Can you guess which contraceptive method Indians want to know about most?
askNivi is LIVE in š®š³India!
Here is what weāve learned so far...
Dear Friends,
After months of hard work by Niviās Global and India Teams, the askNivi service is LIVE in India!
The current version is available on Facebook Messenger in Hindi and English and provides Indian users the opportunity to submit a question and/or explore a menu of conversations about 8 different contraceptive methods. Weāre taking what we learn from these early interactions to a) train our AI, b) plan out and prioritize future content development, and c) learn about our users - who they are, what they want, and how best to engage them!
So what have we learned so far?
1ļøā£Men are curious about family planning methods too!
Knowing that men are often gatekeepers for contraceptive use in India, we designed the experience and ads to engage men as well as women. The result...73% of users are men (this tracks with general social media usage in India) and still 53% of questions are about contraceptive methods and pregnancy. Men are asking questions on behalf of themselves and their partners. Examples:
āI am going to get married and do not want to get pregnant for the next 5 yearsā
āMy wife suffers from irregular periods. Please suggest right directionā
āAre there any side-effects of tablets?ā
Interestingly, men even clicked on ads we had created to engage women (i.e. an ad that features two women chatting about pills) as well as those designed specifically for men. We are now experimenting with different ways to target ads to increase womenās engagement for the times we want to attract more female users.
2ļøā£Facebook is alive and well in UP, Rajasthan, and Bihar!
Meeting with public health implementers and experts, there was healthy skepticism about the ability to reach Niviās target audience with Facebook Messenger, which requires an internet enabled phone. We, too, assumed weād get most users from Tier 1 cities like Delhi and Mumbai, but the data shows otherwise! We ran FB ads across the Hindi-speaking belt and 65% of our users were in Bihar, Rajasthan, and Uttar Pradesh!
Figure 1: askNivi users by Location
Source: Facebook data
Looking ahead, weāre planning to launch Nivi on WhatsApp to further extend access, but we were pleasantly surprised by the reach of our FB Messenger engagement!
3ļøā£Menu selection tells a partial story
Nivi guides users to the information they need in two ways: 1) open conversation-based questions, and 2) menu-driven topic selection. In the conversational format, users can ask a question, Niviās AI classifies their intent, and directs each user to the most relevant content. In the menu-driven format, users search content through a limited list of topics. This latter approach is common among most IVR, USSD, and other mobile services and the analysis of users' menu selection is often used to ādetermineā what is important by NGOs and businesses.
Looking ahead to the flexibility of AI-enhanced conversations, we gathered 2,469 questions in just 2 weeks and are using these questions to train askNiviās AI in Indian English and Hindi. When we collect these questions, we see exactly WHY it is so important to actively engage users based on their questions rather than limiting choice to a fixed menuā¦
While menus show what users value within a predetermined set of topics, analysis fails to understand those topics in the larger context of a personās interests and priorities.
Niviās initial Indian menu offers 8 conversations on methods. Sixty-three percent (63%) of conversations accessed were about natural methods, which would lead one to believe that is the most important concern. While natural methods was a popular intent among questions, making up approximately 6.6% of user questions, we see that there is a much greater interest in "contraceptive method recommendation" (36%), no matter if it is natural or not. We also get greater insight into users' level of knowledge, with many users asking more general and elementary questions about āsexual health, having sexā (16.3%) (an intent we will be breaking down further in our next analysis) and see that āmenās healthā (9.6%) is of great interest and could be used as an inroad to introduce ideas about contraceptives by embedding information in a conversation that those users care about.
Figure 2: Frequency of Conversation Selection via Nivi Menu
% of total conversations accessed
Figure 3: Frequency of User Intent via Questions submitted
% of total questions asked classified with an intent
Weāre excited to deepen our knowledge of our users in India and look forward to working with partners to help you understand the people and markets you want to engage. Letās chat about how your organization might work with Nivi!
Drop us a line at ben@nivi.io and follow us on Medium for the latest insights.
- The Nivi Team