One-Line Summary
Starting from a simple sales feedback form, I redesigned the data flow across showroom visits, customer surveys, CRM, and the quoting tool. The goal: turn first-hand observations that used to live in verbal syncs into data that can be recorded, tracked, and shared across teams.
Background
The showroom has always been one of our most important places for understanding customers.
When customers visit, they often share things that surveys can’t capture: why they’re considering a smart home system, which competitors they’ve looked at, what concerns they have about cost or installation, what’s making them hesitate — and ultimately, why they did or didn’t move forward.
But most of that information used to stay in the conversation.
If marketing wanted to know what was happening on the frontline, someone had to chase down the sales team. If management wanted to understand where customers were getting stuck, someone had to ask the reps to recall and reconstruct. Over time, I started to feel like this was a real drain.
It wasn’t that the sales team wasn’t observing — they were. But their work rhythm doesn’t naturally include sitting down to write a detailed report. A lot of the valuable details only came out when someone specifically asked.
So what I wanted to do was simple: turn those questions into a format that was easier to answer — something sales could fill out after a visit using checkboxes and short notes, without it feeling like extra work.
This project didn’t start as a CRM project.
Originally I just wanted to build a sales feedback form — a quick way for reps to log a few things after each visit:
- What the customer’s main needs were
- Which competitors they’d looked at
- Where things were getting stuck
- Anything worth flagging for marketing or management
The operations team turned out to have a similar need. They wanted to understand what was actually happening on the floor — which questions customers cared most about, why deals weren’t closing, whether there were patterns worth addressing.
Since everyone needed the same information, it made more sense to build one place where it could accumulate and be analyzed, rather than having everyone ask separately.
My Role
This project started from my own observations of the showroom process — it wasn’t a formally assigned task.
My role included:
- Identifying where information was breaking down (between sales feedback, customer surveys, CRM, and the quoting tool)
- Redesigning the data flow logic — deciding what gets filled in by whom, where, and what should be pulled in automatically
- Designing the Sales Tool interface and feedback fields for the sales team
- Adjusting the customer satisfaction survey to reduce duplicate data entry
- Collaborating with AI (Gemini and Claude) to complete the core development
- Coordinating between sales and operations to find a design that served both sides
What I Found
Once I actually started digging in, I realized the problem was bigger than “add a form.”
We already had a customer-facing Google Form, a satisfaction survey, and a Google Sheet where sales sometimes added notes. The data had places to go — but those places weren’t connected.
1. Customers kept re-entering the same information
Customers who booked an appointment had already filled in their contact details. Then they’d have to fill them in again at the showroom, or again in the satisfaction survey — just so we could match it back to the right record.
2. Sales observations weren’t making it into CRM
Whatever a rep noticed during a visit didn’t have a natural path back to the CRM. Without a fixed place for that information, it tended to disappear into chat history.
3. Quote data lived in its own silo
The amounts and customer details generated by the quoting tool weren’t connected to the CRM. Matching them up required manual work.
4. A new form would just be a new silo
If I added a feedback form without fixing the underlying fragmentation, I’d just be creating another disconnected data source — making things more complicated, not less.
How I Approached It
I broke the problem down into a few concrete questions:
- Could we avoid asking customers to re-enter information they’d already provided when booking?
- Could sales reps capture their post-visit observations with less friction?
- In the satisfaction survey, which pieces of information could actually be pulled from existing data — so customers don’t have to fill them in again?
- Could the quote amounts feed back into the CRM?
- When the same customer left different information in different places, how should the system decide which to trust?
Once I had answers to those, I rebuilt the original feedback form into a Sales Tool — and used that as the foundation for redesigning the entire CRM data flow.
What I Built
1. Sales Tool
Sales reps can enter a customer’s phone number, and the system pulls their existing information from the CRM and Google Calendar — reducing duplicate entry. If the customer has a booking that day, the tool can also confirm check-in status.
After the visit, reps can quickly log customer needs, competitor experience, blockers, and additional observations using a structured form with checkboxes and short fields.
The design principle: let reps leave the most useful information with the least effort — not fill out a full report.
2. CRM Data Matching and Pre-fill
Contact details filled in at booking are automatically carried into the Sales Tool. This cuts down on steps for the sales team and makes it easier to keep CRM records consistent.
3. Customer Survey Adjustments
The original satisfaction survey asked customers to re-enter their contact details so we could match the response back to a CRM record. But customers had already filled that in when they booked.
The adjustment: let the CRM and booking data handle the identity matching, so customers only have to answer questions about their actual experience — satisfaction, impressions, things that genuinely need their input.
4. Quoting Tool Integration (in progress)
The PWA quoting tool I built earlier is currently being evaluated for CRM integration. Once quote data can feed back into the CRM, deal amounts, equipment lists, and customer records can all live in the same place — no more manual cross-referencing.
AI Collaboration
Most of the development was done through AI collaboration.
I used Gemini to build the main architecture, then brought in Claude for debugging, code review, CSS details, and errors that Gemini couldn’t resolve. Gemini was better for getting the structure up quickly; Claude was better for refining the details and making the interface look good.
But the part I found genuinely valuable wasn’t the code generation — it was being able to translate my understanding of the real workflow into specific development tasks.
Things like: which fields should the sales rep fill in, which should the customer fill in, which should be pulled automatically from CRM, and when different data sources conflict, which one should the system trust.
If you haven’t thought those things through first, AI will happily build you something that looks functional but won’t actually get used in the field.
Current Status
The main flow and interface are largely complete. I’m currently testing against a CRM copy — not connected to live data yet — to avoid affecting existing records.
Completed so far:
- Sales Tool: customer lookup, CRM matching, appointment check-in confirmation, post-visit feedback fields
- CRM data matching logic: uses phone number as the identifier to match bookings and existing customer records
- Google Calendar integration: confirms customer check-in status on the day of their appointment
- Sales feedback fields: needs checkboxes, competitor experience, blockers, and open-ended notes
- Survey adjustments: removed duplicate contact fields; retained experience and satisfaction questions
Next steps: complete full-flow testing, then evaluate whether to connect to the live CRM and how to integrate the quoting tool.
Information Shouldn’t Just Live in Conversations
What I like about this project is how it started as a single form and gradually became a data flow redesign.
A lot of internal problems look like they’re about missing tools. But often the real issue is that information isn’t being captured in the first place.
The showroom has always had a lot of important information flowing through it — it just used to stop at the conversation, the meeting, or someone’s mental note. Over time, that meant marketing had to keep asking what was happening on the floor, and management had to ask sales to reconstruct things they’d half-forgotten.
What I wanted to fix was simpler than it sounds: could we make it easier for that information to be recorded in the moment, so the people who came after could actually use it?
That’s why I enjoy building internal tools. They don’t have to look impressive. But when they’re designed right, they turn something that used to drain people’s time into something stable, trackable, and useful — something that accumulates into organizational knowledge rather than disappearing into the noise.