People's Demographic Fracture identifies three audiences wearing a trenchcoat: web visitors (45–54), print subscribers (avg 38), and app/social users (Gen Z). The business question is whether one brand can serve three demographics.
The technology question is sharper. Each audience lives on a different identity surface — and the identity system People Inc built to win advertising is architecturally wrong for the products it now wants to sell.
Two Systems
D/Cipher is a content identity system. It identifies what content is about — 12,000 taxonomy terms, 100 million semantic connections, 12 billion intent signals — and serves ads based on that context. It was built to win advertising without knowing the user. No cookies. No login. No user profile. The system works precisely because it doesn't need to know who's reading.
This was a brilliant architectural bet. While the rest of the industry panicked over third-party cookie deprecation, People Inc had already built an ad engine that doesn't depend on user identity. D/Cipher's intent signals outperform cookie-based targeting because context is a better predictor of purchase intent than browsing history.
But the Inversion strategy needs a different system entirely. Turning brands into products — MyRecipes with saved collections, the People App with personalized feeds, InStyle's "The Intern" with viewer engagement — requires knowing who the user is. Purchase history. Preferences. Push targeting. Cross-brand recognition. Everything D/Cipher was designed to ignore.
| Capability | D/Cipher Content Identity |
Inversion Needs User Identity |
|---|---|---|
| What it knows | What the page is about | Who the person is |
| Login required | No | Yes |
| Privacy model | Privacy-first | Consent-gated |
| Scale | 200M monthly visits | 559K app downloads |
| Revenue contribution | $253M/quarter | TBD |
| Cross-brand capability | Full (all 40 brands) | 6 food brands only |
The table tells the whole story. Content identity scales across the portfolio and generates a billion dollars a year. User identity covers six food brands and is still looking for a revenue model.
Three Surfaces, Three Architectures
Web: identity-free by design. 200 million monthly visits. D/Cipher contextual targeting. The Graphene-Selene knowledge graph powers content classification. Data Studio (launched 2020) processes 12 billion intent signals. This surface works — 20% EBITDA margin — because it never asks users to identify themselves. Industry-wide, 90–98% of web visits are anonymous. D/Cipher doesn't fight this. It monetizes it.
App: identity-rich but tiny. 559,000 downloads. Keycloak for authentication. Push notifications, saved content, behavioral profiles. A 65-person team building the product that needs to know its users. MyRecipes expanded this with shared OAuth across six food brands — the first cross-brand identity experiment. But 559K authenticated users against 200M anonymous visits is a ratio problem, not an engineering problem.
Off-platform: zero identity. 21 billion views per quarter on TikTok, Instagram, Apple News — growing 43% annually. People Inc's fastest-growing surface, and the one where they know the least about who's watching. The platforms mediate every interaction. Feedfeed's thousand influencers add another layer of indirection. When someone watches a Food & Wine chef on TikTok, People Inc knows it happened. They don't know who watched.
The gap: D/Cipher covers the web surface completely. The app surface has authenticated identity but no bridge to D/Cipher's intelligence. The off-platform surface has neither. There is no identity system that spans all three.
The MyRecipes Experiment
MyRecipes is the one place where People Inc has tried to bridge the gap. Launched Q4 2024, it layers shared Keycloak authentication across six food brands — Allrecipes, Food & Wine, EatingWell, Southern Living, Better Homes & Gardens, Real Simple. Users get a save button SDK and cross-brand recipe collections. The brands get a shared identity layer.
This is the right architectural pattern. Shared auth, per-brand experiences, a Hightouch CDP composing user signals into actionable segments. The problem is scope. Six food brands out of forty. Recipe saves out of all possible product interactions. No connection to D/Cipher's content intelligence — the system that knows what content is about doesn't talk to the system that knows who saved it.
Extending MyRecipes to the full portfolio means answering questions the food vertical avoids. What does cross-brand identity mean between Investopedia and People magazine? Between Byrdie and The Spruce? Food has natural cross-brand behavior — someone who reads EatingWell recipes also reads Food & Wine restaurant reviews. Most brand pairings don't have that.
What Not to Build
The obvious impulse is a unified identity graph — one system that resolves anonymous web visitors, authenticated app users, and off-platform viewers into a single profile. Every CDP vendor will pitch this. It won't work.
Not because the technology is impossible, but because the surfaces are too different. Web users won't log in — there's no value exchange that justifies it at media scale. Off-platform users can't log in — the platforms won't share identity. And bridging anonymous web behavior to authenticated app behavior requires probabilistic matching that erodes the privacy-first positioning that makes D/Cipher valuable.
The better architecture accepts the fracture:
Don't unify the identity surfaces. Optimize each one independently, and build intelligence that flows between them without requiring identity resolution.
Concretely:
Web — keep D/Cipher contextual. It works. The value is in content intelligence, not user profiles. Push D/Cipher+ to third-party sites to expand the addressable surface. The inference cost is manageable at current scale.
App — build the authenticated identity layer properly. Keycloak → Hightouch CDP → personalization loop. Extend the MyRecipes pattern beyond food where the cross-brand behavior actually exists. Use the app identity to build LiveRamp segments that improve the authenticated portion of programmatic advertising — not to replace D/Cipher, but to complement it.
Off-platform — stop trying to own identity. Focus on content attribution instead: which content formats drive which platform outcomes. The Feedfeed acquisition makes sense as a content strategy, not an identity strategy. Invest in measurement partnerships (the metering problem) rather than identity infrastructure.
The Deeper Problem
In The Integration Machine, I identified three capabilities missing from People Inc's platform: per-brand application logic, unified identity, and independent deployment cadence. The identity gap is the hardest of the three.
Per-brand application logic is a software architecture problem — Graphene-Selene's microservices can probably support it. Independent deployment is a DevOps problem — solvable with investment. But identity is a business model problem. The ad business succeeds by not needing user identity. The product business requires it. These aren't two engineering teams that need to coordinate. They're two business models that make opposing demands on the same infrastructure.
The Integration Machine built uniformity. Every brand on the same CMS, same templates, same ad stack. That uniformity made D/Cipher powerful — more content surfaces mean more contextual signals. Now the Inversion strategy asks some brands to become products, which means building user relationships that D/Cipher was specifically designed to avoid building.
People Inc doesn't have an identity problem. It has two identity systems that serve two different businesses. The fracture isn't a bug. It's the architectural consequence of running an ad business and a product business on the same platform.
The question isn't how to unify them. It's whether you can run both without one starving the other.