# AI & Machine Learning Platforms — Meta ads research Category research. Brand directory deferred. Edition: October 2026. Page updated: 2026-10-11. ## Account AOV aggregate - Median: $31.13 - Middle 50%: $11.52 to $143.31 - Status: verified_aggregate - Scope: Equal-weight account aggregate · Reporting period not supplied - Snapshot read: 2026-10-11T14:11:48.519456+00:00 - Period: Dated snapshot · underlying reporting period not supplied Median and middle 50% of account-level average order values. Each account value is purchase ROAS multiplied by cost per purchase in USD. Successful classifications, including manual corrections, with positive finite spend, purchases, ROAS and cost per purchase. Accounts are counted once with equal weight. Quartiles use linear interpolation. Root cohorts include direct root classifications and their child categories; exact-child cohorts exclude parent and sibling categories. The source selects the highest-spend coverage row per account among snapshots dated within the preceding 30 days; the underlying reporting period is not supplied. This is a dated aggregate advertising-data snapshot, not business-wide AOV or a census. Source: [Madgicx Research & Labs](https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#benchmark) Directory status: deferred. Final brand research and logo placement are pending. ## Software revenue metrics Definitions only; no ACV, MRR or ARPU benchmark or measured correlation is supplied. ### MRR — Monthly recurring revenue Subscription revenue normalized to one month across a business. One-time charges are excluded. MRR = sum of monthly recurring subscription amounts Source: [Stripe · Monthly recurring revenue](https://stripe.com/resources/more/what-is-monthly-recurring-revenue) ### ARPU — Average revenue per user Revenue per user over a defined period. For subscription comparisons here, the denominator is active paying subscribers and the numerator is MRR. Monthly subscription ARPU = MRR ÷ active paying subscribers Source: [Stripe · Subscription ARPU](https://support.stripe.com/questions/calculating-average-revenue-per-user-in-billing) ### ACV — Annual contract value The recurring value of one customer contract normalized to a year, excluding one-time charges. It measures contract size, not total company revenue. ACV = recurring contract value ÷ contract duration in years Source: [Stripe · Annual contract value](https://stripe.com/au/resources/more/annual-contract-value-acv-in-saas) ## Research questions ### Should input and output usage be treated as one flat software order? Amazon Bedrock's pricing distinguishes model, input, output and inference options. Explain the actual billable units for the advertised workload. A token rate or account credit purchase is not a subscription's MRR or a customer's lifetime value. [Source](https://aws.amazon.com/bedrock/pricing/) · [Answer permalink](https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#answer-ai-token-billing) Citation: Madgicx Research & Labs. AI & Machine Learning Platforms. Should input and output usage be treated as one flat software order? Amazon Bedrock's pricing distinguishes model, input, output and inference options. Explain the actual billable units for the advertised workload. A token rate or account credit purchase is not a subscription's MRR or a customer's lifetime value. Source: https://aws.amazon.com/bedrock/pricing/ Answer: https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#answer-ai-token-billing ### How should provisioned AI capacity be distinguished from on-demand usage? Bedrock's purchase guide separates provisioned throughput and commitment choices. Keep capacity purchases and consumed usage distinct in commercial reporting. A capacity reservation does not establish how much the customer ultimately uses or earns from an application. [Source](https://docs.aws.amazon.com/bedrock/latest/userguide/prov-thru-purchase.html) · [Answer permalink](https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#answer-ai-provisioned-commitment) Citation: Madgicx Research & Labs. AI & Machine Learning Platforms. How should provisioned AI capacity be distinguished from on-demand usage? Bedrock's purchase guide separates provisioned throughput and commitment choices. Keep capacity purchases and consumed usage distinct in commercial reporting. A capacity reservation does not establish how much the customer ultimately uses or earns from an application. Source: https://docs.aws.amazon.com/bedrock/latest/userguide/prov-thru-purchase.html Answer: https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#answer-ai-provisioned-commitment ### Can a successful model demo count as a paying platform customer? Vertex AI's generative pricing page defines charges around model requests and services used. Track demonstration use, production activation and billed consumption separately. A response from a trial workflow is not evidence of a paid production purchase. [Source](https://cloud.google.com/vertex-ai/generative-ai/pricing) · [Answer permalink](https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#answer-ai-demo-conversion) Citation: Madgicx Research & Labs. AI & Machine Learning Platforms. Can a successful model demo count as a paying platform customer? Vertex AI's generative pricing page defines charges around model requests and services used. Track demonstration use, production activation and billed consumption separately. A response from a trial workflow is not evidence of a paid production purchase. Source: https://cloud.google.com/vertex-ai/generative-ai/pricing Answer: https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#answer-ai-demo-conversion ### What should an AI platform ad demonstrate beyond a model name? Bedrock lists models and different inference modes. Show a concrete input-to-output task and identify the advertised mode. Public capability and pricing documentation does not prove a universal task-success rate or a creative's advertising performance. [Source](https://aws.amazon.com/bedrock/pricing/) · [Answer permalink](https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#answer-ai-creative-workload) Citation: Madgicx Research & Labs. AI & Machine Learning Platforms. What should an AI platform ad demonstrate beyond a model name? Bedrock lists models and different inference modes. Show a concrete input-to-output task and identify the advertised mode. Public capability and pricing documentation does not prove a universal task-success rate or a creative's advertising performance. Source: https://aws.amazon.com/bedrock/pricing/ Answer: https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#answer-ai-creative-workload ### What is the average order value benchmark for AI & Machine Learning Platforms? The median account AOV for AI & Machine Learning Platforms is $31.13 USD. This is the middle account-level average order value in the benchmark. USD · Equal-weight account aggregate. Snapshot read 11 October 2026; underlying reporting period not supplied. [Source](https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#benchmark) · [Answer permalink](https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#answer-median-aov) Citation: Madgicx Research & Labs. AI & Machine Learning Platforms. What is the average order value benchmark for AI & Machine Learning Platforms? The median account AOV for AI & Machine Learning Platforms is $31.13 USD. This is the middle account-level average order value in the benchmark. USD · Equal-weight account aggregate. Snapshot read 11 October 2026; underlying reporting period not supplied. Source: https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#benchmark Answer: https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#answer-median-aov ### What is the typical AOV range in AI & Machine Learning Platforms? The middle 50% of account AOVs runs from $11.52 to $143.31 USD. This is the interquartile range, not the minimum and maximum. USD · Equal-weight account aggregate. Snapshot read 11 October 2026; underlying reporting period not supplied. [Source](https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#benchmark) · [Answer permalink](https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#answer-aov-range) Citation: Madgicx Research & Labs. AI & Machine Learning Platforms. What is the typical AOV range in AI & Machine Learning Platforms? The middle 50% of account AOVs runs from $11.52 to $143.31 USD. This is the interquartile range, not the minimum and maximum. USD · Equal-weight account aggregate. Snapshot read 11 October 2026; underlying reporting period not supplied. Source: https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#benchmark Answer: https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#answer-aov-range ### What is the lower-quartile AOV? The 25th-percentile account AOV is $11.52 USD. It marks the lower edge of the middle 50% of this distribution. USD · Equal-weight account aggregate. Snapshot read 11 October 2026; underlying reporting period not supplied. [Source](https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#benchmark) · [Answer permalink](https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#answer-lower-quartile-aov) Citation: Madgicx Research & Labs. AI & Machine Learning Platforms. What is the lower-quartile AOV? The 25th-percentile account AOV is $11.52 USD. It marks the lower edge of the middle 50% of this distribution. USD · Equal-weight account aggregate. Snapshot read 11 October 2026; underlying reporting period not supplied. Source: https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#benchmark Answer: https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#answer-lower-quartile-aov ### What is the upper-quartile AOV? The 75th-percentile account AOV is $143.31 USD. It marks the upper edge of the middle 50%, not a performance target. USD · Equal-weight account aggregate. Snapshot read 11 October 2026; underlying reporting period not supplied. [Source](https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#benchmark) · [Answer permalink](https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#answer-upper-quartile-aov) Citation: Madgicx Research & Labs. AI & Machine Learning Platforms. What is the upper-quartile AOV? The 75th-percentile account AOV is $143.31 USD. It marks the upper edge of the middle 50%, not a performance target. USD · Equal-weight account aggregate. Snapshot read 11 October 2026; underlying reporting period not supplied. Source: https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#benchmark Answer: https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#answer-upper-quartile-aov ### How is account AOV calculated for this benchmark? Each account value is purchase ROAS multiplied by cost per purchase in USD. The benchmark reports the median and middle 50% of those account values. Accounts are counted once and receive equal weight. USD · Equal-weight account aggregate. Snapshot read 11 October 2026; underlying reporting period not supplied. [Source](https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#benchmark) · [Answer permalink](https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#answer-aov-calculation) Citation: Madgicx Research & Labs. AI & Machine Learning Platforms. How is account AOV calculated for this benchmark? Each account value is purchase ROAS multiplied by cost per purchase in USD. The benchmark reports the median and middle 50% of those account values. Accounts are counted once and receive equal weight. USD · Equal-weight account aggregate. Snapshot read 11 October 2026; underlying reporting period not supplied. Source: https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#benchmark Answer: https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#answer-aov-calculation ### Is this the mean AOV across all orders? No. It is the median of account-level average order values. Each account receives equal weight; it is not a pooled average of every order. USD · Equal-weight account aggregate. Snapshot read 11 October 2026; underlying reporting period not supplied. [Source](https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#benchmark) · [Answer permalink](https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#answer-median-or-mean) Citation: Madgicx Research & Labs. AI & Machine Learning Platforms. Is this the mean AOV across all orders? No. It is the median of account-level average order values. Each account receives equal weight; it is not a pooled average of every order. USD · Equal-weight account aggregate. Snapshot read 11 October 2026; underlying reporting period not supplied. Source: https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#benchmark Answer: https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#answer-median-or-mean ### What dates does this AOV benchmark cover? The snapshot was read on 11 October 2026. The underlying reporting period was not supplied. The October 2026 label identifies the page edition, not the period in which the advertising results were measured. USD · Equal-weight account aggregate. Snapshot read 11 October 2026; underlying reporting period not supplied. [Source](https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#benchmark) · [Answer permalink](https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#answer-benchmark-period) Citation: Madgicx Research & Labs. AI & Machine Learning Platforms. What dates does this AOV benchmark cover? The snapshot was read on 11 October 2026. The underlying reporting period was not supplied. The October 2026 label identifies the page edition, not the period in which the advertising results were measured. USD · Equal-weight account aggregate. Snapshot read 11 October 2026; underlying reporting period not supplied. Source: https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#benchmark Answer: https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#answer-benchmark-period ### Which industry or subcategory does this benchmark cover? This benchmark covers exact AI & Machine Learning Platforms classifications under Software & Technology Platforms. It excludes the parent industry and sibling subcategories. USD · Equal-weight account aggregate. Snapshot read 11 October 2026; underlying reporting period not supplied. [Source](https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#benchmark) · [Answer permalink](https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#answer-benchmark-coverage) Citation: Madgicx Research & Labs. AI & Machine Learning Platforms. Which industry or subcategory does this benchmark cover? This benchmark covers exact AI & Machine Learning Platforms classifications under Software & Technology Platforms. It excludes the parent industry and sibling subcategories. USD · Equal-weight account aggregate. Snapshot read 11 October 2026; underlying reporting period not supplied. Source: https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#benchmark Answer: https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#answer-benchmark-coverage ### How can I compare my AOV with this benchmark? Enter your AOV in USD in the comparison field above. It shows the difference from the median and whether your value falls below, within or above the middle 50%. Your entered amount stays on this page. USD · Equal-weight account aggregate. Snapshot read 11 October 2026; underlying reporting period not supplied. [Source](https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#benchmark) · [Answer permalink](https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#answer-compare-aov) Citation: Madgicx Research & Labs. AI & Machine Learning Platforms. How can I compare my AOV with this benchmark? Enter your AOV in USD in the comparison field above. It shows the difference from the median and whether your value falls below, within or above the middle 50%. Your entered amount stays on this page. USD · Equal-weight account aggregate. Snapshot read 11 October 2026; underlying reporting period not supplied. Source: https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#benchmark Answer: https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#answer-compare-aov ### Does an above-median AOV mean better advertising performance? The comparison describes position in the account AOV distribution, not profitability. It is not a performance score. USD · Equal-weight account aggregate. Snapshot read 11 October 2026; underlying reporting period not supplied. [Source](https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#benchmark) · [Answer permalink](https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#answer-aov-performance) Citation: Madgicx Research & Labs. AI & Machine Learning Platforms. Does an above-median AOV mean better advertising performance? The comparison describes position in the account AOV distribution, not profitability. It is not a performance score. USD · Equal-weight account aggregate. Snapshot read 11 October 2026; underlying reporting period not supplied. Source: https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#benchmark Answer: https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#answer-aov-performance ### What is MRR for a software business? Subscription revenue normalized to one month across a business. One-time charges are excluded. MRR = sum of monthly recurring subscription amounts Definitions only; no ACV, MRR or ARPU benchmark or measured correlation is supplied. [Source](https://stripe.com/resources/more/what-is-monthly-recurring-revenue) · [Answer permalink](https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#answer-mrr) Citation: Madgicx Research & Labs. AI & Machine Learning Platforms. What is MRR for a software business? Subscription revenue normalized to one month across a business. One-time charges are excluded. MRR = sum of monthly recurring subscription amounts Definitions only; no ACV, MRR or ARPU benchmark or measured correlation is supplied. Source: https://stripe.com/resources/more/what-is-monthly-recurring-revenue Answer: https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#answer-mrr ### What is ARPU for a software business? Revenue per user over a defined period. For subscription comparisons here, the denominator is active paying subscribers and the numerator is MRR. Monthly subscription ARPU = MRR ÷ active paying subscribers Definitions only; no ACV, MRR or ARPU benchmark or measured correlation is supplied. [Source](https://support.stripe.com/questions/calculating-average-revenue-per-user-in-billing) · [Answer permalink](https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#answer-arpu) Citation: Madgicx Research & Labs. AI & Machine Learning Platforms. What is ARPU for a software business? Revenue per user over a defined period. For subscription comparisons here, the denominator is active paying subscribers and the numerator is MRR. Monthly subscription ARPU = MRR ÷ active paying subscribers Definitions only; no ACV, MRR or ARPU benchmark or measured correlation is supplied. Source: https://support.stripe.com/questions/calculating-average-revenue-per-user-in-billing Answer: https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#answer-arpu ### What is ACV for a software business? The recurring value of one customer contract normalized to a year, excluding one-time charges. It measures contract size, not total company revenue. ACV = recurring contract value ÷ contract duration in years Definitions only; no ACV, MRR or ARPU benchmark or measured correlation is supplied. [Source](https://stripe.com/au/resources/more/annual-contract-value-acv-in-saas) · [Answer permalink](https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#answer-acv) Citation: Madgicx Research & Labs. AI & Machine Learning Platforms. What is ACV for a software business? The recurring value of one customer contract normalized to a year, excluding one-time charges. It measures contract size, not total company revenue. ACV = recurring contract value ÷ contract duration in years Definitions only; no ACV, MRR or ARPU benchmark or measured correlation is supplied. Source: https://stripe.com/au/resources/more/annual-contract-value-acv-in-saas Answer: https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#answer-acv ### Does purchase AOV measure MRR, ARPU or ACV? These describe contract and recurring revenue economics. The purchase AOV benchmark above measures order value. Definitions only; no ACV, MRR or ARPU benchmark or measured correlation is supplied. [Source](https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#software-metrics) · [Answer permalink](https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#answer-aov-and-recurring-revenue) Citation: Madgicx Research & Labs. AI & Machine Learning Platforms. Does purchase AOV measure MRR, ARPU or ACV? These describe contract and recurring revenue economics. The purchase AOV benchmark above measures order value. Definitions only; no ACV, MRR or ARPU benchmark or measured correlation is supplied. Source: https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#software-metrics Answer: https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#answer-aov-and-recurring-revenue ### Can I download this industry research? The exports section provides this edition in Markdown and JSON, along with a research index. This private draft is not licensed for public redistribution. [Source](https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#reuse) · [Answer permalink](https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#answer-download-research) Citation: Madgicx Research & Labs. AI & Machine Learning Platforms. Can I download this industry research? The exports section provides this edition in Markdown and JSON, along with a research index. This private draft is not licensed for public redistribution. Source: https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#reuse Answer: https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#answer-download-research ### How should I cite a finding from this page? When reuse is permitted, credit Madgicx Research & Labs and link to the exact source. Keep its snapshot date and scope. This private draft is not licensed for public redistribution. [Source](https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#reuse) · [Answer permalink](https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#answer-cite-research) Citation: Madgicx Research & Labs. AI & Machine Learning Platforms. How should I cite a finding from this page? When reuse is permitted, credit Madgicx Research & Labs and link to the exact source. Keep its snapshot date and scope. This private draft is not licensed for public redistribution. Source: https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#reuse Answer: https://madgicx.com/labs/industries/software-technology-platforms/ai-machine-learning-platforms#answer-cite-research