Persado sells the words that make customers act

The bank had a perfectly respectable sentence. “Access cash from the equity in your home.” It explained the offer. It wasted no words. It sounded like a bank. In a Chase test described by Persado in 2019, a different version performed better: “It’s true - You can unlock cash from the equity in your home.” The offer was still home equity. The customer was being invited to imagine it differently.

That is an awkward result for anyone who has ever won an argument about copy by being the most senior person in the room. A sentence can be clear, tasteful and entirely defensible, yet lose to another sentence. Persado has built its business around making that discomfort useful.

A verb puts taste on trial

The Chase example is interesting because it is so small. Nobody needed a new product, a celebrity or an expensive location shoot to change the proposition’s framing. But the test also changed the call to action. It cannot establish that a single verb caused the improvement. The practical lesson is broader: a piece of creative is a collection of decisions, and those decisions can be examined.

Persado’s language work has long involved breaking messages into components, considering different emotional approaches and testing their effectiveness. Its early marketing materials described a large collection of tagged words and phrases, scored against response data. The unusual asset was the relationship between language and what an audience subsequently did.

A conventional writing assignment asks for a good subject line. Persado’s assignment asks which version makes a particular audience act. Those questions sometimes produce the same answer. Advertising would be a much tidier profession if they always did.

The experiment becomes a company

Alex Vratskides and Assaf Baciu came from mobile marketing company Upstream and founded Persado in 2012. Vratskides remains chief executive; Baciu is president. Their background matters: short digital messages are bounded problems, delivered repeatedly, with responses that can be counted. That is fertile territory for experimentation.

The company announced a $15 million Series A led by Bain Capital Ventures in February 2013, with plans to expand sales, increase research spending and establish its New York headquarters. A $21 million Series B followed in January 2015. Goldman Sachs led a $30 million Series C in April 2016, joined by existing investors including Citi Ventures and American Express Ventures.

Those three announced rounds total $66 million. That is the capital behind the company’s early expansion, rather than a price for using the product. For buyers today, the commercial structure combines enterprise software and managed services. Persado’s Create materials describe credits per asset; the public sales route is a demo and a discussion of scope.

The institutional shape is revealing. This is software sold to businesses trying to influence consumer behavior, with enough campaigns to learn from and enough operational complexity to justify specialist help. A small shop sending an occasional newsletter has a rather different problem.

Ally buys the argument

One reason the Chase story traveled is that other banks recognized the problem. In Persado’s Ally case study, David Hixon describes seeing the Chase announcement, considering its relevance to cross-selling, then receiving the same story from colleagues, including senior leaders. The attractive idea was making better use of existing customer relationships.

“Cross-sell is a foundational strategy for this company.”

David Hixon / Ally, in Persado’s case study

Ally wanted customers to consider more of its financial products. Persado’s published results describe doubled click-through rates, a 57% increase in new accounts in an email experiment, and 16% average conversion improvement across campaigns. These are reported customer results, with different denominators and scopes. They should be read separately.

Ally figures reported in Persado customer materials. Different measures; no shared scale or guaranteed outcome.

A click records interest. An account opening records another kind of action. The distinction is particularly useful when assessing an AI vendor: a remarkable number is more informative once you know what was counted.

The same mechanics extend beyond banking. Persado’s Vodafone Italy materials report a 42% conversion lift across more than 2,000 campaigns, including push and SMS messages used for upselling, cross-selling and winning back former customers. The channel changes; the question about language remains.

The inbox becomes a production line

Persado’s current offering has three doors. Create starts with a brief and supplies finished marketing assets. Optimize takes existing content and produces scored alternatives. Automate connects content selection and experimentation to recurring customer communications.

Create includes coded email, print-ready direct mail, banners, social assets and SMS. Its delivery package includes explanations and audit trails, with a dedicated team described as a content strategist, technical engineer and compliance specialist. The machine has acquired colleagues.

Optimize is aimed at the approved asset that remains disappointing in market. It scores content for performance potential and compliance risk, then suggests changes within defined guardrails. Teams can use self-service Studio or a managed Enterprise service. Persado’s claim that some refreshes avoid another approval cycle depends on staying inside the approved parameters.

Automate serves content through existing messaging systems. Persado lists connections to Salesforce Marketing Cloud, Adobe Campaign, Braze and Cordial. An email service provider still handles the send; Persado supplies variations through template scripts and an API. This places it inside the marketing stack rather than requiring the stack to be discarded.

Dynamic Email, launched in March 2025, makes the operational argument concrete. Teams set up an approved content library, implement the HTML and scripts, then keep updating content and experiments without repeatedly coding the campaign. Its launch announcement reported average conversion gains of 20% and email operations time reductions of up to 75% among financial-services users. Those figures describe the company’s reported experience, not a forecast for every inbox.

Legal joins the experiment

Writing more variants creates another workload: somebody must check them. A bank’s message can be persuasive and still create problems through an unsupported claim, misleading wording or a missing disclosure. Faster drafting can leave the next person in the queue buried.

In May 2025, Persado launched Marketing Compliance AI for financial-services marketing and legal teams. It describes agents that identify relevant regulations, examine content, assess brand and performance, and propose alternatives. Its launch announcement claimed reductions of up to 90% in legal review time. That is a workflow claim, separate from whether recipients buy anything.

The intriguing change is where the check happens. Persado wants regulatory and brand constraints applied while content is being made. A marketing team can then discuss the proposed language and the reason for a flagged issue together, before another round of work accumulates.

This positioning puts Persado between creative agencies, marketing software and regulatory technology. Internal teams and general-purpose language models remain alternatives; personalization products such as Salesforce Einstein and Movable Ink overlap with parts of the job. Persado’s case for purchase rests on combining response data, language experimentation, production and compliance controls. Buyers still need to evaluate those capabilities in their own environment.

Its employer materials offer a small, amusing clue to the company’s self-image: seven values spell PERSADO. They include Science and Owning the Truth, alongside People and Resilience. The stated benefits include hybrid work and learning support. It is an employer’s account of its culture, but the mixture of disciplines fits the product’s peculiar demands.

What the reader can steal

The accessible part of this approach requires a willingness to lose an argument. Keep a control message. Define the business outcome before looking at results. Test a different emotional framing, rather than polishing the same idea five times. Keep the offer and audience comparable where possible. Bring the people who approve claims into the brief.

These are practical inferences from the workflow, not a promise that a small experiment will reproduce a bank’s results. Thin audiences give weak evidence. Poor attribution makes winning difficult to recognize. A better sentence cannot repair an unattractive offer, and a model’s score cannot settle every question of regulatory judgment.

What failed in the opening example was the incumbent creative’s claim to being the best available choice. Evidence gave Chase a reason to try a different approach; the story then gave Ally a reason to investigate. The copyable habit is simple: let customer behavior interrupt the meeting. Sometimes the least fashionable person in the room is the one holding the results.