol and shortcut references, exam facts, single-fact database lookups such as 校名+偏差値)The melting ice cube in its purest form. 99.2% of AI-Overview-triggering keywords are informational, and measured position-one CTR decline on exactly those keywords is 58% globally and 62.7% in Japan as of July 2026. Pew's behavioural panel measured click rate on a traditional result at 8% with an AI summary present against 15% without, with 1% clicking a link inside the summary. howtotypeanything.com at 29x, the recipe sites at 35x, and 校名+偏差値 lookup pages are all the same trade at different prices. No operator skill reverses a SERP feature.
refuseMicro-SaaS under roughly USD 20,000Category average monthly churn is 3.5%, which resets about 35% of the revenue base every twelve months, and above 8% the base resets annually. That converts an owned asset into a continuous demand-creation task, which is the one activity with a 0-for-4 record across roughly USD 42,000 of spend. It also means a support desk, which is a stated hard constraint. Roughly 90% of Acquire.com's in-band inventory is micro-SaaS, with listings logged at 84x, 97x, 500x and 1,111x TTM profit, so the category will dominate any global search and should be filtered out mechanically.
refuseNewsletters, paid communities, YouTube channels and social accountsThey are audience-and-personality assets and there is no audience and no personal brand to substitute. Paid newsletter churn averages 4% monthly and reaches 16.67% in the money vertical, the exact niche with the best economics, so the list must be substantially rebuilt every year by the person whose name is on it. This matters practically in Japan because 47.04% of Rakko M&A's trailing-year listing supply is YouTube channels and 12.38% is SNS accounts, so 59% of the board is inventory you cannot use and effective supply is roughly half what it appears.
refuseEcommerce, dropship and physical product (including supplements)Orders, returns, customer service and supplier management are operations, which is the opposite of the owned asset requested, and dropship margin depends on supplier terms that may not transfer. The Rakko apparel listing at USD 69,128 has the highest absolute cash flow in the Japanese set and no SEO moat at all, so it competes purely on paid-acquisition efficiency against better-capitalised operators while consuming two thirds of liquid capital. Supplements add regulatory exposure on claims. Nine of the twelve in-band Empire Flippers listings are FBA, physical product or YouTube for the same reason.
refusePaid-traffic arbitrage sold as a website (native ads, listing ads, X arbitrage) and downloader or extraction toolsWith paid traffic there is no owned asset: the business stops the day the spend stops, and a '94% margin' on a paid-traffic business usually means ad spend is excluded from the profit figure. TRANBI 20042 states outright that organic is small-keyword only and domain authority is weak, and over 90% of its revenue is one legally sensitive advertising type with 非弁提携 exposure. Downloader sites (Toolzin at 1.6M verified monthly page views and a USD 1.23 RPM) are one DMCA campaign or one platform change from zero, and the low RPM exists precisely because premium ad networks refuse the category, which means the monetisation thesis is not executable.
refuseLink-selling and sponsored-placement publishers (Atamgo at USD 34,559, 0.8x annual, USD 3,776/month is the live example)Considered seriously and rejected. It is the highest absolute profit in Tier B at the lowest multiple in the harvest, and it is the one business model where your day job is the growth plan, since you know from the buy side what placements are worth and which vendors are real. But selling links violates Google's guidelines explicitly, a manual action takes it to zero overnight with no warning and no recourse, 8,572 uniques means buyers are paying for a metric rather than traffic, and revenue in this model typically concentrates in three or four agencies. For a first-time acquirer putting a third of deployable capital into one position, that is the wrong risk to take even with an information edge.
refuseService businesses, brokerages and anything with outbound calling, account management, staff or premisesA stated hard constraint, and the harvest shows these listings disguised as media assets. TRANBI 22777 looks like a pharmacy media site with near-100% margins and is actually a call list of tens of thousands of records plus call scripts, with 有料職業紹介事業許可 licensing exposure flagged by TRANBI itself. Rakko 22390 advertises ¥50-60万/month profit in the past tense, was rebuilt on a new domain in July 2026, has produced exactly 2 leads since, runs on Google Ads, and requires phoning inbound enquirers to qualify job size. Read the detail page before believing a summary line.
Honest risks

What could go wrong

THE BIGGEST ONE, said directly: this research is a comfortable way to avoid the harder problem. Acquisition hunting has all the surface features of productive work (spreadsheets, multiples, boards, diligence checklists) and none of the exposure. It can absorb months without a single dollar changing hands, and at the end of it you would have learned about the market rather than owned anything. The evidence that the risk is live is in this document: four research agents screened roughly 7,000 listings across twelve marketplaces in one day, and the honest output is nine candidates, none of which is unambiguously good. If eight weeks from now the position is 'still looking, the inventory is weak', that is the failure mode, and the countermeasure is the dated commitment in next_actions rather than more screening.
SPENDING MOST OF YOUR CAPITAL ON AN ASSET YOU CANNOT VERIFY IS WORSE THAN DOING NOTHING, and the numbers say so. One position at USD 62,500 gives a p10 of USD 180/month at month 24 and a 38% chance of ending below USD 1,000/month, on an asset where 3% of in-band 50-75k Flippa listings have verified revenue and BATONZ's own listing pages warn that registered financial fields may hold provisional placeholder values. Doing nothing preserves USD 106k and optionality. Buying one big unverifiable thing converts most of that into a property you may not be able to fix, at a moment when the search channel underneath it is measurably contracting. This is the specific reason the recommendation is three at USD 21,000 sequenced rather than one at the ceiling.
THE TARGET IS UNREACHABLE ON THIS CAPITAL IN THIS TIMEFRAME, and hearing it late would be worse than hearing it now. Across 8,000 simulated three-year runs with full reinvestment and no personal draw, the median at month 36 is USD 5,538/month, p25 is USD 3,901, p90 is USD 9,554, and the probability of clearing USD 20,000/month is 0.0%. Extending to five years at a 22x entry gets the median to USD 6,436 and the probability to 0.1%. The only configuration where USD 20k becomes likely is 14.5x entry over five years, which needs roughly 39 completed deals, well beyond a solo workload. If the real objective is USD 20k/month rather than a durable owned asset, the acquisition route on USD 106k is the wrong instrument and that should be decided now rather than in year three.
DRAWING LIVING EXPENSES FROM THE PORTFOLIO DESTROYS IT. A USD 3,000/month draw cuts the three-year median from USD 5,538 to USD 2,031, cuts median deals completed from 12 to 4, and cuts the probability of clearing USD 5,000/month from 58% to 7%. The compounding engine is the reinvestment cadence, so the first three years of income has to come from outside this portfolio entirely. If it cannot, the plan does not work as designed and the target has to be reset before any capital is deployed rather than after.
THE BASE RATE FOR THESE ASSETS IS ROUGHLY A COIN FLIP, AND THE STUDY THAT MEASURED IT IS OPTIMISTIC. The Website Flip tracked 49 Motion Invest listings above USD 10,000 six months after listing: 24 declined, 20 increased, 5 flat, with 12 declining more than 50% and 4 losing over 90%. That cohort is small, single-marketplace, traffic-only rather than revenue, and it predates both the Helpful Content Update and AI Overviews. Applied to three positions it says one is likely to disappoint materially and one is likely to grow. The reserve exists to survive that, and any plan that requires all three to work is already broken.
THE SEARCH CHANNEL UNDERNEATH EVERY CANDIDATE IS CONTRACTING, INDEPENDENTLY OF ANYTHING YOU DO. Pew's behavioural panel of 900 US adults across 68,879 searches measured click rate on a traditional result at 8% with an AI summary present against 15% without, with sessions ending entirely after 26% of AIO searches against 16%. 71% of monitored affiliate domains declined after the March 2026 core update, and only 22% of roughly 400 sites in the HCU recovery study regained 20% or more of what they lost. There is a genuine counter-signal (Seer measured AIO organic CTR rebounding from a 1.3% floor in December 2025 to 2.4% in February 2026, an 85% recovery, and cited pages earn 2.07% against 0.94% uncited), so the honest read is drift rather than collapse. But an asset bought at 26x monthly profit has to grow just to hold its purchase value, and the model applies a continuous monthly bleed for a reason.
YOUR CONTENT PIPELINE IS CLOSER TO A LIABILITY THAN AN EDGE IF APPLIED NAIVELY. The highest-casualty categories named in the March 2026 core update tracking across 600,000+ pages were AI-generated content without substantive human editorial oversight, template-built comparison pages with no original testing, and programmatic aggregators. 25 posts a day lowers your marginal cost of publishing and does nothing for what Google now selects on; unedited, it is the exact fingerprint that got penalised. What it is genuinely worth is speed of hypothesis testing and the ability to run several properties at once, which is why it supports the many-positions answer rather than the content-volume answer. Two shortlist candidates (Rakko 22799, Rakko 21259) already carry declared AI content, so you would be inheriting that exposure rather than creating it.
THE TRANSFER TRAPS ARE WHERE YOUR ACTUAL RISK SITS, BECAUSE THEY DEPEND ON THIRD PARTIES RATHER THAN ON DATA. Your DataForSEO access largely neutralises the measurement traps: you can rebuild the traffic and ranking history without the seller's cooperation. It does nothing about Mediavine refusing your application, Amazon terminating on seven days notice, a closed Japanese ASP declining to register you, or a 特別単価 reverting on transfer. Rakko 22799's special rates are explicitly marked 引継ぎ可否不明, and those rates are plausibly most of its ¥234,018. A deal can pass every data check and still lose half its revenue at handover.
FIRST-DEAL PROCESS COST IS HIGHER THAN THE LEARNING ARGUMENT SUGGESTS. The expensive lessons (diligence process, escrow mechanics, analytics and ad-account migration, seller-handover discipline, how far a stated P&L sits from a bank statement) cost the same hours on a USD 15,000 deal as on a USD 60,000 one. Empire Flippers reports 126 days average from listing to sale and Acquire 81 days on market, and Rakko's median close is 60 to 90 days with a 35.26% success rate on listings older than four months. Three deals is roughly three times the process load for the same lessons. Three still wins, but it wins on variance reduction and on buying more current cash flow per dollar, not on learning.
THE CATEGORY THAT SURVIVES BEST IS THE ONE WITH THE THINNEST INVENTORY. Local service lead generation is the most durable thing measured, and the harvest found exactly one clean example (an Irish driving instructor directory at USD 23,403 with unverified USD 481/month revenue and a 4.1x annual multiple, the most expensive multiple in Tier B). So the correct category and the available inventory are not the same set, and there is a live temptation to buy the best available listing in a category you have decided to avoid. Watch for that specifically with the comparison-format assets, which are the best-looking listings here and the most exposed format measured.
CONSUMPTION TAX METHOD IS AN UNFORCED ERROR WAITING TO HAPPEN. If you are on 簡易課税 or the 2割特例, the 10% on the purchase price is money you never see again, ¥500,000 on a ¥5M site, and the election to switch must be filed before the start of the taxable period, meaning by 2026-12-31 for a 2027 purchase. This is entirely avoidable and entirely dependent on asking one question this week.
MARKET TIMING RISK ON THE JAPANESE BOARDS RUNS BOTH WAYS. Japanese sellers have not repriced for AI risk while Japanese search damage has now passed the global level, which means today's Japanese asking prices embed a risk premium that has not been taken. Either the market reprices downward over the next twelve months, in which case waiting was correct and buying now was expensive, or the informational damage stabilises as the Seer rebound hints, in which case current prices are fair. That variable is genuinely live and should be re-checked quarterly rather than resolved by assertion.
Next actions

Five steps, starting this week

01Send the 税理士 one email with five numbered questions and nothing else: (1) am I on 本則課税, 簡易課税, or the 2割特例 for consumption tax, and what does switching cost and by when must it be filed; (2) model 個人事業主 versus reactivating ムラサキイタリア株式会社 at ¥3M, ¥10M and ¥37M of annual net profit, including 個人事業税 and the KK's 均等割; (3) confirm 営業権 amortises 5 years straight-line for me as an individual and whether first-year proration is monthly; (4) if the KK buys, is the goodwill 資産調整勘定 over 60 months or 営業権, and please cite the checked provision rather than a remembered article number; (5) confirm the post-2026-10-01 免税事業者 credit percentage against enacted law rather than the 大綱.Wednesday 2026-08-26If the answer to (1) is anything other than 本則課税, the 10% consumption tax on every purchase is permanently unrecoverable, ¥500,000 on a ¥5M site, and the election cannot be made retroactively. For a calendar-year individual filer buying in 2027 the deadline is 2026-12-31. This is the cheapest and most time-sensitive item in the entire document and it gates everything downstream.
02Open buyer accounts on ラッコM&A and サイトストック while there is no deal pressure, and record exactly what each demands (Japanese bank account, 本人確認 documents, My Number). Separately, start the My Number Card application, since TRANBI holds the largest Japanese web-media inventory including listing 18129 and per prior vault research will not verify a foreign national on a residence card or passport. Subscribe to Saitoma's 未公開案件 newsletter, since Saitoma states its real book is private and its public board carried only four live listings.Friday 2026-08-28Nothing on either platform's public pages imposes a nationality or residency condition and neither addresses foreign buyers either way, so it is unstated rather than confirmed. Finding a residency block after identifying a target you want is the avoidable version of this problem. The Saitoma newsletter is the single highest-value sourcing action available, because per finding PC-002 entry multiple moves the three-year outcome about four times more than position count, and private inventory is where mid-teens multiples live.
03Run the free DataForSEO pre-contact screen on every shortlist candidate whose domain is knowable, and save every raw response to ~/obsidian-vault/Research/Online Business Acquisition Targets/ per the SEMrush archive rule. For each: ranked-keyword and SERP position history, backlinks_history and timeseries_new_lost, referring_networks, bulk_spam_score, bulk_traffic_estimation, plus a live AI Overview presence check on the top 50 money keywords through serp_google_organic_live. Where the platform gates the domain (Saitoma, Site-Z, Investors Club, Empire Flippers, BATONZ pre-registration), record that as the reason no screen exists rather than skipping the row.Friday 2026-08-28This is the one advantage held over every competing buyer, it needs no seller cooperation, and it disqualifies most candidates before a single conversation. It also produces the AI Overview exposure number that should reorder the shortlist: any candidate already at 80%+ AIO presence on its money keywords has its decline ahead of it and its trailing revenue is the last clean figure it will ever report.
04Contact sellers on exactly three candidates, no more, and lead every conversation with the four questions that kill deals fastest rather than with price. Recommended three: Rakko 22799 (ask first whether the 特別単価 marked 引継ぎ可否不明 will transfer, in writing from each ASP), SiteStock 118807 (ask first whether ドメイン is included in 譲渡対象物, in writing, and walk if it is not), and the Xpert Driving Lessons directory (ask first for the paying-instructor count, churn, and revenue concentration by account). The four standing questions for every seller: your 適格請求書発行事業者 registration number, will you do a live screen share of every revenue dashboard, will each ASP and ad network confirm my eligibility in writing, and what exactly is in the asset schedule.Friday 2026-09-05Three is the number that keeps the diligence honest; screening more listings is the avoidance behaviour named in honest_risks. Each of these three has one specific disclosed fact that can end it in a single message, so leading with that fact converts a week of work into a day. A refusal to screen-share is the cheapest disqualifier available, and with verified revenue at 5.9% across 934 in-band listings it is close to the only real financial check that exists at this price.
05Complete diligence on the surviving candidate and either sign a first purchase of ONE asset at USD 20,000 or below, or write a dated note recording why none of the three cleared and what specifically would have to be true for the next one to. Structure it with platform escrow, an inspection period negotiated up from the 3-7 day default to at least 30 days, a 20-30% holdback released after 60-90 days of verified traffic and revenue, no earnout, an electronic signature so 印紙税 is zero, and a condition precedent making completion contingent on your own ad-network and ASP approvals. Then run it for four to six months before buying the second.Tuesday 2026-09-30Rakko's median close is 60 to 90 days and Empire Flippers averages 126 days listing-to-sale, so a signature by end of September means starting now. The one-at-a-time sequencing costs nothing measurable over three years (the deploy-three-now and deploy-two-now paths land within USD 5/month of each other at the three-year median) while removing the risk of making the same diligence error three times in one month. The written no-go note matters as much as the purchase: it converts a stall into a decision and it is what stops this becoming a research project.
Diligence

The checks that matter, in order

Weighted to the cheap end, where verification is weakest and first-time buyers get robbed.

011. FREE PRE-CONTACT SCREEN, before messaging any seller. Run every candidate through DataForSEO before spending a minute on conversation: labs_google_ranked_keywords and historical SERP position to rebuild the organic curve from Google's side, backlinks_history and backlinks_timeseries_new_lost_summary to date the link acquisition, backlinks_referring_networks and backlinks_bulk_spam_score to detect PBN structure, and labs_google_bulk_traffic_estimation for an independent traffic figure. Then Wayback for niche consistency and WHOIS for a registrant change in the last 90 days. This costs API credits and no seller cooperation, and it disqualifies most Tier A listings before first contact. Save every raw response to the vault per the SEMrush archive rule. This is first because it is the one advantage you hold over every competing buyer.
022. INTENT CLASSIFICATION OF THE REVENUE-DRIVING KEYWORDS. Pull the money keyword set, classify intent, and reject anything above roughly 40% informational regardless of how good the trailing twelve months look. This is the single filter that separates durable from melting, it is mechanical, and it runs across a whole listing feed. Then run a live AI Overview presence check on the top 50 money keywords through the DataForSEO SERP API rather than trusting traffic history: a site already at 80%+ AIO presence has its decline in front of it and the trailing revenue is the last clean number you will ever see. Comparison and 'best X' format triggers AIO at 95.4%, so treat it as informational for this purpose.
033. LIVE SCREEN SHARE OF EVERY REVENUE DASHBOARD, or walk. On a call, with the seller navigating: analytics, the ad network dashboard, the affiliate or ASP dashboards, and the payment processor. Refuse recorded video walkthroughs, which can be manipulated, and refuse screenshots, which are edited with basic HTML tools. With verified revenue at 5.9% across 934 in-band Flippa listings and sellers typing profit figures into a form by hand, this is the only real financial check available at this price point, and a refusal is the cheapest disqualifier you will ever get.
044. GSC VERSUS GA4 DIVERGENCE TEST, the first thing you do with granted access. Plot Search Console clicks against GA4 organic sessions month by month for 24 months on identical date ranges. Search Console filters bot and crawler clicks and GA4 does not, so purchased traffic hitting the site with a spoofed Google referrer appears as a widening gap that starts in exactly the months the seller is selling on. Supporting tells in the same session: no single page above 15% of total traffic, time on page above one minute (under 10 seconds means bought, under 3 seconds means bots), geographic mix matching the niche, and no double GA tracking code producing near-zero bounce with doubled page views.
055. MANUAL ACTIONS AND SECURITY ISSUES PANELS IN SEARCH CONSOLE, shown live. A manual action is a hard stop rather than a negotiating point, because the recovery timeline is unknowable and you buy the penalty with the domain. While you are in there, pull the impressions-versus-clicks by month view, which is the specific signature of AI Overview damage in progress; a seller withholding that chart is withholding the only view that matters.
066. BANK RECONCILIATION OF 12+ MONTHS. Obtain affiliate and advertising reports, merchant processing statements and bank statements, then build a monthly income statement linking each earning to the deposit that cleared, allowing for net-30 terms and payout thresholds that shift income between months. Request the full line-by-line PayPal transaction history rather than a summary. On costs, review statements line by line and independently price every claimed subscription. Then sense-check: any earnings increase without a corresponding traffic increase is a fabrication signal.
077. TRANSFER ELIGIBILITY IN WRITING FROM EVERY MONETISATION COUNTERPARTY, contacted by you, naming the site. Mediavine accounts do not transfer at all: you reapply, they re-run diligence, and approval history, revenue share and account standing do not carry over, so a site that has run their ads for years can be rejected. Raptive requires the site to have been onboarded with them, or associated with the current owner of record, for at least six months. Amazon Associates cannot be assigned without Amazon's express prior written approval and terminates on seven days notice either way, so the seller's earnings history proves nothing about yours and any claimed custom rate almost certainly dies. Do not accept the seller's assurance on any of this.
088. THE JAPANESE ASP TRAPS, which the English literature does not cover and which are where your actual risk concentrates. Rakko's own catalogue of buyer failure cases names four: closed ASP registration blocked because the programme is invitation-only and the new owner simply cannot get in; 特別単価 non-transferable because the rate was negotiated to the seller personally and reverts on transfer; duplicate ASP account registration blocked because the site is still registered under the old owner's account and he must remove it first; and AdSense re-approval failure where AdSense was primary. The verification is to actually attempt registration during handover rather than assume, and to have the seller contact each ASP and confirm the rate in writing before close, then document it in the contract.
099. THE ASSET SCHEDULE, itemised and named, before price is discussed. Domain (confirm ドメイン is explicitly listed in 譲渡対象物; SiteStock 118807 lists content and service-use rights and omits it while the same seller's other listing includes it), registrar account, hosting account, ESP account, every social and video handle, every ASP and ad-network account, content, database and code. Rakko's failure list also names 'associated SNS accounts not transferred' and the forgotten prior-domain 301 that quietly stops being maintained. Anything not named stays with the seller in an 事業譲渡.
1010. BACKLINK PROFILE AND PBN DEPENDENCY. Filter to one link per domain, review the first two to three pages manually, and look for repeated IPs, shared referring networks, thematically unrelated linking domains, spammed international TLDs, adult or gambling sources, over-optimised anchor text, and a heavy concentration of blog comments. Then ask the seller three direct questions: is the PBN content genuine or spammy, do you maintain the links after the sale, and does continued linking require a recurring payment to you. Links arriving in bursts that precede each ranking jump means the rankings are rented rather than owned. Also search Google for the domain plus 'Flippa' to detect prior marketplace listings; a site sold three times in four years is a churn asset regardless of what the current listing says.
1111. EXPIRED-DOMAIN AND CONTENT-ORIGIN AUDIT. A flatline followed by growth in analytics indicates an aged or repurposed domain rather than a site that grew. Confirm in Wayback whether the domain stayed in a consistent niche, because a pivot invalidates the inherited link value and usually means the backlinks point at content that no longer exists. Rakko declares expired-domain use and AI content in the listing itself, which is a real advantage; use it. Separately run content samples through CopyScape (above roughly 5% duplication needs investigation), reverse-image-search with TinEye and demand the actual stock licence records, search USPTO for brand names appearing in URLs, and check the Lumen database for DMCA notices already filed against the domain.
1212. EMAIL LIST CONSENT RECORDS, not the list. Acquiring a list through a transaction does not acquire each person's consent, sending to it violates Klaviyo's Acceptable Use Policy, and under GDPR you must be able to produce the exact signup language, the timestamp, and what was consented to, per subscriber. If the seller cannot produce that, value the list at zero rather than per subscriber. Require transfer of the ESP account itself rather than a CSV export, since an export leaves the seller a copy to mail against a competing property.
1313. REVENUE CONCENTRATION AND REFERRAL OWNERSHIP. Request affiliate reports segmented by referring domain rather than in aggregate, and establish for each major referrer whether the seller has any relationship with it, formal or informal, because a seller who owns or controls the referrer is supplying traffic that leaves with him. On the earnings side, get the split by programme and by advertiser: a single 特単 or a single 純広告 contract carrying most of the revenue makes that contract's remaining term and transferability the entire deal (Rakko 22293 runs ¥122,000/month on one display contract against 2,768 monthly page views).
1414. STOLEN-DOMAIN AND FRAUD SCREEN. Flippa states on its own blog that stolen domains have been listed and names six flags: a domain listing carrying extensive content, a suspiciously low buy-it-now price, recently updated WHOIS, a newly created seller account, deleted comments on the auction, and a request for wire transfer. Any one alone proves nothing; two together means walk. The wire-transfer request is load-bearing on its own, because a legitimate seller has no reason to route around escrow.
1515. LISTING LIVENESS AND CROSS-BOARD PRICE CHECK, which sounds trivial and repeatedly was not in this harvest. Site-Z displays listings as 受付中 whose 交渉期限 expired years ago, including one from October 2019 with 2019 financials still shown. Rakko 22390 advertises ¥50-60万/month profit in the past tense for an asset that has produced 2 leads since a July 2026 domain rebuild. And the same asset appeared simultaneously at ¥1,000,000, ¥2,500,000 and ¥3,300,000 across three boards on the same day. Confirm the deal is live, read the detail page rather than the index row, and search the asset across every board before making an offer.
Japan mechanics

How the purchase actually works

01STRUCTURE: it is 事業譲渡 (business transfer) or bare 資産譲渡, never 株式譲渡, because at this size the seller is a 個人事業主 or a one-man KK selling one property from a portfolio and there is no entity to buy. This is good for a first-time acquirer: hidden liabilities stay with the seller because only named assets and named contracts transfer. The one exception to guard is 会社法22条, under which taking over the seller's 商号 can make you liable for their business debts, so exclude the trade name in the contract or register an explicit 免責.
02NATIONALITY: your US citizenship creates zero Japanese acquisition restriction. Under 外為法26条1項1号 an 外国投資家 is a 非居住者である個人, a non-resident individual. You are a 居住者, so the inward-direct-investment regime does not reach you at all. No 事前届出, no 事後報告, no waiting period. The dormant KK is also not an 外国投資家 because 26条1項3号 only catches companies 50%+ owned by one.
03ENTITY: default to buying as the 個人事業主. It needs no new registration and no new filings, and it keeps the US side simple (a disregarded sole proprietorship on Schedule C with foreign tax credit relief). Two things override that default. A .co.jp target legally cannot be held by an individual (JPRS restricts it to companies registered in Japan; .jp only needs a Japanese address, which you have), and sustained profit past roughly ¥8-10M net starts favouring corporate rates. If the KK becomes the acquirer, you own a CFC: annual Form 5471, GILTI inclusions on essentially all profit because a website has no tangible asset base, and a USD 10,000 minimum penalty for a missed form. Engage a US expat CPA before deciding, and subtract their USD 1,000-3,000/year from the Japanese tax saving before comparing.
04CONSUMPTION TAX, the highest-value item on this list: essentially the entire price of a website deal is 課税資産 (goodwill, domain, content, copyright, code, member data), so plan on paying price x 1.1. Whether that 10% is recoverable depends entirely on your calculation method. On 本則課税 it is reclaimable as 仕入税額控除. On 簡易課税 or the 2割特例 it is a permanent unrecoverable cost, ¥500,000 on a ¥5M site and ¥1,000,000 on a ¥10M one. Your invoice registration T5810247107197 proves you are a 課税事業者 and says nothing about which method you are on. Switching cannot be done retroactively: the 消費税簡易課税制度選択不適用届出書 must be filed before the start of the taxable period, so for a calendar-year individual filer buying in 2027, that means by 2026-12-31, and 簡易課税 carries a two-year lock-in.
05SELLER'S INVOICE STATUS: ask every seller for their 適格請求書発行事業者 registration number before pricing. A registered seller means full input credit. An unregistered 免税事業者 seller caps your credit under the 経過措置 at 80% now, stepping down from 2026-10-01. The 令和8年度税制改正大綱 softened the taper to 80% then 70% then 50% then 30% to nil by 2031-09-30, but NTA's own page still shows the old 80/50 wording and secondary write-ups disagree on the middle years, so have the 税理士 confirm against enacted law. On a ¥10M purchase from an unregistered seller the difference between 80% and 70% is ¥100,000. Put it in the price you offer.
06THE ¥10,000,000 LINE: any single purchase at ¥10M or more tax-exclusive (roughly USD 63,000 at 159 JPY/USD) is a 高額特定資産 and locks you out of 免税事業者 status and 簡易課税 for three taxable periods. Two purchases at ¥6M each do not aggregate, because the test is per 取引の単位. That is a genuine structural argument for buying two or three cheaper properties rather than one at the ceiling, independent of the portfolio maths.
07GOODWILL AMORTISATION: 営業権 is a named 減価償却資産 at 所得税法施行令6条8号ヲ with a statutory 5-year 耐用年数 under 別表第三, using 定額法, and for an individual the deduction is compulsory rather than elective. A ¥6M site bought as goodwill gives ¥1,200,000/year of deduction for five years, prorated monthly in the first year. Because it is compulsory you cannot defer it into a higher-income later year. If the KK buys instead, the goodwill is likely 資産調整勘定 over 60 months rather than 営業権 over 5 years; same duration, different mechanism, and I could not verify the statutory basis for a plain 事業譲受け, so ask the 税理士 and do not accept an article number that has not been checked.
08STAMP DUTY: sign electronically and 印紙税 is zero. A site sale contract on paper is 第1号文書 (無体財産権の譲渡 covering 著作権 and 商号, plus 営業の譲渡) and costs ¥10,000 at ¥5-10M and ¥20,000 at ¥10-50M. Stamp duty attaches to the creation of a paper document under 印紙税法基本通達44-1, so a PDF exchanged or signed in Docusign creates no taxable document. Rakko M&A signs through Docusign at its own cost, so on that platform the question never arises.
09ESCROW: Japanese platforms run real escrow. Rakko M&A holds the purchase price (仮払い), automates deposit confirmation, refunds the buyer in full if the site fails inspection, auto-generates a lawyer-supervised contract you can edit, and provides free lawyer consultation over chat, charging the buyer 5% with a ¥55,000 tax-inclusive minimum while the seller pays nothing. SiteStock runs the same pattern with the escrow itself free. Neither platform states a legal framing (no 資金移動業 registration, no 信託 segregation, no licensed third-party agent mentioned), so these are platform-held funds relying on the platform's solvency. For anything near the ceiling, ask in writing how client funds are segregated before wiring. Never settle off-platform: a wire-transfer request outside escrow is the single most reliable fraud signal at this price point.
10BATONZ FEE WARNING: BATONZ charges a minimum buyer fee of ¥385,000 including tax on deals under ¥10M, which is a 26% surcharge on a ¥1.5M listing. Treat BATONZ as a Tier C-only channel. BATONZ 83045's true all-in cost is ¥11,000,000 including its disclosed ¥1,000,000 Lehman fee, not the ¥10,000,000 headline.
11TRANBI BLOCKER: per prior vault research TRANBI will not verify a foreign national on a residence card or passport and requires a My Number Card. TRANBI holds the largest Japanese web-media inventory, including the best structural fit in this harvest (listing 18129), so if you want that channel open, start the My Number Card application now. Treat it as blocked until it exists.
12CONTRACT: the market-standard サイト売買契約書 is a 17-clause 事業譲渡 document (譲渡対象 covering domain usage right, program, design, content, passwords, all IP, member data and contractual position; receivables and payables up to the 譲渡基準日 staying with the seller; 著作者人格権 handled by a non-exercise clause; 2-year 競業避止 from the 譲渡基準日; 6-month 瑕疵担保; 1 month of free transition support). Its 表明保証 deliberately disclaims responsibility for SEO順位 and 顧客継続, which is exactly the risk you are buying. So the leverage is structural rather than contractual: negotiate a 20-30% holdback released after 60-90 days of verified traffic and revenue, and push the 1 month of transition support to 3 months for a Japanese-language property. Refuse any earnout the seller proposes; it moves their performance risk onto you on metrics you have not yet learned to run.
13CORPORATE SELLER: if the seller is any kind of company, ask for the 株主総会議事録 or board minute authorising the sale (会社法467 requires a special shareholder resolution for a transfer of all or an important part of a business, with 略式 and 簡易 carve-outs at 468 whose numerical thresholds counsel should confirm), and keep the standard warranty that all 会社法 procedure has been completed. If the seller is a 個人事業主, the contract alone is sufficient.
14DOMAIN TRANSFER: two separate procedures, not one, and budget several days. 指定事業者変更 is a 10-step round trip through JPRS, and the AuthCode is per-domain, single-use, and expires at 23:59:59 on the 35th day after issue. 登録者変更 (changing who owns it) is a distinct procedure at the registrar. For gTLDs, ICANN's 60-day transfer lock applies after initial registration or a previous registrar transfer with no workaround, so check WHOIS last-updated before signing. Sequence: contract signed, escrow funded, seller issues AuthCode, domain and hosting move, buyer verifies the live site plus analytics access plus revenue accounts, then escrow releases. Never pay against a promise to transfer next week.
15US SIDE: as a sole proprietor, get a J/USA-6 certificate of coverage to keep 15.3% US self-employment tax off the acquired revenue, since there is no foreign tax credit relief against SE tax. This is a question for a US expat CPA rather than the Japanese 税理士, and which Japanese scheme you are enrolled in (国民年金 versus 厚生年金) affects the paperwork.
16JAPANESE INCOME TAX: acquired revenue flows into 事業所得 on the existing return. All-in marginal rate at the USD 20k/month target (roughly ¥37M/year) is 40% national plus 2.1% surtax, about 10% 住民税, and 5% 個人事業税 (Tokyo 第1種事業, after a ¥2.9M 事業主控除, paid 31 August and 30 November). That is the scale argument for the KK, and it is a scale argument rather than a day-one one.
Where to hunt

Marketplaces worth your time

ラッコM&A (Rakko M&A) Japan Tier A and Bhttps://rakkoma.com/project/listThe single best source in this study and the primary channel to work. 93.92% of the Japanese site-sale market, 6,950 cumulative deals worth ¥4,898,663,791, and 528 live listings across 27 crawlable pages with no login. Uniquely, it publishes before login: month-by-month revenue, Search Console clicks, impressions, CTR and average position, page views with average and peak, article counts, whether the domain was bought expired, whether the content is AI-generated, and the seller's own valuation multiple. That is better pre-contact evidence than any global board offers at any price. Practical notes: the JSON API at api.rakkoma.com caps at 50 rows and cannot enumerate the board, so build any monitor on the paginated HTML at /project/list. 47.04% of trailing-year listing supply is YouTube channels and 12.38% is SNS accounts, so filter those out mechanically or effective supply looks twice as large as it is. Median close 60 to 90 days; 35.26% success rate on listings older than four months. Median 8k-20k multiple is 29.2x, so this is a disclosure advantage rather than a price advantage.Fees: Buyer 5% with a ¥55,000 tax-inclusive minimum; seller pays nothing. Escrow (仮払い) included, plus lawyer-supervised auto-generated contract, Docusign at the platform's cost, and free lawyer consultation by chat.
サイトストック (SiteStock) Japan Tier A and Bhttps://sitestock.jp/sales_web_detail.htmlPublic index with crawlable detail URLs and no login, carrying current dates. Detail pages disclose page views, unique users, page counts, monthly cost breakdown and the 譲渡対象物 schedule. The second-best Japanese channel and the right cross-check on Rakko, since the same asset appeared on both boards at different prices during this survey. Read the 譲渡対象物 line every time: listing 118807 omits ドメイン while the same seller's 118445 explicitly includes it, and that difference is the entire asset. Written disclosures can be as short as two sentences for a ¥6M ask, so treat SiteStock as a sourcing board and do the evidence work yourself.Fees: Escrow service itself is free. Standard flow: contract signed, buyer wires to SiteStock's designated account, SiteStock notifies the seller, seller transfers, buyer confirms, SiteStock remits net of fees.
Flippa Global Tier A and Bhttps://flippa.comThe largest in-band global inventory and the only global board with a working public JSON API (note the filter keys are nested: filter[price][min], not price_min, which 400s). 645 of 1,000 in-band listings carry both revenue and profit. Median multiples measured 2026-08-25: 21.8x in the 8-20k band (n=452, median profit USD 588/month), 23.5x in 20-50k (n=352), 26.5x in 50-75k (n=123). The critical limitation: human vetting applies from USD 50,000 up, so only 5.9% of 934 in-band listings with a stated profit had verified revenue and 36% had verified traffic. Filter hard on has_verified_revenue and has_verified_traffic; six of the strongest global candidates in this study carry both badges. The badges tell you a number was checked, not whether it is rising or falling, so the month-by-month trend still requires an account.Fees: Seller success fee 10% under USD 50,000, which sellers price into the ask. Flippa escrow runs at roughly 0.89% and routes through Escrow.com.
Motion Invest Global Tier A and Bhttps://www.motioninvest.com/marketplace/The specialist in the sub-USD 30,000 band, with 89 active listings of which 14 fall in range, and by far the richest structured data of any global board: price, monthly revenue and profit, income multiple, GA page views and uniques, domain authority, referring domains, article counts, monetisation, seller Q&A and the seller's own growth plan, all readable through the public Supabase REST API behind its front end. Worth screening every week. But the asks run 19x to 36x monthly against Flippa transacting comparable content at 19x to 27x and Investors Club's best at 18x to 21x monthly equivalent, with several assets at 33x to 36x for under USD 400/month. Use it for the data quality and treat the asking prices as unsupported by the rest of the market on the same date.Fees: Not established in this survey.
サイトマ (Saitoma) Japan Tier Bhttps://saitoma.com/website/Worth using for one reason: Saitoma states its real inventory sits in a private 未公開案件 newsletter and the public board is a shop window, which is confirmed by the board carrying only four live listings against a long 成約済 archive. Private and off-market flow is where mid-teens entry multiples live, and entry multiple moves the three-year outcome roughly four times more than position count. Subscribe. The public listings publish profit as a range rather than a point figure (one candidate spans ¥62,391 to ¥582,811 monthly, a 9x swing), so nothing is verifiable pre-contact, but the newsletter is the asset here rather than the board.Fees: Not disclosed publicly; domain gated behind registration.
TRANBI / トランビ Japan Tier Bhttps://www.tranbi.com/buy/list/division/74/The largest Japanese web-media inventory (490 unique listings across 17 crawlable pages in the website and app division, 181 with a point price, 85 in band) and it holds the single best structural fit found anywhere in this study, listing 18129. Its 留意点 section is candid and repeatedly decisive, naming things sellers would rather not volunteer. Two hard limitations. It publishes revenue and operating profit only as bands spanning a hundredfold range (1円〜500万円), so no multiple is computable pre-registration. And per prior vault research it will not verify a foreign national on a residence card or passport, requiring a My Number Card, which makes it inaccessible today. Start the card application if you want this channel.Fees: Not established in this survey.
Empire Flippers Global Tier C onlyhttps://empireflippers.com/listing/The verification standard: they request access to the income sources themselves (AdSense, affiliate dashboards, Stripe, Shopify, Amazon Seller Central) and confirm traffic in Google Analytics before listing, and individual listing pages are fully server-rendered without an account, showing price, monthly multiple, average monthly profit and revenue, profit margin, first-made-money date, the complete monetisation percentage split, the top three traffic channels with exact percentages, and the full asset schedule. Only the domain is gated. The problem is reach: across 200 live listings the cheapest was USD 43,933 and only 11 to 12 sit in band, nine of which are FBA, physical product or YouTube. Use it for two things instead of buying: the published scoreboard (2,666 sales, USD 604,251,848 volume, 26.3x typical against 14.5x distressed) is the map for where to hunt, and the listing pages are the format every other seller should be held to.Fees: 15% commission. Unlocking a listing's domain and full P&L requires government ID and proof of funds.
BATONZ / バトンズ Japan Tier C onlyhttps://batonz.jp/sell_cases/Large inventory (900 unique listings harvested; the transfer_value_ranges and free-text filters both work server-side, contrary to the usual assumption) and it produced the strongest Tier C candidate in the study, listing 83045, because the seller volunteered point figures in the free-text description. But BATONZ withholds 営業利益 publicly, so no multiple is computable unless the seller happens to disclose it, and the listing pages carry an explicit warning that registered financial fields may contain book values or provisional numbers entered only to satisfy required system inputs. The minimum fee makes it structurally wrong for Tier A.Fees: Minimum buyer fee ¥385,000 including tax on deals under ¥10M, which is a 26% surcharge on a ¥1.5M listing. The Tier C candidate here also carries a disclosed ¥1,000,000 Lehman brokerage fee on top of a ¥10,000,000 ask.
サイト売買Z (Site-Z) Japan Tier None; use as a template library onlyhttps://www.site-z.com/detail.htmlIncluded as a warning, because it is the most machine-readable Japanese board and it is largely an archive. Its 受付中 (open) filter is not a liveness filter: every detail page opened during this survey had a 交渉期限 that had already passed, running back to December 2019, and financials are frozen at the listing date. One listing still displayed as open shows 350,000 monthly page views and ¥696,000/month on 2019 figures, which would be the best fit in the entire harvest if it were real. Read it for the shape of asset worth hunting, and confirm liveness with the platform before spending an hour on anything there.Fees: Not applicable.
Marketplaces screened and not worth your time Global and Japan Tier Nonehttps://sitecatcher.net/items/Recording the negative findings so no future session repeats the work. SiteCatcher exposes no crawlable per-listing URL and its price rendering appends a 万円 suffix to rows that are plainly raw yen, making some prices ambiguous by a factor of 10,000. Investors Club gates the domain behind Premium and shows free members new listings 7 to 14 days after Premium members, so anything genuinely underpriced is gone first. Acquire.com has 414 in-band listings of which roughly 90% are micro-SaaS at multiples logged as high as 84x, 97x, 500x and 1,111x TTM profit, and charges USD 390. Niche Investor identifies listings by niche only and never by domain, so nothing there can be diligenced before contact. SideProjectors (3,000 records harvested) and IndieMaker (1,958 listings, where every minimum-revenue filter down to USD 1/month returned zero results) are bazaars for pre-revenue code, and IndieMaker charges USD 25 per unlocked listing. BuyMicroStartups has USD 380k of total deal value across the whole marketplace. Microns publishes only annual revenue without an account and runs a five-day delay on the free tier. Tiny Acquisitions no longer resolves. Odys Global is fully login-walled and exposes no domain, price or DR figure publicly. UREBA is defunct, its inventory transferred out in April 2024.Fees: Varies; several charge for access.