Content production 
- The average number of media uploaded to Commons by the reported GLAM implementation was 213. The number of media placed in Wikimedia articles per implementation was, on average, 50.
Upload counts were available for 46 GLAM implementations (100%). A total of 57,578 images were uploaded to Commons during these implementations. Of those images, 8,747 (15%) had already been used on 255,867 Wikimedia project pages by February 2015 when use was assessed. The number of media added per implementation ranged from 0 to 12,649. The median number of media added per implementation was 213.
The number of unique images used per implementation ranged from 0 to 2,453 and the average was 50. Proportion of media uploaded that was used on Wikimedia article pages ranged from 0 to 99% and the average was 8%. For the middle 50% of implementations, more than 2% but less than 43% of the media the implementation uploaded is used on Wikimedia article pages.
- We were unable to mine pageview counts of the content uploaded during the reported GLAM implementations, but the total content uploaded by 14 partnerships included in this report have been viewed over 5,100,000 times.
We could not obtain the number of pageviews of content contributed as part of the GLAM implementations in this report. However, 14 GLAM partnerships for which implementations are included in this report track the number of times all of the content they have uploaded has been viewed across all wikis. These 14 GLAM partnerships (32%) contributed content that has accumulated over 5,100,000,000 pageviews across the Wikimedia projects since the partnerships’ inceptions. The fewest pageviews were 12,000, to a partnership that has, to date, uploaded 30 media since they began in July 2014. Approximately 4,100,000,000 pageviews were to content from the National Archives and Records Administration of the United States, which has contributed 125,963 images since it began uploading content in July 2011. Another 760,000,000 pageviews were to content from Project Phoebus at the Museum of Toulouse, France, which has uploaded 3,307 media since it began in 2010.
- Of the 57,578 media uploaded to Commons, over 1,800 images were rated as Quality, Featured, or Valued Images.
By February 2015, across the 57,578 media uploads for implementations included in this report, there were 1,839 image ratings assigned:
- Six implementations (17%) had Quality Images. Number of rated photos per implementation ranged from 0 to 450; 
- Three implementations (9%) had Featured Pictures. Number of rated photos per implementation ranged from 0 to 89.
- Five (14%) had Valued Images. Number of rated photos per implementation ranged from 0 to 1,035 per implementation.
Implementations with more media rated as Quality, Featured, or Valued tended to also have higher use of media in articles.
Image Quality Ratings.
Inputs & content created
Cost per article created or improved
- Only two implementations reported data on both budget and media uploaded.
Of the two implementations that reported data on both budget and media uploaded, one implementation had 525 participants, spent $2,400.00 USD and uploaded 7,334 media. The other had one participant, spent $3,201.70 USD, and uploaded 84 media. For the first implementation, 685 unique media are already being used on Wikimedia article pages, while 76 are already being used from the second. Different styles of GLAM partnerships and implementations require different budgets and participants and the number of media uploaded vary widely. We need more data on GLAM budgets before we can explore the impact of budgets and number of participants in different contexts.
Cost per media uploaded
- For the two GLAM implementations that reported their budget and media uploads, the average cost per media uploaded was $19.22 USD.
Two GLAM implementations reported both budget and number of media uploaded. The cost per media uploaded ranged from $0.33 USD to $38.12 USD and the average cost was $19.22 USD. However, more data is necessary before we can understand the costs per media uploaded for the wide variety GLAM implementations.
Participation and Content
- Implementation that uploaded more media also tended to involve more participants, but implementations with more participants did not tend to have more media used on articles.
For the 22 implementations (48%) which reported both participants and media uploaded, there was a positive relationship between participants and media uploaded.  For the 22 implementations (48%) which reported both participants and unique media used on Wikimedia project pages, there was no relationship between participants and unique media used.
File:Participation and content bubble graph
- GLAM program leaders are generally experienced, proactive at producing blogs and other online resources related to their implementations, and are willing to help others start a similar program.
75% of implementations reporting were run by an experienced program leader who can help others in conducting a similar program. You can find which people or organizations ran these implementations in the data table.
For the 12 GLAM implementations in which program leaders reported on these replication strengths, nine (75%) were run by an experienced program leader who can help others in conducting a similar program. (You can find which people or organizations ran these implementations in the data tables). Each of those 12 programs generated shared knowledge in some way: most shared blogs or other online material while it was less common to generate guides or printed resources. The graph below presents the percentage of programs that produced each kind of shared resource.
File:SOME FILE NAME.png
Replication and learning. The bar graph illustrates the percentage of implementations that shared knowledge about how to run an implementation across four different knowledge resources. The majority of GLAM program leaders (75%) considered themselves experienced. It also shows that, of the forms of spreading information about implementation, blogs and online resources were the most commonly used (83%).
GLAM implementations and partnerships, while diverse in their program models, types of partnerships, and institutions, generate a treasure trove of knowledge on how to programs can be replicated or scaled. Further investigation is needed to see how these resources are shared among program leaders in order to learn from one another's experiences and practices.
- ↑ Note: Although "content production" is a direct product of the program implementation itself and technically a program output rather than outcome, most program leaders who participated in the logic modeling session felt this direct product was the target outcome for their programming. To honor this community perspective, we include it as an outcome along with quality improvement and retention of users as active editors.
- ↑ We use “GLAM implementations” here to mean time-bound implementations of partnership agreements. For example, if a partnership has been ongoing for two years, but had two agreements between August 2013 and January 2015 under which new media were uploaded to Commons, we count that here as two implementations.
- ↑ Mean=1,252; SD= 2,390.
- ↑ Mean=236; SD=504.
- ↑ Mean= 25%; SD=32%.
- ↑ Counts are inclusive such that an image that is rated both Quality and Featured is counted once in each category.
- ↑ Median=0; Mean=20; SD=81.
- ↑ Median=0; Mean=3; SD=15.
- ↑ Median=0; Mean=30; SD=175.
- ↑ Quality images to unique media used: Correlation coefficient=0.82; Featured pictures to unique media used: Correlation coefficient=0.82; Valued images to unique media used: Correlation coefficient=0.81
- ↑ Mean=$19.22 USD; SD=$26.72 USD.
- ↑ ’’Correlation coefficient”=.83.
- ↑ ’’Correlation coefficient”=-0.13.