One of the dimensions we looked at in our recent 1000 Programmable APIs study was the primary business areas of the APIs in they list – in other words the type of data or functionality being offered and how this is evolving. Programmable Web provides excellent fine grained classification data on the APIs it lists, however it can be a little hard to get an aggregate picture across the fine grained data – se we dug a little deeper by carrying out our own classification and grouping programmableweb’s tags for this data set.
The most surprising finds in this part of the study were that:
Contrary to what we expected, Infrastructure APIs and SAAS application APIs were very much dominant forces in this group of new APIs – beating out other categories such as Commerce, Media and Data by a significant margin – together making up more than 50% of all the APIs in the list.
Drilling down with the infrastructure area there are a few interesting micro-trends – in particular the emergence of Mobile Backend Service APIs such as those offered by Firebase and Hoppio and others.
The range of SAAS Applications being fitted with APIs is extremely broad – covering nearly 40 categories within ProgrammableWeb’s internal classification network.
As a whole, across the 1000 APIs there is a wide diversity in the types of APIs being launched, however these trends potentially also indicate where the most immediate business opportunities currently lie and which sectors of the Internet economy are rolling out API technology most quickly.
Surveying APIs by Business Area
For the purposes of the study, the APIs were divided into 6 broad classes as defined here. The classes were chosen partly as a grouping of Programmable Web’s own classification tags and on API types we commonly see at 3scale. The categories used were as follows:
Commerce: all APIs concerned with online or off-line including sale of goods (except digital media which was classified as “media”) as well as supporting services such as payment infrastructure and logistics.
Media: all APIs delivering content directly ready for end user consumption (including music, news, video and blogs) as well as media specific support services (such as media search and media metadata). This category however did not include “data sources” as defined in “data” below.
Data: all APIs providing static or dynamic data of some type – including everything from geographical, population and weather data to up-to-the second financial data and congressional spending records. To be included in this category, the API needed to provide an ongoing factual resource – rather than “editorialised” content which fell under the “media” heading.
Infrastructure: this category comprised all APIs providing low level technical functionality which is reusable across a wide range of different applications (was not sector specific). The list included messaging, identity management, search, monitoring, generic content analysis and other similar areas.
SAAS Applications: this category comprised all APIs related to a sector specific applications which emphasize data manipulation, content creation or some other functionality functionality above access to pur emedia or data access. The list included financial planning systems, document management, legal case file management, Marketing support, Dating services and many others.
Scientific: the scientific category covered those APIs which are directly connected to ongoing scientific investigation – including Protein Databases, Chemical compound mappings and many others. Generally these APIs provide core re-usable knowledge (or algorithms) which can be used as component parts by researchers. The category has some overlap with the “data” category, however – information which was deemed relatively well understood and “common knowledge” was classified as data, whereas what appeared to be active areas of research were classified as “scientific”.
Other: while most of the APIs fell relatively clearly into one of the previous categories, a small number (13 of 1000) did not find a home cleanly and were classified as “other” – these included, for example, a number of community efforts to define common interfaces which were not strictly being operated as accessible Web APIs, an API Driven API Explorer, Federal Procurement APIs and a number of others.
Although some APIs may arguably fall into more than one category, each was hard classified into just one. This approach may lead to some subjectivity in results since there are some potential grey areas between types – however, appart from those in the “other” category, on the whole during the classification most APIs found a relatively clear home.
The dominance of Infrastructure and SAAS APIs
The figure below shows the split between the 6 categories within the 1000 API’s surveyed, with Infrastructure and SAAS APIs taking over 50% of the list between them and with Data APIs coming third in the ranking with 18% of APIs.
The results which were surprising compared to what was expected included:
The very low number of APIs which fell into the “media” classification (4% – just 40 out of 1000): It seems likely that this is partly just a facet of the particular data set that the figure is so very low – since historically there have been greater numbers of Media APIs launching. However, the number does indicate that the common perception of Media APIs as one of fastest growing segments may not be correct. A significant number of the older APIs listed on ProgrammableWeb are Media APIs and they were amongst the first APIs to be released – hence there is certainly value in Media APIs. It will be interesting to see over time if the number of APIs in this category climbs again.
While not as low as the percentage of media APIs, the number of Commerce related APIs was also lower than expected. In the sample there were in fact very few true retail APIs (Gilt Group, Mob Commerce and Aisle Commerce fall into this category for example), and the trend in this category is towards “commerce supporting” services such as payment services). In this area it is likely that the technical complexity of running full purchase experiences via API may still be holding back growth.
Correlations between Business Area and Company Size
Analyzing the results further, the figures below show the distribution of company size versus business area. While estimating company size is not always accurate it seems to suggest a number of correlations:
Large and Very Large APIs in the study are primarily found proportionally in the Commerce, Data, Media and Infrastructure Segments.
SAAS App APIs in the sample in particular are primarily dominated by Very Small to Medium organisations.
Government APIs are primarily focused on Data provision and have little presence elsewhere.
(*) It should be noted that in general for Scientific API, the size estimate was taken to be the R&D unit or department publishing the results rather than the overall organisation.
Drilling down into Infrastructure APIs
Infrastructure APIs are a broad category. Drilling down revealed a number of areas of significant growth. The figure shows the different subcategories of infrastructure identified in the survey based on the ProgrammableWeb classification for that API. The figure removes categories with less than 5 APIs present in the 1000.
In the sample, there are five areas with significantly more APIs deployed than the others – Internet, Messaging, Security, Telephony and Tools. The APIs under infrastructure all share the property that they are widely re-usable across wide ranges of applications. In the “Internet” subgroup for example, the APIs cover areas such as Real Time Web Analytics (MapMyUser, Statcounter and others), Cloud Hosting Services (Scalarium and PiCloud for example) and integration platforms such as Boomi Atmosphere.
One area which stood out as a further trend was the emergence of a significant number of new Mobile Backend services. In the 1000 APIs sample, APIs addressing this included Firebase,Kinvey, Hoppio, Kii Cloud, Quickblox, – adding to platforms already out there such as Stackmob, Parse and Cloudmine.
Drilling down into SAAS App APIs
The following figure performs a similar drilldown on SAAS Application APIs. Also here, the diversity is impressive (again the graphs cuts out subcategories with less than 5 APIs in them – in this case around 40 APIs across diverse areas such as Medical Services and Advertising).
The strongest subcategories are enterprise (with APIs such as: Intuit Quickbooks, Desk.com), social (including:,Mixt, Chictopia and Coderwall) and Tools (APIs such as Sen.se or Onuma Architectural Modelling).
Conclusions
Infrastructure and SAAS Applications are both areas which gain clear benefit from API but and it’s interesting to see their surge in growth in the early part of this year. It’s unlikely that this means that there will be a drop of in APIs in other areas – just that growth will be un-evenly distributed. It will be interesting to see if the trends continue or reverse in the future.
In the next installment of the series we’ll dive deeper into the emerging business models behind the 1000 APIs and how they generate value both for their providers and the developers that use them.
Notes
Data refers to the 1000 APIs registered in Programmable Web between the dates of 17/1/2012 and 2/5/2012.
The chosen classification obviously influence outcomes of the study – it could be argued that the “SAAS API” group is too generic and that, subdivided, it would be weaker. However as can be seen in the drill down, there are significant ties between these types of APIs and membership of this groups is typically a good indicator of business model type and operational challenges.
3scale