New York Times co-holders did not cancel The Atlantic more often
Explicit Atlantic cancellation over about 14 months among direct-billed subscribers active in the first half of 2025, split by prior-year New York Times holding
Share with an explicit Atlantic cancellation
Source: YipitData U.S. e-receipt panel. Atlantic subscribers also paying The New York Times n=464, including 50 observed cancellations; Atlantic-only subscribers n=921, including 97 cancellations. Third-party Atlantic billing is excluded.
Chart data
| Series | Share with an explicit Atlantic cancellation |
|---|---|
| Atlantic + New York Times | 10.8% |
| Atlantic only | 10.5% |
How asymmetric was the Atlantic–New York Times overlap?
Across the trailing 18-month wallet window, 38.2% of Atlantic subscribers also paid The New York Times, while 4.3% of New York Times subscribers also paid The Atlantic. The shared group contained about 1,310 subscribers. Atlantic subscribers held an average 1.80 of seven measured news titles, compared with 1.17 among Times subscribers. These figures describe same-window co-holding, not which title came first.
Does the similar cancellation rate prove the titles are complements?
No. The 10.8% and 10.5% rates show that a measured New York Times relationship was not associated with higher Atlantic cancellation in this cohort. They do not establish acquisition sequence, whether one subscription supports the other, which title would be canceled first when spending contracts, or whether co-holding affects pricing sensitivity.
Assumptions & Methodologies
We analyzed YipitData’s U.S. e-receipt panel. Wallet overlap covers Feb. 2025–Aug. 2026 among active-paneled direct-billed subscribers. The cancellation comparison starts with 1,385 Atlantic subscribers active in the first half of 2025, separates 464 who also paid The New York Times in the prior 12 months from 921 with no measured Times relationship, and observes explicit Atlantic cancellation emails through mid-Aug. 2026. Third-party Atlantic billing is excluded because cancellation could not be confirmed consistently. The same cancellation signal is used for both groups, but capture still depends on receipt visibility. Co-holding is measured within a broad window rather than as daily active status. The follow-up contains 50 and 97 cancellation events, the competitive set is limited to seven major news titles, and the panel skews slightly male and upper-middle income. Results describe observed association, not sequence, substitution, complementarity, or causation.