Quick Verdict
Phrasee rebranded as Jacquard in 2024. As of July 2026, the Jacquard site, login, platform documentation, and demo process are the authoritative product surfaces; an old Phrasee article should not be used as a current feature or pricing source. Jacquard is an enterprise language platform for large brands and lifecycle-marketing teams. Its focus is not blog generation. It creates and optimizes high-frequency short messages across email subject lines and bodies, SMS, mobile push, in-app messages, web push, and other listed digital touchpoints, then learns through controlled campaigns and historical outcomes.
The platform fits organizations that already have CRM infrastructure, brand standards, meaningful send volume, and an experimentation program. Current materials emphasize Language² brand guardrails, performance prediction, Audience-Wide Optimization, contextual one-to-one messaging, omnichannel distribution, and language intelligence. It is not an individual copywriting app and does not publish a self-service price card; Book a Demo is the principal buying route. Brand calibration reduces drift but cannot guarantee safety, and a predicted winner is not proven incremental lift. Human approval and randomized evaluation remain necessary.
Best For
- Brands running frequent email, SMS, push, and lifecycle campaigns at sufficient volume.
- Global marketing teams that need a recognizable voice with regional and language variation.
- Organizations already using systems such as Salesforce, Adobe, Braze, Iterable, or Airship.
- Enterprises with CRM operations, data science, brand, privacy, and legal stakeholders.
- Teams seeking a governed short-message system rather than a general-purpose writing assistant.
It is a poor fit for a small blog program, a sender without a stable control group, or a team unable to provide clean historical outcomes. At low volume, random variation can overwhelm language effects. If approval ownership and data permissions are unresolved, automated variant production creates a larger queue rather than a faster system.
Key Features
- Language² and brand calibration: apply style, tone, and prohibited-language constraints before candidate prediction.
- Cross-channel short messaging: generate channel-specific variants from a strategic brief for email, SMS, mobile push, in-app, and web push.
- Audience-Wide Optimization: use experimental design to identify stronger language for a broad audience, not an unquestionable answer.
- Contextual Messaging: combine permitted product and profile attributes into individualized copy, subject to consent and data controls.
- Performance Prediction: rank candidates using historical copy experiments and explain why language may resonate.
- Enterprise governance: current pages describe SSO, roles, approvals, ISO 27001:2022, and privacy controls.
Use Cases
The brief should state the campaign objective, eligible audience, channel, verified product facts, offer, tone, prohibited wording, mandatory disclosures, and primary outcome. After Jacquard generates and calibrates candidates, marketers should inspect hierarchy and channel limits, brand owners should approve voice, and legal or compliance reviewers should validate claims and disclosures. Only approved versions should enter the campaign platform. A subject line must match the body and landing page; a push notification must not manufacture scarcity or use pressure against a vulnerable customer.
For evaluation, keep a stable control, randomize assignment, and hold timing, sender, eligibility, offer, landing page, and permission state as constant as practicable. Predefine the primary metric, sample size, duration, stopping rule, and correction for many variants. Open rates are distorted by mail privacy proxies, while clicks may not indicate qualified conversion. Analyze unsubscribes, complaints, conversion quality, and retention as well. Contextual messages also need cross-channel frequency and contradiction checks. Prediction proposes candidates; a controlled experiment supports causal inference.
Pricing
Jacquard does not publish standard list pricing. Obtain a written quote based on actual modules, channels, regions, languages, integrations, services, and volume rather than carrying forward a legacy Phrasee package.
| Product layer | Primary role | Confirm before purchase |
|---|---|---|
| Core Platform | Brand calibration, generation, and prediction | Calibration timeline, approvals, users, and output limits |
| Audience Optimisation | Broad-audience experimental optimization | Method, minimum volume, control handling, and result export |
| Personalised Campaigns | Contextual one-to-one messaging | Permitted fields, latency, consent, and sensitive-data restrictions |
| Integrations | CRM, ESP, and campaign connections | Supported versions, data flow, fallback, and additional fees |
| Global enterprise deployment | Languages, regions, governance, and services | Localization, SSO, SLA, support, and exit migration |
Review the official platform overview and integration catalog, then pilot the proposed scope with the buyer’s own workflow and control group.
Contextual copy may use product, location, preference, behavioral, or profile attributes. The customer should establish purpose and consent first, send only the minimum fields needed for message selection, and exclude unnecessary names, free text, health information, financial hardship, and other sensitive attributes. Jacquard states that enterprise LLMs are not allowed to train on customer data and describes SSO, roles, and approval controls. Procurement must still verify the data-processing agreement, subprocessors, regions, retention and deletion, logging, tenant isolation, incidents, and the model supply chain.
Brand safety is broader than correct spelling. A message can improve a short-term click metric while damaging trust, increasing complaints, or applying unfair pressure. Establish prohibited strategies, frequency caps, protections for vulnerable users, complaint response, and a human kill switch. Test whether translated rules preserve meaning across regions. Jacquard can provide controls and evidence, but the brand remains responsible for audience selection, truthful offers, privacy permission, and final delivery. Use its privacy policy and legal center as due-diligence inputs.
Pros
- Clear focus on high-volume email, SMS, and push language rather than generic writing.
- Connects brand calibration, generation, prediction, experimentation, and distribution.
- Multi-channel and multilingual governance fits global lifecycle-marketing organizations.
- Integrations can reduce manual movement of approved variants into campaign systems.
- Official materials address permissions, approvals, security certification, and LLM training use.
Cons
- Legacy Phrasee links and claims may be stale after the rebrand.
- No public standard pricing; calibration, integration, and service costs require discovery.
- Prediction needs adequate volume and clean experiment data to be useful.
- One-to-one contextual messaging expands consent, fairness, frequency, and data obligations.
- Guardrails require human judgment and can over-constrain language if configured poorly.
Alternatives
| Tool | Best for | Key difference from Jacquard |
|---|---|---|
| Persado | Regulated omnichannel content supply chains | Greater emphasis on financial compliance, production assets, and managed services |
| Jasper | Broad marketing content and brand knowledge | Wider long- and short-form scope; experiments are less central |
| Copy.ai | Sales and marketing workflow automation | General workflows rather than dedicated short-message language experiments |
| Grammarly Business | Organization-wide writing consistency | Gives inline employee guidance rather than send-time message optimization |
| ContentBot | Self-service blogs and batch drafts | Lower entry barrier and public pricing, with lighter enterprise experimentation |
FAQ
Are Phrasee and Jacquard separate products?
Not as current competing platforms. Phrasee changed its name to Jacquard in 2024. Current evaluation, login, procurement, and contractual claims should use Jacquard sources.
Can Jacquard replace A/B testing?
No prediction should be treated as causal fact. The platform can improve candidate generation and experiment design, but a buyer should retain randomized controls, predefined metrics, and complete analysis.
Does it only optimize email subject lines?
No. The current site lists email, SMS, mobile push, in-app, and web push among its channels. Confirm exact formats and integration behavior during the pilot.
Is customer data used to train models?
Jacquard states that it does not allow LLMs to train on customer data. Buyers should still bind exact models, logs, subprocessors, retention, deletion, and tenant isolation in the data-processing terms.
Can an individual or small team buy it directly?
The public site centers on booking a demo and does not show a self-service price card. Small teams should confirm minimum scope, implementation requirements, and total cost before assuming fit.
Bottom Line
Jacquard carries forward Phrasee’s specialization in brand marketing language while expanding into multichannel generation, audience experiments, and contextual messaging. It is most relevant to large brands that operate email, SMS, and push as a continuous experimental system. Confirm the post-rebrand modules, run a randomized pilot on the brand’s own traffic, and make privacy, frequency controls, brand safety, and human approval release conditions. Vendor case studies show possibilities; they are not causal evidence for a new buyer.