We worked towards improving an existing version of Spotlight to address the evolving needs of users who were just starting out and learning the nuances of digital marketing along the way. In the process we also made it more scalable — ensuring that the system performs efficiently when demands increase.

Role

Product design, user research and testing

Result

Gave sellers more control over their campaigns, increased first time paying users and resulted in around 4.5x revenue uplift in the past 8 months.

What is Top Spotlight?

At the very beginning, Top Spotlight was supposed to pin the listing on the top of the search and browse feed. This meant that buyers browsing in search or category would see the relevant Spotlighted items in the top two slots first before anything else. This gave sellers a competitive edge over others.

The listing stayed at the top untill the bought clicks were delivered or up to maximum 7 days. If the bought clicks weren’t delivered, the remaining coins were refunded to the account which could be used for subsequent purchases.


Issues and insights through research

Before getting into the redesign and changing any feature, we recruited a few existing users of Spotlight, then called Top Spotlight — to know post launch feedback of the initial version. The feature was simple to use so certainly our users valued simplicity. However, we did hear feedbacks around a few recurring themes that included following things:

Users wanted more control over their spending

The current set up was simple but it also meant that sellers had no control over their spending. In order to run a campaign, they were required to pick one of the “packages” that Carousell offered.

Sellers wanted to be shown daily

Some sellers that we talked to wanted their ad (item?) to appear for a while for a certain number of days. Sellers were running a promotion for a certain number of days and sometimes the bought clicks were delivered on the first day.They were looking to spread out the bought clicks. In other words, they wanted to control the number of days their ads was shown.

The existing product offering wasn’t scalable

There were only limited top two slots in search and browse feeds. Once more demand started coming in for Spotlight, we were unable to deliver the clicks. As a result, for most of the ads we were refunding the money because we weren’t delivering the bought clicks. It was an unfavourable situation for both the business as well as the users who were putting their money, effort and time — hoping for the results.

Findings on the Insights Page

Issues on the Stats page

Once the spotlight campaign was running, there was no sense of progress on the stats page. Users had to decipher it themselves with the help of presented raw data whether the campaign was on the right track or something needed to be done. When the campaign ended, we didn’t communicate the cumulative return on investments.

Findings on the Stats Page

Another feedback we often got from sellers was the confusion between views and clicks. Out initial assumption was that since our sellers weren’t pro at understanding jargons like clicks and cpc we stick to widely users terms like views and impressions. But it created more confusion and sellers often assumed that they would be charged even if the same person viewed the listing twice.


Set up page redesign

Problem

We had identified the issues with the previous version of Spotlight where sellers didn’t have control in terms of spending. They also couldn’t set the duration of their Top Spotlight campaign. This was a missed opportunity because their bought clicks were exhausted within a few hours of running.

From the quantitative data, we also noticed that campaigns which ran for more than 3 days had most optimised cost per chat — better returns on investment for sellers.

Approach towards solution

We had some ideas around how to go about solving these pain points. One of the option was to let sellers set their own budget and duration. Setting the budget and duration would allow them to control how frequently they wanted their ad to appear in the marketplace.

First iterations on the Top Spotlight set up

While this was relatively obvious solution, the harder part was to change the offering of the Top Spotlight. As mentioned, there were only limited top two slots and for sellers it wouldn’t matter as long as they got the leads from interested buyers. Of course, there was a pool of sellers who bought Spotlight because they cared about the top two slots but we believed that it was just for vanity reasons. If our system was flexible enough to show the ads to the right buyers at right time — and at the right slots — it would be a win-win situation for all.

The first step was to drop the “Top” from “Top Spotlight.” It started from changing the explainer gif along with copy. We took help from our marketing team to communicate the changes effectively.

Explorations

The first exploration was inspired from the e-commerce checkout flows where buyers have to go through different steps before making a purchase. We tried to replicate that for the Spotlight because in future we were going to add different related features like keyword targeting and bidding into the existing flow.

E-commerce checkout experience

While this approach was exciting, we had some reservations regarding the multiple steps pattern. Since the new features weren’t coming any time soon the multi step felt like an overkill for something as simple as setting the budget and duration. From the past experience, the multiple steps had always proven to be inefficient for new buyers because every step witnessed a drop off.

Another issue with multiple steps was that in order to set one thing you need to go through all the steps which might be a waste of time if sellers want to simply set the budget and duration of their campaigns. Hence we settled with an option where we put every feature on a single screen making them mutually exclusive. When a seller wanted to just set the budget and duration, they can just do that and move on with their purchase.

Dropping the Top from Spotlight


Stats page redesign

Problems

Spotlight experience spans over multiple days and once a campaign is running, stats was going to be the most helpful screen because it has the update about the running campaign. Keeping the research findings into consideration, we came up with a job story to be done.

When sellers visit stats page of their campaign, they want to know how their campaign is doing, so that they can take appropriate actions to improve their campaign.

The other problem sellers were facing was the difficulty to understand that Spotlight attribution. There was no clear indicator to how much additional visibility boost Spotlight was bringing compared to organic means. We believed it was important because this realisation will dictate their future purchase decisions. Knowing what they got in return for their investment was going to make it easier for them to decide if Spotlight was worth it.

We had also found in the research that some sellers have their own idea about which is the right combination of budget and duration and they didn’t change it quite often. So we wanted to make it easy and quick for repeat users to run the campaign with the same settings again.

Approach towards solution

One of the obvious design principles we established for the stats page was that information architecture should easier to scan and comprehend. Some of our sellers were using the app in their offline shop while dealing with the customers at the same time. They didn’t have much time to understand everything. We wanted to help them be more efficient.

Some of our sellers were using the app in their offline shop while dealing with the customers at the same time.

We prioritised couple of data points that were extremely important for our sellers. First was the summary of the campaign: was it going well enough or could it be better or has it been stopped due to bad performance. The other was the progress of the coins spent for that day. Our assumption was that since sellers are setting their budget as — daily budget — they were expecting to see the daily progress too.

We also moved the total coins spent with total budget at one place for better scannability. To remove the confusion between impressions and views, we renamed “views” to “clicks.” We also started using “unique” clicks wherever applicable.

Improved Stats for better scannability

For the attribution we juxtaposed the organic clicks and impressions with the ones gained through paid promotions. To clarify further, we also added a small description to what Impressions or Clicks actually meant. For reference, users were now able to see the previous Spotlights as well. Having the shared components for the previous Spotlight as well helped us speed up the development process.

For instances when sellers’ campaign is stopped — because it’s not performing well — we started providing helpful cues that would be helpful in improving the quality of the listing.

Better attribution and helpful tips


Introducing bidding

About three months after launching the revamped Spotlight we gradually started seeing increasing number of running campaigns everyday. Suddenly in the next few weeks, it went way beyond our expectations and our system started seeing overwhelming number of campaigns.

However, the only problem was that our impressions were limited and could not be increased dynamically based on Spotlight demand. As a result, we were seeing about 40% of unfinished Spotlight campaigns which basically meant we were refunding sellers their coins because we just couldn’t deliver those clicks. This was a good problem to have.

One way to go about solving this problem is to let seller’s decide or bid how much were they willing to pay for each clicks. Sellers who opt in to pay for more are prioritised above everyone else. This allowed us to maximise on our limited supply while giving sellers the control they deserved.

Design goals

One of the design goals that we continued to follow from designing the core Spotlight was making it easy to understand the outcome of the feature. It was evident from the research that majority of the users were trying digital marketing for the first time and they weren’t exactly familiar with the idea of increasing the bid on Cost Per Click (CPC) — the usual way of bidding in Facebook and Google ad campaigns.

Keeping this design goal in mind, we started exploring the solution scope that would allow users to increase their cost per click but in a more intuitive ways.

Approach - 1

The idea behind the first approach was to convey that there were n number of running campaigns and that if you increase the cost per click — aka Prioritise your listing — your listing will be ranked higher in the stack of other running campaigns. This allowed them to increase their CPC but without actually having to do so.

Though a little abstract, while testing users were delighted to see their listing go up and down with the slider. Everyone preferred to move their listing to the top — highest bid.

Approach - 2

For the second approach, we kept the initial idea of prioritising your campaign but gave users control on increasing the cost per click. As they increased that cost per click, the interface would communicate the increase in priority their listing would get.

This approach was more explicit than the previous approach but as we had assumed, “increasing the CPC” didn’t ring much bell while user testing.

Approach - 3

For the third approach we took the best from the first approach and put everything upfront on the same screen. We assumed this was logical flow for the users — set how many clicks as a seller I want and then set how much I want to pay for those clicks. Of course, if I don’t move the slider, I will still end up paying the base charges for those clicks.

For setting the clicks, we made it a single picker instead of slider as previously because we were afraid having two sliders might be confusing for our users.

Post launch feedback

Once we launched an initial version of Spotlight Bidding, we ran a quick survey to collect the feedback from the existing users. One of the common feedback we got was that even after bidding there was no guarantee that the bought clicks would be delivered. One of the ways we are trying to solve this is by having an automatic bidding from our side. Since cost per million (CPM) changes based on demands coming in, how might we automatically change the cost per click of the running campaigns to deliver those clicks or at least utilise the budget in an optimised way.


Product Vision

For our sellers, leads are the metrics that matter the most. Their goal for buying either Spotlight or Bump is to get more leads so that they can sell more. We want to help our users to sell more. Going forward, we would be moving one step closer in the funnel — moving from clicks to chats to guaranteed sales.


Impact

Spotlight is one of the longest project that I have worked on (it’s still work in progress). With each new feature and optimisation we were able to add a new set of paying users. In the last twelve months, we increased the first time paying users by 30%. Our team was able to increase the Spotlight revenue by about 4.5x in the last eight months.


Learnings

Scrappy is good; sloppy is bad

At this point, I have to admit that the solutions and the explorations presented here aren’t the best and if given enough time, these can be refined further — especially the visuals. However, given the nature of startups, we are always under a tight deadline to fulfil business goals while solving user problems. We focused on making something scrappy that solves core users problems well enough. Sometimes that meant going ahead and testing an unfinished prototype with users to get their initial feedback — other teams were conducting the research and we utilised those appointments instead of using our time to schedule ones. At other times that meant engineering hacks to minimise the development efforts that was tactical, balanced with strategic thinking.

Scrappy prototype

However, we were always mindful about not making any sloppy choices along the way and it proved to be beneficial for the users and business. Even with the minor thoughtful optimisations we could see the needles moving. Working within the constraints made us realise scrappy is good; sloppy is bad.

⦿