marketing · cheap to verify
Connectivity sales and retention steered by AI models (acquisition, upsell, win-back, retention)
Not sized
The Q2 call, release and 10-Q do not name AI in marketing, sales or retention; the head of Connectivity & Platforms credits data and analytics, without naming AI, for more targeted responses (claim c4).
inscrutabledisclosure: not mentioned· motive: exploratory· before LLMs: relabelled
Q2 is silent on the AI models. Broadband losses improved by to ; domestic broadband revenue changed to , and Connectivity & Platforms marketing and promotion rose to , which the 10-Q explains without AI (claim c6).
Evidence: 0 quotes, 8 figures, 3 confounds, 2 from before coverage
Broadband and wireless revenue gained or kept because AI models choose whom to target and with what offer. On the Q1 2026 call the head of Connectivity & Platforms says the company uses AI to improve transactional outcomes, running many models over many customer attributes to optimize acquisition, upsell, win-back and retention, and is enhancing its marketing tech stack for customization and personalization. He gives it as one example in a list of changes (data use, customer experience, purchase flows, unassisted channels, network reliability, sales effectiveness), and the CFO credits most of the quarter's improvement in broadband losses to the Legendary February offers (claim cmcsa-2026-cq1-c3). No source gives a measure for the models. The money would sit in domestic broadband and wireless service revenue; the cost of running the models sits, unreported, in Connectivity & Platforms marketing and promotion and other expenses. This is the door the ledger reads at Verizon as acquisition and retention spend lowered by AI micro-segmentation; Comcast speaks of outcomes (connects, upsell, retention) rather than cost, so it is read here as revenue.
Why this motive
Exploratory, carried from Q1; silence is not evidence about motive.
Before LLMs: relabelled
At the anchor the activity sits in Connectivity & Platforms marketing and promotion, which the fiscal 2024 10-K describes as the costs of attracting new customers and promoting service offerings (anchor claim cmcsa-anchor-c2): for FY2024, beside domestic broadband revenue of . The same 10-K named AI only as a technology the company risked failing to employ (anchor claim cmcsa-anchor-c1); neither anchor source describes models for acquisition or retention. Models that score customers for acquisition and retention predate large language models; the covered call describes them under the AI name with no line, price or volume credited to them, so the tag is relabelled (the tie-break).
“If we choose technology or equipment that is not as effective or attractive to consumers as that employed by our competitors, if we fail to employ technologies desired by consumers or that enhance our business operations, such as through the use of AI, or if we fail to execute effectively on our technology initiatives, our businesses and results of operations could be adversely affected.”
“Marketing and promotion expenses primarily consists of the costs associated with attracting new customers and promoting our service offerings.”
Figures
- Improvement in domestic broadband subscriber net losses, year over year · 2026-CQ2
- Domestic broadband subscriber net losses in the quarter · 2026-CQ2
- Domestic broadband revenue · 2026-CQ2
- Domestic broadband revenue, prior year (same row, 2025 three-month column, recast for the 2026 presentation) · 2025-CQ2
- Domestic broadband revenue change, year over year · 2026-CQ2
- Connectivity & Platforms marketing and promotion · 2026-CQ2
- Connectivity & Platforms marketing and promotion, prior year (same row, 2025 three-month column) · 2025-CQ2
- Connectivity & Platforms marketing and promotion growth, year over year · 2026-CQ2
What else could explain it
- transformation program: The go-to-market pivot credited in Q1 (claim c4) continues: the CFO says investment in the customer experience and go-to-market capabilities contributed to the decline in Connectivity & Platforms EBITDA (claim c7).
- other: Data and analytics, faster time to market and more targeted responses are named without AI (claim c4).
- relabel: Models that score customers for acquisition and retention predate large language models, and the anchor 10-K named AI only as a technology the company risked failing to employ (anchor claim cmcsa-anchor-c1).
By quarter
- Q1 2026described · described, no size · exploratory
- Q2 2026not mentioned · inscrutable · exploratory