Getting feedback from sellers

Strategic impact

• Identified a major blind spot in post-sale transactions where Motorway had little visibility
• Proved there was value in formalising post-sale seller feedback rather than leaving it in disconnected survey tools
• Demonstrated a low-cost way to test a new post-sale service before full product investment
• Helped shift Dealer Ratings from a deprioritised idea into a business capability with cross-functional adoption
• Allowed operational teams to intervene in cases of bad dealer behaviour
• Improved Trustpilot targeting and contributed to an increase to 4.62 score (approached 100k reviews)
• Created the evidence base for several 2026 product features, delivering across H1 2026

Context

How Motorway works

Motorway is an online marketplace for selling used cars. Private sellers profile their vehicle, it goes into a blind auction where car dealers bid. When the auction closes, the seller can accept or decline the top bid.
Once a sale is agreed, the dealer either collects the car themselves or uses Motorway Move — Motorway's optional collection service — to have it inspected and transported. Payment happens either by direct bank transfer, or through Motorway's payment product.

The problem

As of Q2 2025, c.89% of sellers on Motorway went through the direct collection path. In this journey there was no post-auction  product pages. Everything happened offline between the seller and dealer, giving Motorway very little insight into the completion of the sale.

Below was the last screen these sellers would see — a static page basically wishing them luck with the collection and dealing with the buying dealership.

Design for this page was done by another Designer c.2024 I've included it to illustrate the deadend we left customers at in the product.

While we had no insight into the collection step. Half of transactions that did complete via a direct payment method, we had no idea about the final price paid. This route was causing a lot of problems for Motorway as we had no presence in the key points of the transaction, but plenty of vocal sellers and dealers complaining about the process.

The remaining 11% of our sales went through our Motorway Move collection service, so we knew what was happening with them.

My role

Lead Service Designer, this case study is an example of working across various departments, pulling work previously done in silos, and delivering work that impacts roadmaps.

Identifying the various threads

In Q1 2025, our User Researcher had launched a Braze CRM email and Typeform survey. This fired two days after a seller accepted a bid. The questions asked if they'd either had their car collected already, had an upcoming collection, or had no contact at all yet. The business priority at the time was to catch dealers who had not contacted sellers yet (considered a big pain point at the time).

The CRM and Typeform had excellent engagement — 78% clickthrough rate and about 50,000 responses. Working with the CRM technical team I was able to export the data to analyse and identify what customers were telling us.

I mapped out the live questions and also the volume of users through the various paths.

My map of legacy typeform introduced by UX research and ops in March 2025, with analysis on volume through the various

Initial data analysis

I took 7 months of sales data and mapped the various seller experiences through the two potential collection methods, and the potential payment methods. I used our Motorway Move transport service and Motorway Pay payment service data.

This visualised the scale of the knowledge gaps, illustrated by the red boxes in the image below.

I mapped 7 months of the seller journeys. The red blocks are the primary areas I wanted to begin to collect data on.

My proposal was to update the Typeform or embed this in product, and iterate out some additional comms.

Mapping out the logic on new typeform questions and re-engagement

For sellers who had told us their car was already collected, we would ask what they were paid and then they would be asked to rate their dealer.

For sellers with upcoming collection dates (the largest cohort), we would use that date to trigger future CRM to them. In the event a Motorway Pay event fired, then we'd remove them from the follow-up Braze canvas. This way we could ask at the best time if their collection had happened. If it was paid (full or amended) or cancelled. Finally, we'd ask them to rate their dealer.

We launched the first CRM and Typeform changes  and started to collect our first dealer ratings.

The pitch to Product leadership

I wanted to ideally have the Typeform embedded in producty, or the questions productised into a series of post-sale pages.

The issue was, due to a re-org, there was no owner for that part of the product. I had discussions with an engineering team that had capacity to help. Through conversations with them and some high level estimates we formulated a basic first build.

Our pitch

• Sellers are engaging with the current forms and are very keen to give us this information
• We build a few lightweight post-sale pages, these would be foundational pages other teams could build on too
• We leverage existing CRM, point it to the product pages instead of Typeform
• When customers give us this info, we store it in our DB and use it, or our existing payment events to trigger follow-up comms via Braze to essentially close the loop
• If no event had fired on the day, then we'd ask if they sold, if yes what they got paid, and then to rate their dealer

Proposal on how we might ask sellers for this information.

Who could benefit from this work

• A dealer product team had shipped a basic dashboard and I thought this could be a useful thing to surface to dealers, hoping to change behaviour of bad actors
• We could also show sellers the ratings a dealer has to build confidence and trust in the transaction
• Our data and operational teams would have better insight into live transaction data
• Marketing would have better seller segmentation to customise downstream comms

We were unable to convince senior stakeholders of the value of this work within product, and the engineering team was put onto some tech debt tasks. A large re-org followed a week later and our UX Researcher, some of the product and CRM team were let go.

Product leadership decline the work

I'd learnt of a couple of other teams in the company looking at related work themes, and so started discussions with them about how we might work together. The CX team wanted a way to better segment sellers to protect our Trustpilot score. They wanted to only invite sellers who had had a positive experience to leave a Trustpilot review.

At the time they had no way of intelligently segmenting sellers, and the basic methods they had led to a lot of negative reviews being left from invited sellers.

My suggestion was to remove all the legacy segmentation logic, funnel as many sellers through our dealer ratings tool, and use that sentiment as the primary data point for further segmentation. There was some agreement on this approach, so we began work on implementing this.

I brought together a small project team — a CX Ops person, 2 technical CRM Engineers and myself.

Finding a way to build it anyway

Using the product work I'd done we mapped out an improved Typeform, and went live with basic changes to see the results.

I made the decision to leave the design of the emails unchanged. Though they looked pretty terrible, the conversion was excellent. We focused on building the CRM segmentation logic.

A week later we went live with re-targeting upcoming collections. Re-engagement CRMs saw slightly lower click-through rate.

Now we had data coming through. One of the team set up a Slack channel which updated when reviews came through; this was shared with the wider business, which helped promote the work internally.

Adding in re-engagement logic for future dated collections, and their injection intom the mid point of the survey questions.

The data was exported to Google Sheets from Typeform for analysis. I worked with the seller and dealer customer support teams to promote this work and help them understand how they could use this new information. Dealer teams used it for enforcing against bad actors (dealers intimidating sellers in person, unprofessional and sexist language and behaviours).

I'd created a reporting dashboard in DataStudio, after getting the help from Analytics to pipe the data to BigQuery.

To get even better visibility on the work, I got help from our Data Analytics team. Together we built out a Looker dashboard, and had the Analytics Engineering team pipe the data into BigQuery so that the data could be used more broadly by other teams. I designed the report to be high-level, with the ability to drill down to individual dealership level and link this data to a Trustpilot dataset.

Making the case to target the sellers getting paid via our MotorwayPay feature (CX needed alot of convincing)

We were able to make a link in the data strong enough to validate some assumptions around the segmentation of sellers who had had a positive experience.

We used existing events occurring on the platform to engage a new cohort of sellers. If a seller had been paid via Motorway Pay and had the Motorway Pay amended event fire, we'd invite them to the dealer rating page as we knew the outcome of their sale. Sellers who were paid the full amount were excluded from this and sent straight to our Braze Trustpilot invite canvas (this was the only legacy cohort that had high conversion to strong Trustpilot reviews).

Legacy logic for Trustpilot: only sellers who had an auction price above their initial estimate were invited. They were not all sent a Trustpilot invite directly; the goal was to reduce the number of people who had what the business considered a 'negative experience', and therefore reduce negative review volume.

Outcomes

Operations and CX

• By early December we had around 10,000 dealer ratings (for context, at the time we had ~90,000 Trustpilot reviews from over 7 years of operations)
• This started another discovery theme of work for myself and other PMs, looking into how we might create a better and more transparent vehicle collection and handover process for direct collections.
• I explored how the collection dates and dealer ratings survey could exist in the lead-up to this critical event.
• An Operations team wascreated to support this new handover feature, using the collection date info gathered in our survey to help forecast headcount.
• An experiment on the seller CRM was run, showing the sellers the dealer ratings (if they were above a certain threshold). This was expected to reduce cancellations. At the time of writing, this has remained inconclusive.

Product

• In Q1 2026, I led some discovery on a post sale initiative with a new PM, looking at how we improve the Direct Collection experience, before handing over to a dedicated designer who joined the team
• That same team went on to release the 'Handover' feature in Q2 2026, example design below (this was done by one of our Senior designers)
• Scoping out how we capture Collection dates in product as part of the Handover feature for Q3, and migrate away from Typeform
• In Q2 2026, a product team implemented showing sellers a winning dealers summarised survey results, to build trust in the transaction (experimeng ongoing as at time of writing)

Handshake app using the learning from dealer ratings. Final design by Harry in our team.

Dealer account page, showing them their ratings left by sellers. Final design by Josh in our team.