AIDMA HOLDINGS INC

Company Overview

AIDMA Holdings is a Japanese small- and medium-sized enterprise (SME) services company. It sells outsourced sales, hiring, and business-support services, combining proprietary sales data and software with remote workers and operating staff. Its central economic proposition is simple: SMEs that cannot hire enough sales or administrative staff can buy an external “unit” of sales capacity and related support instead.

Data freshness matters unusually much here. The latest clean official annual base is FY2025, ended 31 August 2025, filed on 28 November 2025. More recent data is partial, unaudited, or estimated. The latest official update is the nine months to 31 May 2026, announced 10 July 2026; the attached quarterly financial statements received no auditor review. This update is material because it included a large reduction in full-year guidance.

FY2025 audited annual data: revenue was ¥13.270 billion, operating profit ¥3.103 billion, and net income attributable to owners ¥1.955 billion. Operating cash flow was ¥2.480 billion. The supplied FY2025 owner-earnings measure is ¥1.615 billion, but this is a model estimate, not an audited accounting line.

For the market snapshot dated 31 July 2026, the supplied share price is ¥946 and market capitalization is ¥14.060 billion. The FY2025 balance-sheet snapshot showed net cash of ¥5.923 billion. More recent unaudited quarterly data at 31 May 2026 showed cash plus securities of ¥5.797 billion and interest-bearing debt of ¥72 million, or net cash of approximately ¥5.725 billion. The balance sheet remains strong; the question is not solvency but whether the prior growth engine was supported by weak customer-quality controls.

The latest nine-month results are weaker than the FY2025 annual base: revenue increased 12.6% year on year to ¥10.853 billion, but operating profit fell 19.8% to ¥1.850 billion and net income fell 27.3% to ¥996 million. This is not a case where an old annual report can be treated as current.

How the Company Makes Money

The company reports one accounting segment, Work Innovation, but its operating model has three practical revenue streams. The largest is sales support: it uses its Sales Crowd system, business database, marketing methods, call operations, and remote workers to design and execute sales campaigns for corporate clients. The second is human-resources support, including the Mama Works remote-work platform and related hiring and workforce-support services. The remainder includes adjacent business-support products, such as online meeting tools, executive networks, media, and acquired services.

In FY2025, company management data showed sales-support revenue of ¥8.795 billion, human-resources-support revenue of ¥3.924 billion, and other revenue of ¥549 million. These are management service-category disclosures rather than audited reportable segments. The company’s audited annual report describes a single segment.

The sales-support service is not merely a software subscription. A client pays for a package that includes customer targeting, sales scripts, test marketing, remote calling or other outreach, campaign measurement, and a recommended sales process. The company says it has supported more than 15,500 clients cumulatively and maintains a database of roughly 2.1 million registered corporations. Its human-resources side monetizes recruitment advertising and services linking SMEs to remote workers.

This is a relatively asset-light model in physical-capital terms. FY2025 gross margin was 69.9%, only modestly below 71.6% in FY2024. The principal costs are people, outsourced labor, customer acquisition, technology, and quality control rather than factories or inventory. The more important capital requirement is working capital and credit discipline: a contract has little economic value if a customer does not start the service, cannot pay, or stops paying while work is under way.

The franchise has some real advantages: an accumulated sales-playbook database, workflow software, a remote-workforce ecosystem, and accumulated SME customer knowledge. But these are not yet proof of a high-return compounder. The business still depends on sales execution, service quality, customer willingness to pay, and underwriting of small customers. The FY2026 problem demonstrates that weak customer selection can overwhelm attractive gross margins.

Why the Stock Fell

The dominant observed catalyst was the 10 July 2026 disclosure. Management cut FY2026 guidance from revenue of ¥17.0 billion to ¥14.0 billion, operating profit from ¥4.0 billion to ¥2.2 billion, and net income from ¥2.3 billion to ¥1.2 billion. These are reductions of 17.6%, 45.0%, and 47.8%, respectively. This was not a routine quarter-to-quarter variance; it was an admission that the company’s prior order-to-revenue conversion assumptions were wrong.

Management attributed the shortfall primarily to a change in the customer mix during FY2025 and FY2026’s first quarter. The share of individual proprietors and very small owner-managed customers rose, service-start lead times lengthened, first payments were not collected in some cases, and collections deteriorated during service in others. Management therefore reduced or stopped booking certain higher-risk customers, required upfront payment from individual proprietors, and tightened the definition of an order to cases where service starts within 20 business days.

The quarterly evidence supports the seriousness of the issue. In the standalone third quarter, derived from the disclosed cumulative figures, revenue was ¥3.604 billion, up 6.1% year on year, but operating profit was only ¥330 million, down about 60%, with operating margin falling to 9.2% from 24.4%. The nine-month allowance for doubtful accounts rose to ¥987 million from ¥327 million at FY2025 year-end. Higher hiring costs, acquisition-related goodwill amortization, and the provision also increased selling and administrative expense.

The company simultaneously authorized a buyback of up to 450,000 shares and ¥900 million on 10 July 2026. This is a capital-allocation response, not evidence that the operational issue is resolved; the stated purposes also include M&A and incentives.

The supplied data gives a 31 July 2026 market price of ¥946, but does not provide a verified daily price series. I therefore do not attribute a specific percentage stock-price decline to a particular day. The evidence does establish that the July guidance cut, the collection problem, and the collapse in quarterly margin are the dominant market-relevant events.

What the Market Is Assuming

The market is no longer pricing AIDMA as a clean 25%-plus earnings grower. At ¥14.060 billion of equity value, it is assigning limited value to a return to the FY2025 earnings trajectory after recognizing that a meaningful portion of recent order growth did not convert into collectible revenue.

• One-time or repairable factors. Part of the FY2026 profit decline reflects a doubtful-account provision, April hiring costs, and acquisition-related amortization. The underlying gross margin did not collapse: it was 70.0% in the first nine months of FY2026 versus 69.1% in the comparable prior period. This suggests the service itself remains economically viable before overhead and credit losses.
• Medium-term headwind. The revised FY2026 revenue guide of ¥14.0 billion is only 5.5% above FY2025’s ¥13.270 billion, while operating profit is guided 29.1% below FY2025. The problem is not simply a delayed quarter. The company expects fewer orders after imposing stricter customer-quality and service-start rules, and explicitly says the lower order backlog may affect the next fiscal year as well.
• Long-term risk. The market may be assuming that sales outsourcing and SME lead generation have low switching costs and can be increasingly automated by generative AI. This is plausible. AIDMA itself says its AI calling tool currently delivers roughly one-third of human performance at approximately one-seventh of the cost. That is potentially margin-enhancing for AIDMA, but it also makes the underlying work easier for competitors and clients to automate.

The multi-period customer-quality data is the key check. Management disclosed that the individual-proprietor and owner-manager mix rose from 39.2% in FY2023 to 48.9% in FY2024 and 59.1% in FY2025. Over the same period, average lead time reportedly rose from 50.3 days to 55.7 days and then 104.4 days, while the unstarted rate for these customers rose from 41.9% to 51.1% and then 60.8%. This points to a build-up during FY2025, not an isolated third-quarter miss.

The revised FY2026 guide is numerically achievable from the reported nine-month base: it implies fourth-quarter revenue of roughly ¥3.147 billion and operating profit of ¥350 million, or an 11.1% operating margin. That is above the third quarter’s 9.2% margin but far below FY2025’s 23.4% full-year margin. The guide is therefore not heroic, but it still assumes that the credit problem does not worsen.

The critical evidence gap is cash collection. There is no nine-month FY2026 cash-flow statement, no disclosed receivables-aging schedule by customer cohort, and no detailed disclosure separating provisions, actual write-offs, recoveries, and expected losses. The improvement management claims from new payment and service-start rules is encouraging but remains a company update, not independently tested evidence of restored unit economics.

Temporary or Structural?

Customer-credit and order-conversion failure: structural but survivable. The damaged mechanism is not demand alone; it is the conversion of signed orders into started, paid, and profitable service contracts. A company selling to small businesses cannot compound earnings if order growth is achieved by accepting customers who do not start or pay. This is reversible within three years only if underwriting, payment terms, service-start discipline, and customer segmentation become embedded operating controls rather than a temporary management reaction. The balance sheet provides time to repair it. The absence of current cash-conversion evidence prevents treating the repair as accomplished.

Lower near-term growth from stricter customer selection: not truly structural. Requiring prepayment and limiting orders to prompt service starts will reduce reported order volume and backlog in the near term. That is an economically rational trade if it improves cash realization. It becomes structural damage only if the higher-quality SME customer base is too small to support acceptable growth and margins. Evidence is incomplete: the company has not yet disclosed post-rule-change cohort retention, collection, customer-acquisition cost, or lifetime value.

Labor intensity and sales-service commoditization: structural but survivable. The model’s high gross margin comes from combining software, data, process, and flexible labor. However, the service is exposed to wage inflation, lower remote-worker productivity, and customer insourcing. AI can improve AIDMA’s cost base but can also lower barriers to entry. AIDMA’s proprietary data and workflow may preserve an advantage, but this is a narrower moat than a mission-critical system of record with deeply embedded switching costs.

AI disintermediation: a genuine structural risk. If AI agents make prospecting, calling, list building, and campaign optimization cheap and reliable, the value of a human-led outsourced sales layer declines. The company’s own AI investment partially hedges this risk. It does not eliminate it. The relevant test is whether AIDMA can own the client workflow and outcome data rather than simply provide labor-enhanced outreach.

Time-as-a-moat test. A well-funded competitor could recreate a credible sales-outreach product, buy corporate data, use general-purpose AI, and recruit freelance labor within two years. In five years, AIDMA’s historical campaign data, customer relationships, and operating playbooks could matter if they consistently produce better client outcomes. In ten years, the position is durable only if its data flywheel and software become meaningfully embedded in client workflows. Current evidence supports an operating advantage, not an unassailable moat.

The appropriate diagnosis is therefore partly TIME and partly ESSENCE. The immediate earnings reset and provision are repairable. The revealed weakness in underwriting small customers and the limited durability of the service moat are economic issues that should permanently lower the multiple and required yield assigned to the business until disproven.

Is the Market Wrong? By How Much?

The market appears to be pricing a severe but not terminal impairment. The balance sheet gives substantial downside support: the supplied FY2025 net-cash figure of ¥5.923 billion equaled 42.1% of the 31 July 2026 market capitalization, and the updated 31 May 2026 unaudited net-cash calculation is still ¥5.725 billion. The supplied FY2025 NCAV estimate is ¥4.106 billion, or 29.2% of market capitalization. NCAV is not a liquidation value. A formal liquidation estimate is unavailable because inventory and real-estate inputs were missing, and unavailable data must not be treated as zero.

The worth-alive evidence is also attractive but requires a haircut. The supplied FY2025 owner-earnings estimate is ¥1.615 billion, implying an 11.49% yield on the 31 July market capitalization. That measure was produced from the FY2025 annual base and should not be treated as current earnings after the FY2026 reset. Its reported cash conversion of 125.52% is historical; the current receivables and allowance issue makes it unsafe to extrapolate.

Moat-and-mispricing score: 5/10. The stock has tangible balance-sheet support and a plausible earnings-value gap, but the business has just failed a basic quality-of-revenue test. The valuation is attractive only if the revised customer-selection rules contain losses and preserve at least modest normalized owner earnings. It is not a clean high-quality compounding thesis at this stage.

The following is an intrinsic-value estimate, not a price target. It is FY2025-based, adjusted for the nine-month FY2026 update and the revised FY2026 company guidance. Owner earnings are my estimates below, not company facts. I retain stock-based compensation as an expense; it is not added back. The supplied owner-earnings calculation does not disclose enough detail to independently rebuild maintenance capital expenditure or stock-based compensation, so the estimates deliberately use wide required-yield ranges.

The bridge uses approximately 14.863 million economic shares, implied by the supplied market capitalization and price. Net cash is the 31 May 2026 unaudited figure of ¥5.725 billion: cash plus securities less short- and long-term interest-bearing debt. To avoid double counting, the operating owner-earnings figures exclude a de minimis estimate of net financial income. No valuation case depends on multiple expansion.

The base case says the market is modestly too pessimistic, by about ¥3.1 billion of equity value, but this is not a large margin of safety against a second failure in collections. If only 80% of the reported net cash is genuinely surplus after working-capital needs, follow-on provisions, and acquisitions, each valuation value falls by roughly ¥77 per share. The market is clearly wrong only if the company can show that post-rule-change contracts convert to cash at normal rates while preserving an operating margin materially above the revised 15.7% guidance margin.

Key Facts, Estimates, and Judgments

The strongest falsification checks are operational rather than macroeconomic. First, the next annual results must show whether the doubtful-account allowance stabilizes or rises again. Second, management should disclose post-policy cohorts: service-start rates, payment collection, write-offs, customer mix, lead time, and repeat or cross-sell behavior. Third, cash flow must validate earnings; FY2026 cash generation cannot be assumed from FY2025. Fourth, the FY2027 outlook must show whether tighter underwriting caused a temporary reset or permanently reduced the addressable revenue base. Evidence of further provision growth, another guide cut, sub-15% normalized operating margins, or continued contraction in higher-quality orders would invalidate the base case.