I Wish to Develop More Products that Customers Want with iKala's Technology

Human-centered is considering AI technology as a complement to human-beings instead of a substitute.

Dictated by Finn Yeh, Senior Product Manager, KOL Radar
Written by Iris Hung

The customer journey in influencer marketing includes developing the strategy, setting the goals, choosing the key opinion leader (KOL), signing the contract with the KOL, outlining the scope of our cooperation and content of our ads, and then finally evaluating the results. KOL Radar started by helping customers to "search" for KOLs. After a period of development and exploration, I began to think further ahead: how can we provide a more valuable service to our users? This led my train of thought back to the beginning. What are the goals and vision of our products? Who are we supposed to serve? By analyzing the user flow, we found that besides the search function, the customer's biggest pain point was evaluating the results of their campaigns. We needed a way to make the final report more visualized, more comprehensible. So, we decided to develop the "Deep Report".

In the past, customers used Excel spreadsheets to calculate the results. However, when you are working with many KOLs, you waste a lot of manpower updating the spreadsheets manually, and you cannot see the results in real-time. This limited any attempts at analytics to a very surface level. With Deep Report, the customer can accurately track the result of each influencer marketing campaign. By analyzing multidimensional data, we are able to solve the real problems encountered by our end customers. Not only are we able to free up manpower with the power of technology, we are also able to expand the scope of our customer's imagination. For instance, a word cloud can be used to dissect user-generated content (UGC), so we can understand what the audience cares about, what their thoughts are about our sponsored content, whether their reaction is positive or negative. Our customers can use these insights to further adjust and improve their marketing strategy.

After a round of user interviews and tests, we found that "activation" was the most difficult part. What could we do to convince our customers to change their current routine and try out Deep Report? It turns out everyone was afraid they would need to put more effort into compiling data, without even knowing what the end results would be. Therefore, we added an "import data" function to help customers import their existing spreadsheets directly into Deep Report. So long as the categories in the spreadsheet were not too different from Deep Report's template, the system was able to automatically match the corresponding data. The customer could quickly pour their data into the system and just sit back as the system automatically generated a new report.

Compared with my experience as a product manager (PM) in other companies, iKala has a much clearer division of responsibilities, allowing me to focus more of my attention on the product. Also, whether it's between departments or within the KOL Radar team, the resources and support systems are much better. When we are faced with a new problem, members of the team are eager to share their counsel, and there are many colleagues with different specialties to turn to for advice. I hope we can continue to explore the market in the future, so we can collect more useful feedback from our customers, which will help us develop more products that users want. That is my vision as a product manager.


Note: Deep Report uses technologies such as web crawling, text segmentation module, and sentiment analysis module. A web crawler is a highly efficient automated system that uses Proxy Server to analyze the API of different social media platforms and regularly capture the latest numbers from a sponsored post. The text segmentation module and the sentiment analysis module both rely on NLP technology. The text segmentation module is an improved Jieba system that converts text into meaningful phrases. The sentiment analysis module uses ML technology to study a lot of text messages and learn how to differentiate between positive, negative, and neutral tones.

Leave a Reply

Your email address will not be published. Required fields are marked *