top of page

A data collection tool that distributes polls through the chatbots of online shops.

Status  /  Pollar is the grantee of the Cyberport Creative Mircofund (CCMF).

Brief  /  CCMF sponsors high potential tech start-up projects and business ideas with a HK$100,000 grant. Participants use the grant over six months to produce proof of concepts and prototypes.

Role  /  Design founder, led the design process from start to end.

Team  /  Alan, Engineer founder; Michelle, Operation founder

Timeline  /  6 months

Tools  /  Invision, Illustrator, XD


The quality of online survey research is low. 

Survey panels could only reach a sample of "professional" respondents, which are typically in the lower income group, and they would falsify information just to take more paid survey.


Of researchers find it difficult to recruit reliable respondents

Source: GRIT Report 2015


Of market researchers considered data quality as the top challenge at work 

Source: GRIT Report 2015


Online survey market is expected to grow to $7b by 2020 at CAGR of 11.25%

Source: Knowledge Sourcing Intelligence LLP

How might we
improve the reliability and reach of online research through polls?
Product demo
Screenshot 2019-08-30 at 12.27.47 PM.png
Organic respondents

Through Pollar’s e-commerce partners, researchers could reach a much broader demographic who doesn't typically do online surveys.

Precise and reliable targeting

Through targeting specific online stores, researchers can segment respondents based on their purchase behaviour. 

Pollar research
Screenshot 2019-08-28 at 4.00.51 AM.png
Transparent recruitment practice

Pollar pays the online shops that distribute our polls. The fee and process is completely transparent,  so researchers could trust the quality of our responses 100%. 

Pollar's story


Our team designed Pollar with the philosophy of building a lean startup. 


We build first, then test it with users, and iterate upon feedback.

lean approach-07.png


Initial concept


Our first idea, named Adpoll, pays people to engage with online ads disguised as polls. 

The Adpolls would be distributed in news and media websites.


Concept testing

We found out that a few dollars is not rewarding enough for people to deviate from their browsing habits and start interacting with ads. 

However, we could not afford to pay the respondents more as well, it is simply not financially viable. 


Screenshot 2019-09-25 at 3.22.39 PM.png
Screenshot 2019-09-25 at 3.22.35 PM.png

Reframing our enquiry

This was the aha! moment for the team.

Since the premise of our idea is to exchange data for cash, we realized that our concept would be the most attractive at the point of check-out.

Screenshot 2019-09-27 at 2.23.00 PM.png
Screenshot 2019-09-27 at 2.21.28 PM.png

There is a gap in the incentive structure for data collection. 

Those who actually provide the data only gets a fraction to none of the value of it. Bridging that gap could improve both the quantity and quality of data collected. 

The pivot

User testing with online shoppers

How do we support growth in self awareness?


High fidelity prototype

Pollar version 1

4 action framework

Offer discount at checkout

We refined our idea according to the new finding.

Shoppers who complete polls at the checkout page of e-commerces would get instant dollar off their purchase.


The polls could be used for research or marketing. 

Screenshot 2019-08-20 at 12.49.04 PM.png

Early concept sketch

We tested our prototype with 20 online shoppers, and we received positive responses almost unanimously.


Our test focused on observing their reaction with the offer at checkout and the percentage of user who are willing to complete the polls for the discount. 


of users were able and willing to complete the poll without major pauses


of users rated the service as "desirable" or "very desirable"


of users said they would sign up for this service

The challenge


Defining the business model

Screenshot 2019-08-29 at 1.37.36 PM.png

Our current revenue model is not sustainable at a small scale, so e-commerce recruitment should be our immediate priority.

Overlooked stakeholder

We realised that the net worth of Pollar highly depends on our e-commerce network. However, our current model is not optimised to serve their interest. 

Basic poll


+ 5% service charge

Targeted poll


+ 5% service charge

Screenshot 2019-09-25 at 6.24.47 PM.png

Pricing model

Final iteration


Understanding the needs of online shop owners

Modifying the checkout page is a major concern to most online shop owners. 

I would give it a try if you do all the set up for me. And you must never make it a pop-up. I can't have it annoy my customers.

Screenshot 2019-08-29 at 2.36.36 PM.png

Well, if there's a way to provide the discount before the checkout I might consider it. 

Screenshot 2019-08-29 at 2.36.36 PM.png

Maybe talk to shops that are smaller?

The benefit seems too petty for my shop considering the extra time I need to spend.

I don't want to install things on my checkout page. Shopify doesn't allow changes on the checkout page too.

Screenshot 2019-08-29 at 2.36.36 PM.png
Screenshot 2019-08-29 at 2.36.36 PM.png

From a checkout page plug-in to chatbot

Screenshot 2019-09-26 at 12.42.07 PM.png
Screenshot 2019-09-26 at 12.40.57 PM.png

Less potential revenue for online shops

Pollar pays the shops based on the number of polls completed. 

If the poll is only visible to customers that purchase, they get less responses and less revenue.

Higher conversion for online shops

After respondents get a cash coupons upon poll completion, it is more likely that they will purchase at the store. 

Screenshot 2019-09-26 at 1.22.23 PM.png

Effortless experience

Minimal effort is needed to discover and complete the poll. The conversational design adds delight to the responding experience.

Screenshot 2019-09-26 at 1.23.26 PM.png

Integration to coupon code system

Pollar make it easy for online shops to track and see the number of polls or discount given from Pollar.


Win-win-win situation

At the end of the project, we successfully designed a data collection system which benefits all the parties involved. 

While the financial viability of the business still needs to be validated, our team is confident that Pollar would inspire a new wave of data collection tools that value quality and ethics. 

What I learned


User-centred does not equals to end-user centred.


I should identify the needs and relationship of all stakeholders involved, and see the systemic effect of applying pressure to one of the component. 

Research, research, research. 

As great as it is to iterate upon feedback, it would save us a lot of time if we talk to our stakeholders in earlier stages. If I were to do this again, I would invite at least one market researcher and online shop owner to every ideation meeting we had. 

bottom of page