Mike HickmanGroup Creative Director, AKQA

Case 03 · Nike · 2018

Without missing a beat.

Giving the mastery somewhere to go.

The agents in Iowa were the best I’d seen.

Not in spite of the tools. Because of what they’d taught themselves to do around them. Twenty-plus pieces of software (inventory systems, order trackers, a standard calculator) running simultaneously, and these agents moved between them without missing a beat. Copying from here, pasting to there, waiting for results, holding the conversational space while they worked. Servicing one customer or five depending on the channel, never losing the thread.

It was something to watch. The distance between what they were working with and what Nike had asked us to build toward was enormous.

The brief was a consumer-level experience for internal agents. I walked in thinking the work was about the interface, bringing Nike’s design standard inward, raising the visual quality of the tools to match the consumer products.

By the end of the first morning, the brief looked wrong.

Fig.01 — One request, across the field
Twenty-plus applications open at once · nine named Inventory Order tracker Calculator CRM Returns DB Shoe catalog Email Phone log Notes One tick = one hand-off: Copy, paste, wait for results. Cut · by the end of the first morning. The brief was the interface. The copying and pasting had become most of the job. Twenty-plus open at once Nine named Inventory Order tracker Calculator CRM Returns DB Shoe catalog Email Phone log Notes One tick = one hand-off: Copy, paste, wait for results. Cut · the brief was the interface. The copying and pasting had become most of the job.

One tick = one hand-off between applications; nine stops, eight hand-offs, one request. The nine windows and their positions are the case’s recorded before-state; the order of the traverse is not — the case names the moves, not their sequence. Dashed outlines are the brief as written: raise the visual quality of each window. The field is flat because the field was flat: a workstation is a surface, and a request crossing it is an adjacency, not a depth. The route is laid down once, at reading pace, and rests; each pause is one hand-off, and they are equal because the record does not time them.

The mastery was real, but it was aimed at the wrong target. Their skill was compensating for fragmentation instead of reaching the customer. Watch an agent long enough and you stop seeing a workaround and start seeing the role itself. The copying and pasting had become most of the job.

We needed to build something that gave the mastery somewhere to go.

Fig.02 — The workstation, reassigned
Before: twenty-plus applications open at once, nine shown. After: four zones, each naming what it absorbed. Customer — CRM, order tracker, notes — order history, loyalty, years with Nike, surfaced before the first question, marked “before you ask”. Conversation — email, phone log — one channel, one thread. Snippets — shoe catalog — agent knowledge, made reusable. Tools — inventory, calculator, returns DB — unified into one surface. BEFORE 20+ OPEN AT ONCE · NINE SHOWN AFTER FOUR ZONES · EACH NAMES WHAT IT ABSORBED CUSTOMER CRM · ORDER TRACKER · NOTES ORDER HISTORY · LOYALTY · YEARS WITH NIKESURFACED BEFORE THE FIRST QUESTION CONVERSATION EMAIL · PHONE LOG ONE CHANNELONE THREAD SNIPPETS SHOE CATALOG AGENT KNOWLEDGE,MADE REUSABLE TOOLS INVENTORY · CALCULATOR · RETURNS DB UNIFIED INTO ONE SURFACE BEFORE YOU ASK BEFORE 20+ OPEN AT ONCE · NINE SHOWN AFTER · FOUR ZONES CUSTOMER CRM · ORDER TRACKER,NOTES ORDER HISTORY · LOYALTY ·YEARS WITH NIKE, SURFACEDBEFORE THE FIRST QUESTION CONVERSATION EMAIL · PHONE LOG ONE CHANNEL, ONE THREAD SNIPPETS SHOE CATALOG AGENT KNOWLEDGE,MADE REUSABLE TOOLS INVENTORY · CALCULATOR,RETURNS DB UNIFIED INTO ONE SURFACE BEFORE YOU ASK

One block = one zone of the rebuilt workstation; the names inside it are the applications it absorbed. Nine of twenty-plus are shown — the nine the case’s before-state figure names. Average handle time fell five minutes, measured across three centers.

We unified what they’d been copying and pasting between. The anecdotes agents had accumulated over years (about shoe models, release histories, things no customer would think to ask) we surfaced as snippets, so more customers got the benefit of what the agents already knew. Amplify what made them good rather than replace it.

The real fight was the automation.

A phone number (or an email address in the email channel) could pull up a customer’s entire history. Order records. Brand loyalty. Years of relationship with Nike. We wanted that to surface automatically, before the agent had to ask a single question, so they could arrive in the conversation already knowing who they were talking to. Make customers feel known without crossing into surveillance territory. That line became a design principle.

Fig.03 — What surfaces, and where it stops
SURFACED BEFORE THE AGENT ASKS A SINGLE QUESTION PHONE NUMBER OR EMAIL, IN THE EMAIL CHANNEL ORDER RECORDS BRAND LOYALTY YEARS OF RELATIONSHIP FEEL KNOWN SURVEILLANCE NOTHING BEYOND THE LINE IS NAMED IN THE RECORD. THAT LINE BECAME A DESIGN PRINCIPLE. SURFACED BEFORE THE AGENT ASKS A SINGLE QUESTION PHONE NUMBER OR EMAIL, IN THE EMAIL CHANNEL ORDER RECORDS BRAND LOYALTY YEARS OF RELATIONSHIP FEEL KNOWN SURVEILLANCE NOTHING BEYOND THE LINE IS NAMED IN THE RECORD.

One card = one thing the agent already knew before asking. The three shown are the three the case names. What sat beyond the line was never written down, so the far side is drawn as an empty slot rather than filled in.

IT hadn’t scoped it. Hadn’t planned for it. So we built the argument and took it to product and leadership — the case that the return accumulated over time rather than landing as a single line item.

We found out afterward how close it came.

Without that argument, the project would have been shelved, millions already invested, organizational will exhausted. The design didn’t save it. Winning the argument did.

The platform shipped. Standard adoption across every Nike customer service center and warehouse.

Fig.04 — The argument, and the measurement
WHAT WAS ARGUED, AND WHAT WAS MEASURED ONE TICK = 30 SECONDS OF HANDLE TIME, PER CALL ARGUED 30 SEC / CALL MEASURED 5 MIN / CALL THREE CENTERS DEVELOPMENT · ONE-TIME IT HADN’T SCOPED IT — NO FIGURE IN THE RECORD × EVERY CUSTOMER SERVICE INTERACTION WHAT WAS ARGUED, AND WHAT WAS MEASURED ONE TICK = 30 SECONDS, PER CALL ARGUED · 30 SEC / CALL MEASURED · 5 MIN / CALL THREE CENTERS DEVELOPMENT · ONE-TIME NO FIGURE IN THE RECORD × EVERY CUSTOMER SERVICE INTERACTION

One tick = 30 seconds of handle time on one call — the unit the argument was made in. Argued: one tick, against a one-time development cost IT had not scoped, so its height is left unset. Measured afterwards: ten ticks, across three centers. The case states both numbers; it does not claim one produced the other.

I’ve carried that distinction since.

Back to the work.