Customer concentration can materially shape the quality and durability of a company's reported revenue, but the evidence is often buried inside annual filings rather than presented as a clean analytical series. An analyst reviewing a 10-K may need to search customer-risk notes, revenue disclosures, and accounts-receivable sections to understand how much of the business depends on a small number of customers.
Those disclosures do not always measure the same exposure. A customer accounting for a large share of year-end receivables may represent a different share of annual revenue. A customer that disappears from a disclosure threshold in the following year should not automatically be assigned a zero percentage, and language such as “significant customer” does not support a numerical estimate.
In this article, we use Claude with Financial Modeling Prep to extract disclosed customer concentration from two annual filings for Applied Optoelectronics (AAOI). The example separates revenue concentration from receivables concentration, calculates approximate revenue exposure where the filing supports it, and compares the disclosures across FY2023 and FY2024. Conflicting or non-comparable evidence remains visible for analyst review.
FMP Data and Claude MCP Setup
The workflow uses FMP's MCP server to retrieve structured annual-report content and filing records. An analyst then checks the relevant primary filing when an extraction omits context or different sections report conflicting values. This reduces the amount of manual searching while preserving the source review needed to interpret a disclosure.
To reproduce the analysis, obtain an FMP API key, follow the Claude connection instructions, and confirm that the connector can retrieve a basic company record before running the analytical prompt. Check dataset access for your plan before starting. No separate Python workflow is required for this example.
The Form 10-K financial-report JSON is the primary filing-content source. SEC filing records establish the corresponding annual filing and its provenance. Consolidated annual revenue comes from the “Revenue, net” line in each year's own filing. Company identifier data is needed only if the company or filing identity requires clarification.
Customer Concentration Comparison Methodology
FY2023 and FY2024 are intentionally retained as a historical validation case, not as the latest available annual periods. The FY2024 filing is useful because its customer disclosures require consistency checks across different sections. The example's named-customer and Top 5 comparisons remain within this two-year window.
Keep the Denominators Separate
|
Disclosure type |
Denominator and period |
|
Revenue |
Consolidated annual revenue or net sales for the fiscal year. |
|
Receivables |
Total accounts receivable at the fiscal year-end date. |
|
Other |
Another explicitly identified denominator, retained separately. |
Review Required is a separate evidence-control flag. It identifies uncertainty about the value, denominator, customer identity, source, or comparison rather than serving as a disclosure type. Revenue and receivables percentages remain separate throughout the extraction and interpretation.
Preserve the Source Hierarchy
Each fiscal year's own annual filing supplies the primary evidence for that year's audited disclosure. The audited financial-statement customer-concentration note takes priority when it provides a percentage for the customer, fiscal year, and denominator being analyzed. Another quantified section of the same filing may supplement it when those details are explicit and the note does not provide a conflicting value.
A later filing's comparative disclosure may supply an earlier-year percentage when it explicitly identifies the customer, year, denominator, and percentage. Its later-filing provenance must remain visible. If the audited note and another section disagree about the same observation, the workflow retains the audited-note figure, records the alternative, and marks the observation Review Required; it does not average the values or silently choose a resolution.
Use Disclosed Percentages and Supported Bounds
Use a disclosed percentage when the filing provides one. If the filing states that no other customer exceeded 10% and does not quantify a particular customer elsewhere, retain the supported bound of “Not greater than 10% (≤10%).” Do not substitute 10%, 9.9%, a midpoint, or zero. If neither a percentage nor a bound can be established, report Not quantified and flag the observation for review.
In the prompt below, “exact” distinguishes an explicitly stated percentage from a threshold bound or an inferred estimate. It does not mean that the underlying company percentage is unrounded. This distinction also applies to a reported 0%: it should be reproduced at the filing's stated precision rather than treated as a separately verified dollar amount.
Implied Revenue Exposure = Annual Consolidated Revenue × Disclosed Revenue Concentration %
Calculate implied exposure only for explicitly disclosed revenue percentages with a clear consolidated-revenue denominator. Keep the result approximate because the percentage may be rounded, use the full reported revenue amount, and round the final result to the nearest $0.1 million. Do not multiply a receivables percentage by annual revenue or calculate exposure from a threshold-only observation.
Compare Like Measures Across Years
Change in Percentage Points = FY2024 Percentage − FY2023 Percentage
The comparison classifies supported numerical movements as Increased, Decreased, or Unchanged. When an observation crosses the disclosed threshold, it uses Entered disclosed threshold or Fell below disclosed threshold and leaves the exact percentage-point change as N/A. In this example, the latter label means the customer is no longer above 10%; the supported bound remains ≤10%, rather than strictly less than 10%.
Named customers can be aligned when identity is established. Anonymous labels such as Customer A and Customer B cannot be assumed to identify the same counterparty in successive filings. A change in Top 5 membership, however, does not invalidate a comparison of the aggregate ratio when its definition and denominator remain consistent; it limits attribution of the aggregate movement to particular customers.
Comparability Status and Review Required answer different questions. A pair of observations can use matching customers, periods, and denominators while still requiring verification of a conflicting source value. An increase in concentration, a decrease, or a supported threshold crossing does not itself trigger Review Required.
Claude Prompt for Customer Concentration Extraction
Copy the complete prompt below into the connected Claude session. The published results that follow arrange the evidence into narrower table panels and source notes for readability.
Use the connected Financial Modeling Prep (FMP) MCP server to extract and compare disclosed customer concentration for Applied Optoelectronics (AAOI) across FY2023 and FY2024 annual filings.
This is intentionally a historical validation case, not a latest-period screen. FY2023 and FY2024 are retained because the FY2024 filing contains customer-concentration disclosures that are useful for demonstrating source hierarchy, cross-section consistency checks, and Review Required. FY2025 is outside this demonstration's defined comparison window.
|
Use the connected Financial Modeling Prep (FMP) MCP server to extract and compare disclosed customer concentration for Applied Optoelectronics (AAOI) across FY2023 and FY2024 annual filings. This is intentionally a historical validation case, not a latest-period screen. FY2023 and FY2024 are retained because the FY2024 filing contains customer-concentration disclosures that are useful for demonstrating source hierarchy, cross-section consistency checks, and Review Required. FY2025 is outside this demonstration's defined comparison window. Use FMP MCP as the primary retrieval layer. Do not use general web search, analyst estimates, quarterly data, third-party sources, or general knowledge to fill missing information. If the FMP structured filing extraction does not preserve a relevant filing section, you may use the primary SEC filing linked through the FMP filing record only as supplemental validation. Do not use third-party reproductions of the filing. Keep the workflow limited to AAOI, FY2023, FY2024, and the customer-concentration evidence required below. 1. Retrieve the Two Annual Filings Retrieve AAOI's FY2023 and FY2024 annual 10-K information using FMP MCP. Use:
For each fiscal year retain:
Use each fiscal year's own 10-K as the primary source for that year's audited disclosure. Do not silently substitute a later filing for the corresponding year's own filing. If filing identity or provenance cannot be established reliably, set Review Required = Yes. 2. Locate Customer-Concentration Disclosures Search the filing evidence for quantified disclosures relating to:
For each relevant observation retain:
Classify Disclosure Type as:
For Revenue, the denominator must be consolidated annual revenue or net sales. For Receivables, the denominator must be total accounts receivable. For Other, state the denominator explicitly. Never combine revenue and receivables percentages. 3. Source Hierarchy Use the audited financial-statement customer-concentration note as the primary quantitative source when it provides an exact value for the customer, fiscal year, and denominator being analyzed. However, do not reduce an observation to a threshold bound merely because the audited note does not list that exact customer percentage. Another quantified section of the same annual filing may supplement the audited note when it explicitly identifies:
Use that exact figure when the audited note does not provide a conflicting exact value for the same customer, fiscal year, and denominator. A later filing's comparative disclosure may also be used for an earlier-year customer percentage when the later filing explicitly identifies:
When using such a comparative figure, preserve its provenance explicitly. Do not present it as though it came from the earlier filing's audited note. If the audited note and another section of the same filing report different exact percentages for the same customer, fiscal year, and denominator:
4. Handle Threshold-Based Disclosures Conservatively Use an exact percentage when the filing evidence provides one. Use a threshold bound only when an exact percentage cannot be established. Where the filing supports only that a customer was not above a 10% disclosure threshold, report: Not greater than 10% (≤10%) Do not convert that into:
If neither an exact percentage nor a supported threshold bound can be established, report: Not quantified and set: Review Required = Yes Do not infer that an omitted customer had zero concentration unless a filing explicitly reports 0%. 5. Retrieve Consolidated Annual Revenue Retrieve AAOI consolidated annual revenue for FY2023 and FY2024 from each fiscal year's annual financial statement / structured annual filing. Retain:
Do not use:
If consolidated annual revenue cannot be established reliably, do not estimate it. 6. Calculate Implied Revenue Exposure For each exact Revenue concentration percentage, calculate: Implied Revenue Exposure = Annual Consolidated Revenue × Disclosed Revenue Concentration % Treat the result as approximate because disclosed customer percentages may be rounded. Round the implied exposure to the nearest $0.1 million only after completing the full calculation. Do not calculate implied revenue exposure for:
Never multiply a receivables percentage by annual revenue. 7. Compare Named Customers Across Years Compare a named customer across FY2023 and FY2024 only when:
For two exact comparable percentages calculate: Change in Percentage Points = FY2024 Percentage − FY2023 Percentage Classify:
When one period contains only a threshold-supported value, use:
only when supported by the filing evidence. Do not calculate an exact percentage-point change when either value is only a threshold bound. 8. Top 5 Customers and Other Aggregate Ratios Treat an aggregate metric such as Top 5 Customers as comparable across periods when the aggregate definition and denominator remain consistent, even if the identities of the customers within the group change. Changing membership does not make the aggregate concentration ratio itself non-comparable. For comparable Top 5 ratios:
Changing group composition limits customer-level attribution of the aggregate movement. Do not infer which individual customers caused an aggregate increase or decrease unless the filing separately establishes their contribution. Apply this rule separately to:
9. Customer Identity Rules Use customer names exactly as disclosed. Named customers may be aligned across periods when identity is clear. Do not assume anonymous labels such as Customer A or Customer B represent the same counterparty across filings unless the filing establishes that identity. If identity cannot be established reliably:
Changing membership of a consistently defined aggregate group is not, by itself, a customer-identity failure for the aggregate ratio. 10. Review Required Rules Set Review Required = Yes when:
Do not set Review Required merely because:
11. Required Output Return the results in the following order. Table 1 — Disclosed Customer Concentration Evidence Create: | Company | Fiscal Year | Filing Source | Customer / Customer Group | Disclosed Percentage | Disclosure Type | Denominator | Annual Revenue | Implied Revenue Exposure | Source Section / Evidence | Review Required | Evidence Note | Requirements:
Table 2 — Year-over-Year Customer Concentration Comparison Create: | Customer / Customer Group | Disclosure Type | FY2023 Percentage | FY2024 Percentage | Change in Percentage Points | Year-over-Year Direction | FY2023 Implied Revenue Exposure | FY2024 Implied Revenue Exposure | Comparability Status | Review Required | Analyst Follow-Up Action | Use:
For Top 5 aggregate ratios, use Comparable when the aggregate definition and denominator remain consistent. For exact comparable percentages, calculate the percentage-point change directly. For threshold-based comparisons, show: Change in Percentage Points = N/A Populate implied exposure only for exact Revenue percentages. Filing Provenance Summary Create: | Fiscal Year | Form | Accession Number | Filing Acceptance Timestamp | Filing | Filing-Content MCP Tool / Operation | Use descriptive filing hyperlinks. Do not guess any accession number, timestamp, URL, tool name, operation, or parameter. 12. Interpretation After the tables, provide a concise interpretation covering:
Do not characterize a concentration increase or decrease as inherently positive or negative. Do not infer:
13. Data Retrieval Note End with a short Data Retrieval Note identifying:
If a broader raw filing-text retrieval fails, state the limitation factually. Do not interpret an FMP MCP extraction limitation as evidence that the disclosure is absent from the underlying SEC filing. Distinguish between:
if both can be established reliably. Do not invent missing tool parameters or call-trace details. Final Consistency Check Before returning the response, verify that:
Do not introduce additional companies, fiscal years, metrics, or post-filing performance analysis. |
Results from the AAOI Annual Filings
The reviewed example preserves the original FY2023 and FY2024 comparison. The tables below present the named-customer and Top 5 observations used in that comparison, with revenue and receivables in separate panels. They are an editorial presentation of the reviewed evidence rather than a claim that every disclosure in the filings is reproduced here.
The annual revenue inputs are $217,646,000 for FY2023 and $249,365,000 for FY2024, reported as “Revenue, net” in the consolidated statements of operations, page F-5 of the respective FY2023 filing and FY2024 filing. The calculations use these full amounts. All implied revenue exposures below are approximate and rounded to the nearest $0.1 million.
Table 1A Disclosed Revenue Concentration
All rows relate to AAOI. The denominator is consolidated annual revenue for the stated fiscal year. Source codes identify the filing and section in the notes below.
|
Fiscal year |
Customer or group |
Disclosed share |
Implied revenue |
Source |
Review required |
|
FY2023 |
Top 5 Customers |
85.6% |
$186.3M |
A |
No |
|
FY2023 |
Microsoft |
46.6% |
$101.4M |
A |
No |
|
FY2023 |
ATX Networks Corporation |
15.6% |
$34.0M |
A |
No |
|
FY2023 |
Digicomm |
11.3% |
$24.6M |
A |
No |
|
FY2023 |
Oracle |
8.8% |
$19.2M |
C |
No |
|
FY2024 |
Top 5 Customers |
92.9% |
$231.7M |
B |
No |
|
FY2024 |
Microsoft |
43.7% |
$109.0M |
B |
No |
|
FY2024 |
Digicomm |
35.1%* |
$87.5M* |
B; conflict in C and D |
Yes |
|
FY2024 |
Oracle |
12.4% |
$30.9M |
B |
No |
|
FY2024 |
ATX Networks Corporation |
0%† |
$0.0M† |
C |
No |
* Digicomm: The 35.1% value and $87.5M estimate use the audited-note disclosure. Other sections report 34.1%; the percentage and derived exposure remain subject to review.
† ATX: The filing explicitly reports 0%. The displayed $0.0M is the calculation from that reported percentage, not a separate disclosure of an exact customer-revenue amount.
Table 1B Disclosed Receivables Concentration
The denominator is total accounts receivable at each December 31 date. No revenue-exposure calculation applies to these rows. The two ≤10% observations are supported bounds used in the comparison, not estimated percentages.
|
Balance date |
Customer or group |
Disclosed share |
Source |
Review required |
|
Dec 31, 2023 |
Top 5 Customers |
75.4% |
A |
No |
|
Dec 31, 2023 |
Digicomm |
35.2% |
A |
No |
|
Dec 31, 2023 |
ATX Networks Corporation |
21.4% |
A |
No |
|
Dec 31, 2023 |
Microsoft |
≤10% |
A; other-customer threshold statement |
No |
|
Dec 31, 2024 |
Top 5 Customers |
95.5% |
B |
No |
|
Dec 31, 2024 |
Digicomm |
66.6% |
B |
No |
|
Dec 31, 2024 |
Microsoft |
16.8% |
B |
No |
|
Dec 31, 2024 |
ATX Networks Corporation |
≤10% |
B; other-customer threshold statement |
No |
Source Notes for Table 1
- FY2023 audited concentration note: Note B, subsection 8, Concentration of Credit Risk and Significant Customers, page F-10. This supplies the FY2023 named-customer revenue percentages, Top 5 ratios, receivables percentages, and the statement supporting Microsoft's ≤10% receivables bound.
- FY2024 audited concentration note: Note B, subsection 8, Concentration of Credit Risk and Significant Customers, page F-10. This supplies the FY2024 Top 5 ratios, Microsoft, Digicomm, and Oracle revenue percentages, the receivables percentages, and the statement supporting ATX's ≤10% receivables bound.
- FY2024 business discussion: Item 1, Business, page 4. This explicitly reports Oracle at 12.4%, 8.8%, and 5.9% for FY2024, FY2023, and FY2022, respectively, and ATX at 0%, 15.6%, and 47.3%. Only the FY2023 and FY2024 observations enter this comparison. Oracle's FY2023 8.8% is therefore attributed to the FY2024 comparative disclosure. This section also reports Digicomm at 34.1% for FY2024, creating the conflict with Note B.
- FY2024 management discussion: Item 7, Management's Discussion and Analysis, Overview, page 30, also reports Digicomm at 34.1%. The workflow records that conflicting value without averaging it with 35.1% or treating the audited-note choice as a resolved factual discrepancy.
Table 2A Year over Year Revenue Comparison
Percentage-point changes use the disclosed percentages. The corresponding approximate dollar exposures appear in Table 1A. Comparable describes the alignment of the customer, measure, and periods; Review Required separately identifies the source conflict.
|
Customer or group |
FY2023 |
FY2024 |
Change |
Direction |
Comparability |
Review required |
|
Microsoft |
46.6% |
43.7% |
−2.9 pp |
Decreased |
Comparable |
No |
|
Digicomm |
11.3% |
35.1%* |
+23.8 pp* |
Increased* |
Comparable |
Yes |
|
ATX Networks Corporation |
15.6% |
0%† |
−15.6 pp† |
Decreased |
Comparable |
No |
|
Oracle |
8.8% |
12.4% |
+3.6 pp |
Increased |
Comparable |
No |
|
Top 5 Customers |
85.6% |
92.9% |
+7.3 pp |
Increased |
Comparable |
No |
* Digicomm: The change is calculated using the audited-note FY2024 value of 35.1%; it remains subject to the disclosed conflict. † ATX: Calculations retain the reported 0% at its stated precision.
Table 2B Year over Year Receivables Comparison
FY2023 and FY2024 columns refer to December 31 balances. Implied revenue exposure is N/A for every receivables row.
|
Customer or group |
FY2023 |
FY2024 |
Change |
Direction |
Comparability |
Review required |
|
Digicomm |
35.2% |
66.6% |
+31.4 pp |
Increased |
Comparable |
No |
|
ATX Networks Corporation |
21.4% |
≤10% |
N/A |
Fell below disclosed threshold |
Threshold Comparable |
No |
|
Microsoft |
≤10% |
16.8% |
N/A |
Entered disclosed threshold |
Threshold Comparable |
No |
|
Top 5 Customers |
75.4% |
95.5% |
+20.1 pp |
Increased |
Comparable |
No |
Threshold interpretation: “Fell below disclosed threshold” retains the prompt's classification label. For ATX, the supported statement is that its FY2024 receivables share was not greater than 10%; the evidence does not establish a precise percentage or a strict <10% bound.
Analyst follow-up: Verify the conflicting FY2024 Digicomm revenue percentages before treating its precise concentration or implied exposure as settled. No additional verification action is identified for the other comparisons. Oracle's later-filing provenance and possible Top 5 membership changes remain interpretation notes rather than unresolved comparison failures.
What the Comparisons Establish
Microsoft remained AAOI's largest individually disclosed revenue customer, although its share declined from 46.6% to 43.7%, a decrease of 2.9 percentage points. Because consolidated revenue increased, its approximate implied revenue exposure rose from $101.4 million to $109.0 million. A lower concentration percentage therefore did not indicate a lower implied dollar contribution in this case.
Using the audited-note figure, Digicomm's disclosed revenue share increased from 11.3% to 35.1%, or 23.8 percentage points, and its approximate implied exposure increased from $24.6 million to $87.5 million. Those precise FY2024 results remain qualified by the 34.1% figure reported elsewhere in the same filing. The workflow's source hierarchy makes the calculation traceable without resolving the inconsistency.
ATX's reported revenue share declined from 15.6% to 0%, while Oracle's increased from 8.8% to 12.4%. The Oracle comparison uses the expressly identified FY2023 percentage in the FY2024 filing. It does not rely on assuming that an unnamed customer in the earlier filing was Oracle.
Receivables show a different exposure profile. Digicomm's share increased from 35.2% to 66.6%, or 31.4 percentage points. Microsoft moved from a supported ≤10% bound to 16.8%, while ATX moved from 21.4% to a supported ≤10% bound. These are shares of outstanding year-end receivables, not annual revenue and not evidence of customer-level credit losses.
Top 5 revenue concentration increased from 85.6% to 92.9%, or 7.3 percentage points. Top 5 receivables concentration increased from 75.4% to 95.5%, or 20.1 percentage points. The aggregate ratios remain comparable because their definitions and denominators are consistent, but changing group membership prevents attributing the aggregate movement to individual customers without additional evidence.
Data Retrieval and Filing Provenance
The documented analysis used FMP MCP for filing identification and structured annual-report retrieval. Claude accessed filing provenance through mcp__FMP__secFilings using search-by-symbol, and retrieved the annual reports through mcp__FMP__statements using financial-reports-form-10-k-json. These names describe the operations recorded in the original session.
The MCP operation differs from the documented REST route. The Financial Reports Form 10-K JSON API uses /stable/financial-reports-json, while the session exposed financial-reports-form-10-k-json. Preserving that distinction avoids presenting a documentation route as though it were the name returned by the MCP session.
|
Fiscal year and filing |
Accession number |
Filing date |
Accepted timestamp |
|
0001437749-24-005367 |
2024-02-23 |
2024-02-23 16:00:57 |
|
|
0001437749-25-005575 |
2025-02-28 |
2025-02-28 07:38:21 |
Acceptance timestamps are reproduced as displayed on the FY2023 SEC filing detail and FY2024 SEC filing detail pages. Fiscal year ends are December 31, 2023, and December 31, 2024, respectively.
Structured retrieval used symbol=AAOI, year=2023, period=FY and symbol=AAOI, year=2024, period=FY. The retained Claude output confirms AAOI as the filing-search symbol but does not preserve every pagination or date-window parameter from the filing-identification call. Those missing parameters are not reconstructed here.
The original retrieval note reports that the FY2024 Microsoft and Digicomm receivables percentages were available through structured filing/XBRL data even though the surrounding narrative extraction did not preserve the complete numerical sentences. Both percentages can also be read in Note B, subsection 8, of the linked primary SEC filing. This documents a limitation of the session's extraction, not an absence of the disclosure from the filing itself.
A targeted attempt to retrieve broader FY2024 filing text through financials-raw-text did not return that text because the retrieved filing record lacked a baseRawLink. Where relevant customer discussion was not preserved by that route, the linked primary SEC filing supplied supplemental validation. The source notes distinguish those sections from the audited concentration note.
Each year's own filing remains the primary source for its audited concentration note and consolidated revenue. Oracle's FY2023 8.8% is a separately identified comparative disclosure from the FY2024 filing. The published tables reflect source verification and editorial formatting of the documented example, not a newly executed MCP session.
Using the Evidence in Further Research
Customer concentration describes disclosed dependency on customers; it does not by itself establish future churn, profitability, credit losses, or revenue loss. Its interpretation depends on the denominator, reporting date or period, customer identity, and source reliability. Keeping those attributes alongside each percentage makes it easier to see when a comparison is supported and when it remains uncertain.
The AAOI case also shows why a single concentration ratio cannot replace a broader review of revenue and cash conversion. Microsoft's share declined while its implied dollar contribution rose, and the receivables distribution differed from the annual revenue distribution. The remaining Digicomm inconsistency identifies a specific source question for follow-up rather than a reason to discard every supported observation.
