Postscript Updates Unified Profiles, Memories, Multi-Model Attribution and SMS Analytics MCP
In October 2026, Postscript introduced four updates. On October 23, the company added three features to the Postscript 2.0 Agentic Context Platform (ACP): Unified Profiles to consolidate customer records, Memories to preserve interaction context, and Multi-Model Attribution to compare five models for assigning revenue credit across marketing touchpoints. On October 29, Postscript separately introduced Postscript 1.0 MCP, a read-only connection that lets AI assistants analyze SMS campaigns. The first three updates expand personalization across SMS, RCS, and email, while MCP offers another way to explore SMS marketing performance without modifying campaigns.
Unified Profiles: Connecting Customer Data Across Channels
Postscript introduced Unified Profiles, a mechanism for automatically detecting and merging duplicate customer profiles.
The problem arises when the same shopper uses multiple email addresses, phone numbers, or accounts when making purchases. Traditional marketing systems may store those interactions in separate profiles even though they belong to the same person.
As a result, marketers see an incomplete customer history: some orders may be tied to a work email address, others to a personal address, while SMS conversations are stored separately.
Unified Profiles automatically merges detected duplicates in the background. After consolidation, SMS, RCS, and email use a shared customer record and interaction context.
For example, a shopper who previously purchased under different email addresses may be recognized as an existing customer rather than two new subscribers. This helps prevent repeat first-purchase offers and supports a more accurate understanding of order history.
The feature is also intended to reduce costs associated with storing and paying for duplicate profiles.
However, automatic merging depends on the system's ability to establish that different identifiers belong to the same person. Postscript does not claim that every anonymous visitor can be identified automatically or that Unified Profiles removes the need to obtain consent for marketing communications.
Memories: Preserving Customer Conversation Context and Preferences
The second update, Memories, is designed to work with information shoppers share with a brand in text conversations, reviews, and surveys.
This information often remains unstructured and is not used in subsequent marketing interactions.
For example, a shopper might mention in an SMS conversation that they are looking for a gift for a relative with specific dietary restrictions. In a conventional CRM, that detail may never become an attribute available for personalizing future messages.
Memories extracts meaningful information from conversations and other sources and saves it to the customer profile.
The system distinguishes between long-term and temporary facts. Stable preferences may remain relevant for an extended period, while short-term circumstances gradually receive less weight in subsequent communications.
For instance, information about a lasting dietary restriction may remain useful, while details about a recent wedding should not have the same influence on personalization a year later.
This approach allows the platform to use not just predefined profile properties but also context developed through the shopper's conversations with the brand.
Unlike conventional segmentation, in which a marketer defines a set of conditions in advance, Memories allows AI agents to consider an individual customer's circumstances when selecting message content.
Preserving context does not guarantee that every extracted statement is accurate or remains current. Personalization quality depends on the accuracy of the original information and the system's ability to interpret changes over time.
Multi-Model Attribution: Five Ways to Assign Revenue Credit
The third update focuses on marketing analytics.
Postscript added Multi-Model Attribution, a tool that lets marketers switch between five attribution models and compare their results within a single platform.
The supported models are:
- Last touch: All attributed revenue is assigned to the final interaction before a purchase.
- First touch: Revenue is assigned to the first interaction.
- Linear: Credit is divided equally among the included touchpoints.
- Position-based: More weight is assigned to the first and last interactions.
- Time decay: Interactions closer to the purchase receive greater weight.
Marketers can switch models and compare how the contribution of individual channels to attributed revenue changes.
For example, a shopper might first visit a website from a Meta ad, then receive an SMS from the brand, and finally open an email shortly before making a purchase.
Last touch assigns the result to the final interaction, while Linear distributes credit across all included touchpoints. Under Time decay, later interactions receive greater weight.
Multi-Model Attribution also offers configurable attribution windows. Users can inspect the journey behind a particular order and see which interactions were included in the revenue allocation.
This gives marketers multiple ways to analyze the same results instead of relying on a single measurement rule.
Comparing SMS, RCS, and Email in a Shared Analytics Framework
Multi-Model Attribution is part of the broader Agentic Context Platform architecture.
Postscript combines customer data and communication outcomes across SMS, RCS, and email to support a common personalization and analytics framework.
For marketers, this means evaluating customer interactions across several channels within one customer profile.
For example, Last touch may associate a large share of revenue with email if subscribers often open messages just before buying. First touch or Position-based models, however, may show a more substantial contribution from an earlier SMS interaction.
Comparing models helps reveal these differences and avoid management decisions based exclusively on one attribution rule.
Multi-Model Attribution is not incremental measurement. All five models redistribute credit for purchases that have already been recorded according to predefined rules. They cannot determine whether a purchase would have occurred without the marketing exposure.
Evaluating additional revenue caused by a channel still requires experimental methods, including control groups.
The new capabilities therefore improve attribution transparency but do not turn rule-based attribution into a causal evaluation of marketing effectiveness.
Postscript 1.0 MCP: Analyzing SMS Performance With AI Assistants
On October 29, 2026, Postscript introduced Postscript 1.0 MCP, a separate Model Context Protocol connection for examining SMS marketing results with compatible AI assistants. This fourth update complements the October changes to customer data and attribution in Postscript 2.0.
The integration operates in read-only mode. An assistant can access the available SMS performance data and help marketers investigate campaign results. For example, a user might request an analysis of metric trends, a comparison of campaigns, or identification of messages needing further review. The specific data available depends on the capabilities exposed by the connected MCP server.
The MCP connection cannot edit campaigns, change automation settings, or send messages on the user's behalf. Its purpose is analytical access, not autonomous SMS campaign management.
Postscript published setup instructions in its help center, but the announcement did not specify plan requirements. Availability for a particular account should therefore be checked separately. It is important to distinguish the products: Unified Profiles, Memories, and Multi-Model Attribution belong to Postscript 2.0 and its Agentic Context Platform, while the analytics MCP connection is identified as Postscript 1.0.
The October 29 update expands the ways marketers can work with campaign data. Postscript 2.0 builds a shared foundation for customer context and attribution models, while Postscript 1.0 MCP provides AI assistants with a separate read-only interface for exploring SMS results.
Agentic Context Platform: A Shared Foundation for AI Agents
Unified Profiles, Memories, and Multi-Model Attribution operate within the Agentic Context Platform introduced by Postscript in September 2026.
The company describes ACP as an architecture that not only stores customer data but makes it available to AI agents making decisions about marketing communications.
The platform brings together three categories of information: customer context, brand context, and results from marketing experiments.
Customer context includes purchases, website behavior, and information gathered from shopper conversations. Brand context covers products, promotional offers, business rules, and communication tone. Experiment results allow agents to consider the performance of previously tested message variations.
Using this information, Postscript's AI agents can select recipients, send times, and message content.
In particular, Postscript is developing Shopper for individualized customer interactions and Infinity Testing for automated testing of marketing message variations.
Within this architecture, Unified Profiles consolidates identifiers, Memories preserves relevant customer context, and Multi-Model Attribution analyzes how revenue is credited across interactions.
All three features are included in Postscript 2.0 plans. The company states that ACP is provided without a separate fee as a core component of the new platform.
About Postscript and Postscript 2.0
Postscript is an ecommerce marketing communications platform that originally specialized in SMS marketing for Shopify stores. With Postscript 2.0, the company is expanding into RCS and email, bringing these channels together with AI personalization, customer data management, and marketing analytics tools.
The Agentic Context Platform serves as shared infrastructure for customer profiles, brand context, and AI agents that manage marketing communications.
October's releases cover four areas. Unified Profiles reduces customer data fragmentation across identifiers and channels. Memories makes information from conversations, reviews, and surveys available for future interactions. Multi-Model Attribution provides five ways to analyze the contribution of marketing touchpoints to revenue. Separately, Postscript 1.0 MCP lets AI assistants explore SMS campaign performance in read-only mode.
For ecommerce marketers, the most significant shift is from segmentation based on a limited set of structured attributes toward the use of richer customer context in automated communications.
The effectiveness of this approach depends on accurate profile matching, reliable extracted data, and the rules governing AI agent behavior. Differences among attribution models also require careful interpretation, since none independently measures incremental impact.
Together, Unified Profiles, Memories, and Multi-Model Attribution advance Postscript 2.0 as a unified system for customer context, message personalization, and revenue attribution. Postscript 1.0 MCP, released on October 29, supplements those changes with a separate read-only interface for SMS analysis through AI assistants without granting permission to modify campaigns.
Trust Us Transform Your Performance Marketing.
Connect customer data, messaging performance, and attribution insights to make more informed retention marketing decisions.
Explore Affect Group ServicesSources
- Postscript. Product Releases: Unified Profiles, Memories and Multi-Model Attribution, October 23, 2026. Postscript product releases.
- Postscript. Beyond the CDP: How the Agentic Context Platform Puts Customer Data to Work. Postscript blog.
- Postscript. Agentic Context Platform. ACP overview.
- Postscript. Postscript Launches Email on a New AI-Native Retention Marketing Platform. Postscript announcement.







