Instagram is an idiosyncratic and fun way to share your life with friends through an array of pictures. The company was a startup launched in October 2010 that has grown so fast until the sensational Facebook’s $715 million acquisition 28 months later. It was only 13 employees by the time of acquisition, but it already had 100 million monthly active users. Most of the avid users use Instagram to snap a photo by their smart phone, then pick a filter to beautify the image and transform into memory to share their life. Some of them just love to showcase their traveling journey and adventure.
But do you know some of the users use Instagram daily as their online stores? This is unintended use of its service, and it is always a cool thing to notice that a product can do more than accomplish its intended goals. From company point of view, paying attention to that unconventional use is a way to get products to appeal to customers in a whole different way. Where are these online sellers found? What kinds of products are often sold?
Here is some example on online shop posts in Instagram. Some are sold in elegant way but many posts look so fancy.
Using the Instagram API, I wrote Python script that collected hashtags and crawled user data into text files for further analyses. In early February 2015, I ran the program to collect ‘online shop’ related hashtags to see which countries are most of the online shop posts from. According to Mashable article, similar to Twitter, small business is highly encouraged to boost their market reach by using hashtags, since this improves the chance of being found in targeted searches. Therefore, it is reasonable to start my findings from the hashtags, where I can find small businesses in Instagram.
Kevin Systrom, the CEO of Instagram, reported on the company’s blog that there are now more than 300 million Instagrammers that share more than 70 million photos and videos each day. The script returned the small subset of that: more than 32 million Instagram posts are associated with 50 different online shop hashtags. From the collected hashtags, I picked four represented cities of the country where top posts are coming and crawled them including user info, such as username, likes, comments and timestamp. The posts were cleaned using NLTK library by removing emoticons, unrelated links, numbers, stopwords and converting to lower cases to make the analyses process easier. In the end of the crawling, there are 127,708 posts collected from the four cities.
The most online sellers are in Indonesia
The first thing I set of to investigate was the online shop hashtags list. The word cloud below shows the collected online shop hashtags:
The above word clouds depict the hashtags with font size indicating number of posts. The bigger sizes tell us that particular hashtag included in more posts, such as “#onlineshop”, “#onlineshopping” and “#onlineshopindo”. From the above viz, we understand that people append the place’s name initial into the hashtags as well. For example, “#onlineshopjakarta” indicates Jakarta and “#onlineshopcebu” indicates Cebu, a place in Philippines. Thus, we can see which countries are they belong to from the hashtags, without looking at posts’ geo-tag location. Many posts are not found to include geo-tag location, since adding location to the posts is turned off by default in Instagram. Thus, finding the countries from geo-tag is not a good idea. Through the hashtags, I wanted to see where are the top posters coming from.
From the collected hashtags, we can group them by countries through the initials. However, it is hard to learn the country from general hashtags like “#onlineshop” and “#onlineshopping”, so we need to neglect them and normalize total post for each country by the remaining posts of 10 millions. The hierarchical bars below tell us that most online sellers are found in Indonesia, where 65% of the posts with online shop hashtags are found.
Fashion and beauty are the most common products in Indonesia
I then turned my analysis to the products that are sold in Instagram. Since Indonesia is the country where the most posts from, I am more interested in the products sold in Indonesia. The posts with four major cities (Jakarta, Bandung, Jogja, Bali) hashtags are collected that I hope to get representative posts from the country. In order to get fair amount proportion of the cities, the posts are limited to approximately 30,000 posts for each city. The four major cities hashtags are chosen due to their high penetration of social media users as well as highest ability to spend and propensity to consume. A report by McKinsey mentions the four cities are categorized as Tier 1, a category with the greatest ability to consume and the most willingness to do so. Moreover, according to Forbes article, Jakarta is the most active Twitter city in the world and Bandung is in the 6th position just ahead of Paris and Los Angeles. Each of the collected posts is then classified into 9 product categories by looking at the keywords contained in the posts (this includes the hashtags after removing the pound symbols, because the seller often put keywords on the hashtags as well). For example, posts that contain lipstick, shampoo, and eyeliner are counted as beauty product, while posts containing earrings, necklace, and eyeglasses will be counted as accessory. The counted posts for each category is then normalized by the total collected posts of 92,815. Below is the result:
As we can see from the viz, fashion category like clothing, footwear, bag, and accessory as well as health and beauty products are dominating the Instagram online shop world in Indonesia. These proved McKinsey report that mentions accessories and apparel as categories of goods in which Indonesian people spend time more on their consumption attitudes. Footwear and clothing that includes dress, t-shirt, skirt, etc. are the most popular product categories, just ahead of health and beauty products that contain skin care and makeups. Example on fashion product category can be found in the first part of this blog, but here is what health and beauty posts look like in Instagram.
Gift, food and home categories are not pretty popular, but they exist to enrich the product varieties in Instagram. According to my friend who is an avid seller in Instagram, food (e.g. cupcake) and gift (e.g. personalized popup card) are new business in Instagram and can be considered as premium products that are a bit costly, these are only purchased in niche market segment. Thus, unlike the fashion and beauty products, these categories have less number of posts. Here is example of gift and food products in Instagram.
The vizzies tell us that the online sellers in Instagram are found the most in Indonesia, with fashion and beauty products as the most popular product categories. However, there are several limitations exist in this findings. First is the Hashtag bias. There might be a lot of posts that use hashtags other than “#onlineshop” to sell products, such as “#olstore” or “#shop”. Posts might also sell stuffs without any hashtags that are not covered in this project. Moreover, the general hashtags that are removed might impact the post ratio on each country, or even add more countries to the account. The API call also could only return 50 hashtags without any pagination for other hashtags, which limit the findings.
Second thing is the Lack of product keywords. From the 120 thousands crawled posts, there are around 30 thousands posts did not match with any of the 9 product categories. If there are more related keywords inputted in the classification process, we can reduce this number to get better result.
Finally, it is also interesting to see the gender proportion on buyers/sellers in Instagram, but currently there is no way to get the gender from the API.
In conclusion, for those of you who want to start opening an online store in Instagram, consider to sell products other than fashion and beauty, to avoid competitors that have already been in this business for a while. Moreover, the products sold in Instagram seem to target middle-low class people, as can be seen by the product screenshots that do not look elegant and are less costly compare to similar products sold in department store or mall. So, are you interested to sell your products in Instagram?
This project was made by Giri Kuncoro as part of Computational Method for Data Science course assignment, a graduate course taught by professor Dan Cosley at Cornell. The script I wrote for this project can be found in my Github, that uses awesome visualization libraries such as D3.js and matplotlib. Data courtesy of Instagram, taken by using Instagram API. If you have any questions, please leave me a message on this page.