c2c06353-00b9-4782-b198-861b52e29d1bIMD JOB

Do headline share buybacks actually reduce the share count? Compare Apple and Microsoft using fiscal years 2023, 2024 and 2025.

Use each company's official annual reports/Form 10-K filings and their original financial statement notes as the primary evidence. Record the exact fiscal year-end dates; Apple's and Microsoft's fiscal years are not identical calendar periods. Useful starting points are Apple's 2025 Form 10-K at https://www.sec.gov/Archives/edgar/data/320193/000032019325000079/aapl-20250927.htm and Microsoft's 2025 annual report at https://www.microsoft.com/investor/reports/ar25/ . Retrieve the other annual filings needed for the requested historical comparison. Prefer later restated comparative figures when appropriate, but identify and explain restatements and differences rather than mixing incompatible values. Record your research cutoff in UTC.

For each company-year, extract cash spent on share repurchases, number of shares repurchased if disclosed, period-end common shares outstanding, the opening share count for the same period, weighted-average diluted shares used for EPS, and stock-based compensation expense. Keep period-end counts separate from weighted-average counts and separate dollars from shares. Cite the exact report section, note or table for every quantitative series. Identify whether repurchase amounts include taxes, fees, accelerated repurchase settlements or other adjustments and reconcile cash-flow versus equity-note figures where they differ. Do not substitute an authorized buyback amount for purchases actually executed.

Calculate the change in period-end shares outstanding in shares and percent, using consistent units and split-adjusted comparisons where applicable. Where the share roll-forward supports it, reconcile repurchased shares with employee share issuance, shares withheld for taxes, acquisitions or other disclosed share changes. Explain how gross repurchases can differ from the net reduction in shares. If a complete reconciliation is not disclosed or cannot be reproduced, label the residual unexplained rather than attributing all of it to employee compensation. Stock-based compensation expense is an accounting expense, not the number of new shares issued; do not subtract that dollar expense from repurchased share counts or present a dollar ratio as a direct dilution measure.

Produce a six-row company-year comparison table and a short explanation of which company reduced its share count more over the requested periods, what the filings can explain about the difference, and what cannot be inferred. Discuss why repurchases, share-count reductions and shareholder value are different questions: this historical accounting exercise cannot establish whether the stock was undervalued, predict returns, or prove that repurchases alone improved the business. Avoid a stock-picking recommendation or a causal EPS claim unsupported by a proper counterfactual.

Deliver artifacts/report.md with a plain-English summary, the comparison table, a worked reconciliation example when supported, and limitations; keep the main narrative under 1,800 words excluding references. Deliver artifacts/evidence.json containing the six observations, units, fiscal dates, source URLs and table references, reported values, derived calculations and any unresolved differences. Use at least five distinct relevant primary-source URLs, preferably the companies' annual filings for the specified years. Verify that the JSON parses and that signs, units, totals and percentage calculations agree with the source statements and report. This is a one-time historical public-information study. Do not trade, connect brokerage accounts, use paid services or set up monitoring.