# From Excel to applause Heaps – case study by Seemant Shekhar. https://www.seemant.design/work/from-excel-to-applause Call-audit feedback at Heaps travelled by a daily Excel email: auditor to vendor team lead to caller. I replaced it with a dashboard where every piece of feedback has an owner, a deadline and a close. _Screens are blurred under NDA. I’m happy to walk through the real designs in a conversation._ - **Role**: Design lead - **Team**: Senior PM, ops and quality teams - **Users**: Auditors, vendor team leads, callers - **Status**: Shipped, 2026 - **Areas of influence**: Research with auditors and callers; Product rules with ops leads; Interaction design; Prototype in Lovable - **Highlights**: 20–30 min a day back for every auditor; Every item has an owner, a deadline and a close; Callers hear about their good calls too ## 01. The problem **Everything was written twice, then emailed.** Auditors scored each call in the CRM, then copied the same details into an Excel tracker and emailed the day’s errors to each vendor’s team lead. Nobody could see what happened after that. 1. Audit the call – Mark parameters and comment in the CRM 2. Copy into Excel – Same UMR, date and comment, again (the pain point) 3. Email the TL – One sheet per vendor, end of day 4. TL tells callers – Forwarded or said in person > **20–30m** – “After saving the audit, I copy-paste the details into the Excel sheet, which usually takes 20–30 minutes a day.” – Auditor ### What auditors and ops told me > “Maintaining data in both the CRM and the Excel tracker is time-consuming.” – Auditor > “Vendor TLs circulate the daily feedback report internally within their calling teams.” – Ops lead ## 02. Listening to all three sides **Rules first, screens second.** I asked auditors to walk me through their day, ran a session with callers alongside our PM, and put five decisions to the ops leads before drawing anything. Their answers became the product’s rules. ### The four rules - **Created automatically**: A feedback item appears the moment any parameter is marked No. - **Overdue after 2 working days**: So a team lead can’t sit on feedback. - **A closing note is required**: Resolving means saying what was shared. - **All five columns stay**: Call date, audit date, UMR, care manager, top errors. ## 03. The key decision **Resolve per call, or per caller?** A caller can have no feedback one day and three the next. Resolving call by call meant team leads going back and forth to the same person. Grouping by caller lets them share everything at once and close it together, so that’s what I chose. ### Both options, and the score brackets - **Option 1 · Per call**: Open each record, share it, resolve it. Simple, but a caller with three audits means three trips and no overview. - **Option 2 · Per caller · chosen**: See every open item for one caller, have one conversation, resolve them together with one note. - **Scores set the urgency**: Using brackets the ops team defined: 90%+ needs no action, below that the feedback must be shared, and below 60% can lead to a performance plan. ## 04. Iterations **Three rounds, then a working prototype.** The first list showed callers but hid the reason to act. I also explored working call by call in a side drawer, with a required closing note and a "Shared with caller" check. The final design kept the note and moved to one view per caller. Before the final layout, I tested the flow as a clickable prototype built in Lovable. - Image: First iteration: a plain list of callers with feedback counts, audit dates and scores. Round 1: a list, but no sense of what needs action first. Blurred under NDA. - Image: Second approach: a side drawer for one audited call with failed parameters, the auditor comment, a required closure note and a Shared with caller checkbox. Explored: one call at a time, in a drawer. Blurred under NDA. ## 05. The dashboard **One caller, one conversation, one close.** Team leads land on the callers who need action, with sortable average and lowest scores. Opening a caller shows every unresolved audit with its failed parameters and the auditor’s words, and one resolution note closes them together. Resolution history keeps the trail. - Video: A 21-second walkthrough built from the shipped screens, blurred under NDA. Chapters: 0:00 Callers to action; 0:02 Open a caller; 0:04 See what failed; 0:08 One note for all; 0:14 Resolve; 0:18 Back to the list. - View "Callers to action": Feedback Dashboard listing callers with unresolved feedback, their feedback count, latest audit date, average and lowest quality score. Unresolved callers first, with their average and lowest score. Blurred under NDA. - View "One caller": One caller view with three unresolved feedbacks showing quality score, failed parameter tags and auditor comments, a resolution notes box and a Review and Resolve button. Every open audit for one caller, failed parameters inline, one note to resolve. Blurred under NDA. - View "Resolution history": Resolution history showing a past resolution with date, who resolved it, scores, resolution notes and the feedbacks it covered. What was shared, when, and by whom. Blurred under NDA. ## 06. What’s next **From closing feedback to tracking improvement.** The ops team asked for the next step themselves: flag callers who repeat the same error, so coaching starts before it becomes a performance plan. Some of that can be detected automatically and some will stay a human call, which is the design problem I’d take on next. --- Next case study: [Try first, connect later](https://www.seemant.design/work/try-first-connect-later)