“Ancient” is not necessarily a bad thing, for Celtic or Viking artifacts — and the same is true of “old money”. But you don’t want your old money — in the new financial management account you just opened — either to sit idle or to be vulnerable to hackers. What to do, what to do….?!

The project

The project was challenging not so much because of the scope or sheer size — it was actually relatively small — but because of the fluidity of the requirements; they changed rapidly and often, and crucially important stakeholders were not always brought in until very late, when they would sometimes have game-changing input. Much of the design then had to be based on assumption based on scant information that might change the next day.

Similarly, the business rules that governed this seemingly small, simple application were astonishingly complex. Security was an enormous concern for the client, enough so that we spend quite a bit of time devision a method of referring to Active bank accounts that the end user had added to our client’s financial-management system, with vague and even evasive terminology, and only allowing live agents to provide any information regarding such accounts. This had the effect of creating a potentially very confusing situation in which the announced number of accounts could have a mismatch against the number of accounts the end user knew himself to have. This became the primary UX challenge to be met.

Finally, the list logic to be used in the event of multiple accounts was extraordinarily complex. The system needed to announce first the total number of accounts, then obliquely refer to Active, Suspended, or Needs Reauthorization accounts as something to be discussed “in a moment”, moving on to accounts either Pending Signature or Pending Verification. These accounts were then grouped by 3s, with any “leftovers” played out at the end. The user could choose to receive or skip details about either status, and could likewise choose to repeat any group of 3, move on to the next group, or stop hearing any additional statuses. The flow then had to contend with any other statuses, offering a live agent to the caller, to hear about those statuses, if they existed. Very complex, very challenging, but a very nice final user experience.

The modalities

This project was entirely for AI-assisted Voice IVR.

​The strategy

The design strategy focused on the complexity of providing enough information to be useful, without risking a potential breach of security. Within the information provided, the list navigation became the primary driver of the design. The attached call flow diagram shows just how unusually complex the design and the list logic needed to be, in order to sound simple, natural, easy to use, and conversational.

The outcome

After extensive and iterative revisions, to meet requirements of both the business needs and security needs of the client, the project goes live on 16 NOV 2024.

Take a listen

The sound file below is a simulation of one the more complex use cases we expect the system to encounter. (The “Westwind Financial Services” name is entirely fictitious, as is the web URL provided in the audio clip.)