Financial services · Enterprise workflows

Turning 100s of pages of specs into a campaign tool

Designed for the team that sends a financial services firm's advice to thousands of clients.

The campaign list after review: a band counting drafts, pending QA, pending approval and ready to launch, and rows with content type, status, launch, scope, channel and language.

My role & impact

Senior product designer. Led design with engineering, analysts and project leadership.

  • Four tools to one

    Designed one campaign tool to replace four separate tools and a process run by hand.

  • Aligned leadership on one flow

    Project leadership agreed to replace the signed-off spec's three flows with one. No requirement changed.

  • Rules others design by

    Set the design rules for campaign creation and onboarded a partner-firm designer to them.

Project
6 months · In build
A blocked campaign, with a banner saying it cannot be launched. The audience screen: an exclusions panel beside the client list, with included and excluded tabs. The campaign list: a band counting drafts, pending QA, pending approval and ready to launch, above rows with content type, status, scope, channel and language. Client details are blurred.

PROBLEM: What was broken?

Four tools, one campaign

The team that reaches the firm's clients assembled every campaign by hand: client data pulled from one place, filtered in another, drafted somewhere else, approved over email. One weekly campaign took days.

Before a campaign goes out, every client on the list has to be checked: have they consented to be contacted this way, and are they allowed to be offered this product. The team ran that check by hand, one client at a time. A single miss meant advice reaching a client who should never have received it, which is a compliance breach for the firm. Approvals happened over email, so there was no record of who had signed off.

INSIGHT: Do three campaign types need three flows?

Three types, same inputs

The signed-off spec described three campaign types as three flows. The requirements were long and time was short, so I used an approved AI tool to condense them and lay them against the UI fields.

Read step by step, the three types looked like three different things. Listed by what each step needs, the same inputs appeared under all three. Only the starting point differs: a piece of content, a group of clients, or a portfolio event.

The repetition had a source. The spec was written around an earlier prototype that already contained it, so the agreed document carried the prototype's structure along with the requirements.

DECISION: One flow or three?

One flow, one branch

The choice was to build the spec as agreed or to restructure it. I restructured: one flow where you choose how the campaign starts, and only the few things that differ change. Fields appear when a campaign needs them, a pattern taken from campaign tools that do the same job.

Hiding steps has a cost. Someone who never goes down a path may not learn the option exists. I took that trade because a small team uses the tool every day, and most campaigns are one-off sends built by someone who already knows which type they want.

That cost won't show in usage data, so the check I'd push for is a session with the team a few weeks after launch, on what they expected to find and couldn't.

Aligning leadership on one flow

A shorter flow no longer matched what had been signed off. Project leadership pushed to follow the spec. From where they stood, a shorter flow looked incomplete.

The screens couldn't settle it. On screen, a step that is gone and a step that hasn't been asked for yet look the same. So I mapped the old flow against the new one, requirement by requirement. Every line of the spec landed on a step, and showed up when a campaign needed it.

Leadership agreed to the restructure. No requirement changed, so the signed document stood. The functional spec now records where each requirement surfaces.

Every exclusion has a reason

Before

The first layout: included and excluded clients in two lists side by side, each row with a name and an email.

After

The final audience screen: an exclusions panel on the left with switches, and the client list with Included clients and Excluded clients tabs, rows by client ID.
  1. The team's choices and the campaign's scope. The team can switch these off.
  2. Who is in and who is out, one tab each. They are different checks.
  3. Rows show client IDs, because the bank masks personal details. Each excluded row says why.
  4. Bank rules: eligibility, consent and staff accounts. Locked, but kept visible with counts, so the approver can account for every removal.
The audience screen, from the first layout to the final one. Client details are blurred.

A campaign starts from a client scope, then people are added and removed until the list is right. A client can be left out for three reasons: the team's choice, how the campaign is aimed, or a bank rule. The team can reverse the first two. Bank rules stay locked.

The locked rules stay on screen anyway, with their counts. Nobody can act on them, but the approver reads the same list afterwards, and hiding those removals leaves a number they can't account for.

The first layout showed both lists side by side. Checking who is in and checking who is out are different jobs, so they became two tabs. The bank masks personal details, so rows show client IDs. That took the readable label off a screen where thousands of rows get checked, so the exclusion reason had to carry more.

A campaign can also be stopped at any step, from QA review to launch. Every stop shows the same blocked state, with a link to the reason.

The top of a blocked campaign: a Blocked pill under the campaign name, and a grey banner saying the campaign cannot be launched, with a See why link.
  1. The pill names the state, whichever step stopped the campaign.
  2. The banner says it cannot go out. See why opens the workflow, where the comments name the step.
The top of a blocked campaign.

REVIEW: What did client review change?

A view of where each campaign is

Before review

The campaign list before review: status filters above rows that show a name, status, dates, owner and channel.

After review

The campaign list after review: a band counting drafts, pending QA, pending approval and ready to launch, and rows with content type, status, launch, scope, channel and language.
  1. A band counts what is waiting: drafts, QA, approval and ready to launch.
  2. Each row now carries content type, number of clients and languages.
The campaign list, before and after client review.

The client reviewed the designs as they developed. The team said that once a campaign went for approval, they lost track of where it was. The campaign list gained a band that counts campaigns at each stage, and each row now shows its content type, number of clients and languages.

The platform is in build. At launch I would watch two things: whether a simple campaign gets built faster than a complex one, and how often a campaign comes back for rework. A shorter flow should not produce sloppier campaigns.

REFLECTION: What did I learn?

Look closely, and draw what you find.

  • Listing each step's inputs exposed the repetition.

    The spec was written around an earlier prototype and repeated its steps as requirements. I now list what each step needs before I design from a spec.

  • The map won leadership's agreement.

    It showed every requirement landing on a step in the new flow. Next time I'd draw it before showing any screens.

Say helloroopshri.ux@gmail.com