Jonathan Yu - instagram challenge

#1

Kirsten’s current visitation status among countries of the world.

Dashboards:

https://public.tableau.com/profile/jons1691#!/vizhome/kirstenalana/Dashboard1

https://public.tableau.com/profile/jons1691#!/vizhome/kirstenalana/Dashboard2

SQL: selected everything

#2

I didn’t know I could only upload a single image at a time. Also, I now realize that this is literally the worst ordering of the images possible.

Just a few of the socio-economic and demographic map layers available in Tableau. These are at the level of zip codes, and it demonstrates how quickly you can get an idea of your current audience (especially easy to see if you turn the jitter up a little).

#3

I probably should have posted this image first as an introduction, then the visitation status map for comparison.

#4

The dashboards allow you to adjust amount of jitter and play chronologically.

https://public.tableau.com/profile/jons1691#!/vizhome/kirstenalana/Dashboard1

https://public.tableau.com/profile/jons1691#!/vizhome/kirstenalana/Dashboard2

Generally, it seems like tight clusters of users appear almost simultaneously. Even within countries, the clusters have a tendency to stay within their local subregions. I’d think that this would actually make it easier to identify which specific areas to target. The global coverage of users seen in the data reflects Kirsten’s current visitation status of countries around the world.

More broadly, the number of countries with similar socio-economic and demographic populations is limited. I would recommend identifying promising areas on a much finer scale (subregions/states/etc.), and also exploring options for more precise targeting of ads and hashtags (not absolutely sure, but I’m assuming you can get more precision with the hashtags).

#5

Not sure if this is the right place for random, non-data-backed speculations, but I thought I’d write them down here since I’m still exploring the data (and in case kirstenalana happens to read this). I (would like to) believe that human intuition still plays an important role in data analysis–if not in the analysis itself, then at least during discovery.

So I’ve looked over some of her posts that have received a proportionately high/low number of likes/comments, and these are my thoughts:

-posts regarding contests/giveaways were clear positive outliers; similar events could be explored as a strategy to reengage or attract more followers
-strongly suspect the actual subject/content of images influence likes/comments. besides some obvious stuff like exotic > ordinary (e.g. sunset at a tropical beach vs. a shot of breakfast), but i bet including the images themselves in analyses would probably reveal some interesting insights
-as a user, i feel that many of the more engaging posts are written in a particular style, almost as if the image was just taken and the post written on the spot. i know that’s not practical most of the time, but i’d try crafting the posts as if that were the case (e.g. describing emotions at the time, slightly higher energy, more conversational tone, etc.)

1 Like
#6

Another thought: I mentioned that it might be more effective to target smaller geographic areas as the number of large regions to target shrink.

The number of english-speaking areas left should be even smaller. For example, some attractive areas to target (just off the top of my head and based on the map) might include the Scandinavian countries, some North-East Asian cities, New Zealand, etc.–of those, only New Zealand is officially english-speaking.

[deleted because]: Instagram is already capable of auto-translations. But realistically, many, if not most, of the engaging rhetorical qualities found in a well-written post would likely be lost in auto-translation (source: Google Translate). Human translation should definitely be considered, especially for the major posts made in applicable non-english-speaking areas (which might not necessarily be the official language of the country).

I don’t know if it’s possible in Instagram to manually override auto-translations and replace them with customized human translation, but if not, I suppose it’d be easy to just add it to the post as long as it’s not too long. Assuming you can find translators/translating friends/friendly locals/etc.that you trust to accurately convey your “essence” or whatever you intend to express. Either way, the cost of a test run is virtually negligible, so there’s not much to lose by trying.

Here’s an example from the first paragraph in this post, https://www.instagram.com/p/Bq2OieYhEQR/?utm_source=ig_web_button_share_sheet :

I could hardly breathe without feeling pain it was so cold (as I remember it) but Central Park was still in the grips of a beautiful autumn blanket. Trees were the colors of a bag of Starburst if you only had cherry, orange and lemon left. Holiday windows were already causing lines of tourists to jam up Fifth Avenue. I had to buy a hat from one of the street vendors because my ears were so cold I thought they’d fall off if I didn’t cover my head. The scent of roasting chestnuts mixed with the odor of the horses, the taxi exhaust, the grit that just is New York’s smell.

Then translated into Hmong:

Kuv yuav ua tsis taus pa tsis hnov ​​mob heev li (raws li kuv nco nws), tiamsis Central Park tseem nyob rau hauv qhov sib txuas ntawm lub caij nplooj zeeg zoo nkauj. Cov ntoo yog cov xim ntawm lub hnab ntawm Starburst yog tias koj tsuas muaj lws suav, txiv kab ntxwv thiab txiv qaub laug. Lub caij so qhov khoom plig tau ua rau cov neeg tuaj ncig xyuas tebchaws tuaj yeem mus txog Fifth Avenue. Kuv yuav tsum tau yuav ib lub kaus mom los ntawm ib qho ntawm cov neeg muag khoom vim tias kuv pob ntseg kub heev kuv xav tias lawv yuav poob yog tias kuv tsis npog kuv taub hau. Cov ntxhiab ntawm cov kua txiv hmab txiv ntoo tov nrog cov ntxhiab ntawm cov nees, lub tsheb tav xis tsheb, cov pob zeb uas yog New York tsis hnov ​​tsw.

Which looks reasonable at first, but when translated back to English:

I can not breathe as much (as I remember it), but Central Park is still in the fall of the fall. The trees are the colors of the bag of Starburst if you only have tomato, orange and lemon. Holiday breaks for visitors can travel to Fifth Avenue. I had to buy a hat through one of the vendors because my ears were so hot I thought they would fall if I did not cover my head. The scent of juice mixed with the fragrance of the horse, the car, the stone from New York smells.

Yes, I know this doesn’t accurately reflect the same translation process, but I think the same limitations are clearly illustrated. I mean, I’d probably still like a post just based on the photography, but how does one engage with that? Commenting is technically possible, but at that point, it’s probably going to be a one-off comment rather than the beginning of any meaningful exchange.