Finding that Spatial someone: using analytics to find love

“Love is a numbers game” has never been more true than today. In 2021, over 323 million people used dating apps - a figure which is projected to grow to 440 million by 2027. Today, this global market is worth $4475.28 million. 

Location is a key element driving dating apps. Whether you’re looking for something casual or more serious, it’s unlikely you’ll want to travel far to meet someone you’ve met online. But location in dating is about more than convenience. Your chances for finding love will be governed by the number of compatible singles in your areas, whether you define that by age, sexual orientation, interests or personal compatibility.

So in this Valentine’s Day special, we’re looking at how data can be used to find that spatial someone.

…We promise that’s the last time we’ll make that pun.

Location is key to calculating an app’s Total Addressable Market (TAM). At its most simple, this could be seen as the number of single adults in an area. In reality, this is far more complicated.

For instance, for heterosexual relationships you’d typically require a balanced number of men and women. While on a global scale, the percentage of the population who are single is roughly the same, this isn’t reflected locally - which is far more relevant to dating apps. 

Take this example of Texas. We’ve sourced data on marital status from the US Census Bureau at Census Tract level. As this data is large and complex, we’ve converted the data with our SQL enrichment tools to a H3 Spatial Index to allow us to quickly render and analyze this data.

The results? Pretty interesting!

Open in full screen here.

The purple areas are where there are more single women, and the green more single men - with taller peaks indicating a greater difference. Generally, there are fewer single men in large metropolitan areas like Dallas and Houston - although as you zoom in there are some localized discrepancies.

Dating app companies could use this information to focus and adapt their marketing strategies to attract more men in these imbalanced areas, or perhaps focus their efforts on more “balanced” locations.

Of course, compatibility is about much more than marital status and gender. One far less tangible element of compatibility is people’s interests. Does location play a role in that?

Geosegmentation takes data streams such as detailed demographics, consumer spend and internet behavior, and ranks an area against different “social indexes.” This allows data scientists to predict the behavior of residents from that area - whether they’re likely to buy a certain product, visit a certain store or enjoy a certain activity. You can learn more about geosegmentation in this recent webinar with CACI which delves into the cost of living crisis. 

So, if you can map out people’s expected behaviors and interests, you can see how likely someone is to find compatible singles in that area. With that information, a dating app company could advise users on how far to set their distance threshold, suggest relevant data locations… the list goes on!

Let’s check out an example.

Spatial.ai is one of CARTO’s data partners which specializes in geosegmentation. Their Proximity dataset - available for the USA, UK and Canada via our Spatial Data Catalog - ranks areas against 72 social indexes based on their activity on social networks. A score of 99 - for example - means they are more aligned with that index than 99% of the rest of the country.

One of these segments is “Happily Ever After:”

“Whether it's the stress of wedding planning or looking back on that special day, this segment has their eyes and hearts focused on lifelong commitment.”

Dating apps which specialize in helping users find long-term romance may wish to target marketing efforts in these areas, whilst organizations focusing on more casual users would probably find more success elsewhere.

Above: The spatial distribution of the "Happily Ever After" Index across California

A great way to quantify this is the Getis-Ord* tool from our Analytics Toolbox. This is a spatial hotspot tool which allows you to test variations in the statistical strength of spatial trends in your data. Hotspot tools are an incredibly useful tool in spatial analysis, so much so that we’ve written a dedicated guide to them here.   

The example SQL code for running this is below - you’ll need to convert your data to a Spatial Index first - find out how here. You could also plug this into a Custom SQL module in CARTO Workflows to embed it into a wider process. To try this out, make sure you grab yourself a free two-week trial!

   
  WITH
--run getis-ord*
getis_ord AS (
 SELECT `carto-un`.carto.GETIS_ORD_H3(
ARRAY_AGG(STRUCT(h3, EC02_happily_ev_8f6801ec_avg)),
     3, 'gaussian') AS output
 FROM `yourproject.yourdataset.yourh3segments`),


 --the output to this is an array, so it needs to be unnested
GIunnest as (SELECT unnested.INDEX AS h3, unnested.gi, unnested.p_value FROM getis_ord, UNNEST(getis_ord.output) AS unnested)


--rejoin the results to the source data to better understand the analysis
SELECT 
GIunnest.*, 
seg.EC02_happily_ev_8f6801ec_avg FROM GIunnest
LEFT JOIN `yourproject.yourdataset.yourh3segments` seg
ON GIunnest.h3 = seg.h3

   

Check out the documentation for this analysis for further details here - and we can see the results of this below!

Open this map fullscreen. 

Only areas with a P value < 0.05 and a positive GI* have been included, i.e. locations which we can confidently conclude are hotspots. Redder areas indicate areas of a higher population density; we can see that actually a lot of these hotspots are in sparsely populated suburban and rural areas, where the target addressable market is lower. 

So within this analysis, a highly populated area with a strong “Happily ever after” trait would be a great bet for an app targeting users looking for commitment. This intelligence could inform strategies for market expansion or geomarketing activities, capitalizing on the fact that they’re in a national hotspot for people looking for “the one!” 

It’s by bringing together both the Total Addressable Market and more qualitative resident characteristics that we’re able to derive really localized actionable insights like these.

With approximately 45 million active users globally, Bumble Inc.’s mission is to create a world where all relationships are healthy and equitable through their apps, Bumble and Badoo.

Is 1.5 million active users in the US a lot? That is Bumble’s key challenge; contextualizing their market share to identify areas for growth. To address this, they harnessed the power of Location Intelligence to understand their market penetration, contextualized within their Total Addressable Market (TAM).

One of the problems Bumble face’s is defining a geographical market area; generally city boundary polygons are based on administrative or statistical zones, and don’t reflect where people actually live. To deal with this, Bumble built a model based on DBSCAN clustering and Concave hulls that automatically captured the most densely populated areas. 

With the market footprints defined, Bumble enriched these with our Spatial Features demographic data from our Data Observatory in order to extract each city’s population disaggregated by age and gender. This provided a city-level Total Addressable Market size, which could be compared with their own first-party data to calculate market penetration.

Calculating these statistics enabled a lot of new projects and innovation within the company as they understood the population at more granular levels, allowing them to reduce their acquisition costs and focus their marketing budgets on:

• International expansion, by analyzing country-level TAM.
• Generation-based strategic marketing efforts.

Want to know more about how Bumble is leveraging Location Intelligence to drive growth? Check out their full story here. 

Still searching for “the one” spatial analysis platform for you? Sign up to our free two-week trial here!

Want to get started?

Sign up for a free 14-day trial

‍