From First Edit to Lasting Contribution: Mapping Wikimedia Brasil’s Editor Community

For more than a decade, Wikimedia Brasil has organized edit-a-thons, photography campaigns, education programs, wikicontests and community events to bring new people into the Wikimedia movement and strengthen existing contributors. But what actually happens after someone joins an activity? Do they continue editing? Do they return for other events? Which activities create long-term engagement, and which mainly attract one-time participants? 

Data was collected from S.A.R.A. (which stands for Sistema de Avaliação de Resultados e Aprendizados in Portuguese, or Results and Learning Evaluation System), from the old Activities Control spreadsheet Wikimedia Brasil used from 2020 to 2022, and the reports available on Meta-Wiki since 2014. These sources were further cross-referenced and analysed with public Wikimedia data where applicable.

This report looks at these questions through the analysis of 12,797 editors registered in Wikimedia Brasil activities between 2014 and April 2026. By combining internal activity records with Wikimedia public data, the study maps participation patterns across five dimensions (breadth, tenure, experience, persistence and frequency) revealing how editors enter, engage with and sometimes leave the Wikimedia ecosystem. The results offer one of the most detailed portraits yet of Wikimedia participation in Brazil, while also raising broader questions about community growth, retention and the challenges of sustaining volunteer contribution online in the Wikimedia Movement.

Editors 16,243	Reports 2,173
Unique editors 12,797	Dimensions 5

Editorship Dimensions

Before proceeding, it is important to define the dimensions used in this analysis so one can better understand what is actually being measured.

• Persistence measures the continuity of an editor’s contributions following their first registered Wikimedia Brasil activity.
• Frequency measures the number of edits an editor makes in a given date.
• Breadth measures the number of distinct Wikimedia Brasil activities an editor registers for.
• Tenure measures the age of an editor’s account at the time of registration for an Wikimedia Brasil activity.
• Experience measures an editor’s wiki skills and understanding of editorial policies and practices, as reflected by their edit count up until a given date.

These five dimensions form the foundation of this analysis. Ideally, they are orthogonal to one another; however, depending on how each one of them is defined, some overlap may occur in practice.

Dimension	Metric	Definition
–	Editor	Any person that contributes with Wikimedia in the context of activities promoted or supported by Wikimedia Brasil
Persistence	Churned editor	Someone that has made at least one edit during or before their first Wikimedia Brasil activity, but has not made any edit after its end date
Persistence	Retained editor	Someone that registered for at least one Wikimedia Brasil activity and has made at least one edit after the end of their first Wikimedia Brasil activity
Frequency	Occasional editor	Someone that makes between 1 and 4 edits during a given period, usually a month
Frequency	Active editor	Someone that makes between 5 and 99 edits to content namespaces during a given period, usually a month1
Frequency	Very active editor	Someone that makes at least 100 edits to content namespaces during a given period, usually a month2
Frequency	Lapsed editor Inactive editor3	Someone that was an active or very active editor in the prior given period, but did not edit more than 4 edits in this given period, usually a month4
Frequency	Former editor Departed editor5	Someone that was an active or very active editor, but now does not wish to edit anymore
Breadth	Single-activity editor	Someone that registered for only one Wikimedia Brasil activity and didn’t participate in any other activity
Breadth	Multi-activity editor6 Returning editor	Someone that registered for more than one Wikimedia Brasil activity
Tenure	Newcomer editor	Someone that created their Wikimedia account up to 30 days before their first Wikimedia Brasil activity
Tenure	Established editor	Someone that created their Wikimedia account more than 30 days and up to one year before their first Wikimedia Brasil activity
Tenure	Long-term editor	Someone that created their Wikimedia account more than a year before their first Wikimedia Brasil activity
Experience	Editor 0 Registered non-editor	Someone that has created their account, registered in a Wikimedia Brasil activity, but have never completed an edit up to a given date
Experience	Editor 1-99 New editor Inexperienced editor	Someone that has created their account, registered in a Wikimedia Brasil activity and has completed at least one edit but fewer than 100 edits in total up to a given date
Experience	Editor 100-999 Experienced editor7	Someone that has created their account, registered in a Wikimedia Brasil activity and has completed between 100 and 999 edits in total up to a given date. It is assumed they have basic wiki skills and understand basic editorial policies and practices
Experience	Editor 1000+ Very experienced editors8	Someone that has created their account, registered in a Wikimedia Brasil activity and has completed at least 1,000 edits up to a given date. It is assumed they have good wiki skills and a good understanding of editorial policies and practices

Breadth

Breadth measures the number of distinct Wikimedia Brasil activities an editor registers for. An editor who participated in only one activity is classified as a single-activity editor, while one who participated in more than one is classified as a multi-activity editor (also known as a returning editor).

Some of the questions that guided this analysis are:

1. What is the proportion of editors that participated in just one activity versus the ones that participated in multiple activities?
2. Which activities attract the most single-activity editors?
3. Which activities attract the most multi-activity editors?
4. How did the breadth split evolve over time?
5. How does the returning rate vary by year?

Overview

Between 2014 and April 2026, Wikimedia Brasil registered 12,797 unique editors across its activities. Of these, 11,417 (89.2%) participated in only one activity, while 1,380 (10.8%) went on to participate in more than one activity. That means that approximately one in ten editors who engaged with Wikimedia Brasil returned for a second activity.

Evolution over time

The returning rate has not remained static across the analysis period. Early cohorts (between 2014 and 2019) showed returning rates ranging from 3.6% (2014 and 2015) to 16.8% (2016), with no consistent upward trend. The 2016 cohort stands out as an early outlier, though this may reflect the smaller number of editors then rather than a genuine surge in engagement.

The returning rate by year is based on each editor’s first Wikimedia Brasil activity year, so an editor who appeared in year X and later returned in Y is counted in the X cohort, not Y. Each editor is counted only once, in the year of their first activity. The chart below shows for each column the proportion of editors first engaged that year that registered for one or more activities afterwards. It also shows, through the curve, the number of editors first engaged in each year.

The chart shows that although 2020 –the first year of the COVID-19 global pandemic– had a significantly higher number of editors first engaged then (1,970), only 7.5% of them returned to register for more activities. The cohorts for 2023 through 2025 –the years of Wikimedia Brasil’s strategy cycle– have increased and stabilized in the return rate close to 15%. It is important to note that the later the cohort was engaged, the less time they had to return, so returning rates for them might be artificially deflated and reflect the situation at the time of this analysis – values might change in the future.

Activity breakdown

Not all the activities attract and engage the same profile of editors. Some might only be engaged by specific thematic activities, while others might only be engaged through in-person activities, for example, which could result in higher number of single-activity editors. By a significant margin, the activities with the largest absolute number of single-activity editors are Portuguese Wikipedia thematic Wikicontests, which ranges from 193 (Wikicontest Brazilian Women in Sport, in 2023) and 785 editors (Wikicontest New Ipiranga Museum, in 2020). These competitions are effective at bringing large numbers of new participants into the Wikimedia ecosystem, but the vast majority do not return. One hypothesis is that editors register because they are more interested in the theme of the wikicontest than to contribute to the Wikimedia ecosystem. Thematic wikicontests represent at least 30% of the single-activity editors.

At the other side of the breadth split, the activities with the highest absolute number of multiple-activity editors –those who return– are more diverse. The common links between those activities are their recurrent thematic and community-centric nature: yearly Wiki Loves Monuments campaigns (101 editors in 2024 and 96 in 2025), yearly Every Book Its Reader campaigns (93 editors) and structured community events (wiki encounters, larger and structured edit-a-thons) rank higher in the list with a range between 66 (Edit-a-thon Afropédia – Balaiada, in 2025) and 101 editors (Wiki Loves Monuments 2024). These activities are able to engage a significant volume of editors while also keeping a meaningful number of returning editors.

Wiki Loves Monuments (2024) and Wikicontests New Ipiranga Museum (2020) and Brazilian House (2021) appear in both lists. These campaigns had significant infrastructure and engaged several institutional partners behind their promotion, with events happening in several cities of the country for Wiki Loves Monuments, and the engagement power of one of the major history museums in the country –Museu do Ipiranga.

Tenure

Tenure measures the age of an editor’s Wikimedia account at the time of their first registered Wikimedia Brasil activity. Editors are classified into three groups: newcomers, whose accounts were created up to 30 days before their first activity; established editors, whose accounts were created between 31 days and one year before their first activity; and long-term editors, whose accounts were created more than one year before their first activity.

Some of the questions that guided this analysis are:

1. What is the proportion of newcomers, established, long-term editors at their first Wikimedia Brasil activity?
2. How has the tenure distribution evolved over time?
3. Which activities attract the most newcomers?
4. Which activities attract the most established editors?
5. Which activities attract the most long-term editors?
6. Has the percentage of newcomers grown over time?

Overview

Of the 12,797 editors analysed, 4,273 (33.4%) were newcomers at the time of their first Wikimedia Brasil activity, 6,528 (51%) were established editors, and 1,182 (9.2% were long term editors. A further 814 editors (6.4%) could not be classified, either because their account creation date was unavailable or because their information could not be matched to a Wikimedia account. The more prominent profile is therefore the established editor.

Evolution over time

The tenure distribution has shifted over the analysis period. Between 2014 and 2018, Wikimedia Brasil’s activities were dominated by established editors, with rates ranging from 93.7% in 2014 to 71.2% in 2018. Newcomers represented a small –but growing–  share in these early years, and still way higher than long-term editors, who have always been the less prominent group. In 2020, the number of each category reached a then all-time high, with 727 newcomers, 1,035 established and 181 long-term editors engaged in their first Wikimedia Brasil activity. That coincided with the transition to online activities during the peak of the lockdown of the COVID-19 pandemic. This transition significantly lowered participation barriers and broadened the organization’s geographic reach.

The following years, 2021 and 2022, the established editors group represented more than 70% of the editors engaged in their first Wikimedia activity, but later on, this trend shifted again, when newcomers became the largest single tenure group for the first time since 2019. In 2024, Wikimedia Brasil had 965 newcomers (56.5%) joining, out of 1,708 editors, and in 2025, the figure reached 1,002 out of 2,003 (50%). Long-term editors, while growing in absolute numbers –from 3 in 2014 to 238 in 2024–, have remained a consistently very small share of the total, never exceeding 14% in any given year.

Activity breakdown

The activities engaging the largest absolute numbers of newcomers are led by the Wikicontest New Ipiranga Museum (286 newcomers, or 35%) and the Wikidata Lab XLI: Wikidata IOLab, an event in 2024 that engaged participants of the International Linguistics Olympiad into knowing about the ontology of languages in Wikidata. (179 newcomers, or 96.2%). In terms of proportions, highlights include the editions of Wiki Loves Folklore in 2024 (82.6%) and in 2025 (84.9%) and the Arts+Feminisms in the Lusophone (84%). These activities function primarily as entry points into the Wikimedia ecosystem.

Established editors are mostly distributed into Wikicontests (Brazilian House in 2021, with 526 established editors, and New Ipiranga Museum in 2020, with 457) and Education programs, in particular the journalism-related activities at Cásper Líbero College; these show virtually no newcomers (90.4% established editors in the Introduction to Science Journalism of 2025 and 100% in the Education Program “Political Science (Journalism)” of 2016), mainly due to the delay between registering for these activities and the collection of data from them. Introduction to Science Journalism, for example, ran from January 1st until December 31 2025, so an editor that registered in February would be considered to have an 11 months old account.

Long-term editors are present in smaller absolute numbers across all activities. The activities that engage the highest number of long-term editors tend to be thematically specialized or community-oriented. Wiki Loves Monuments 2019 stands out with 30.1% long-term editors (it was the first year of Wikimedia Brasil managing the contest, and the outreach work was more endogenous). The other activities on the list are wikicontests, which suggests that activities with a strong community component or specific editorial focus on Wikipedia are more likely to engage long-term editors.

Cross referencing tenure and breadth dimensions reveals a non intuitive pattern. Long-term editors show the highest returning rate of any tenure group, with 24.9% going to participate in more than one Wikimedia Brasil activity. Newcomers follow at 14.2%, while established editors show the lowest returning rate, at just 6.5%. This pattern may reflect the motivational profile of each group. Newcomers who engage with Wikimedia Brasil early in their contribution path may be more likely to adopt it as a central part of their Wikimedia involvement, while established editors may participate out of curiosity or for a specific project without developing a lasting connection to the organization’s activities.

A complementary question is whether editors who do not return for a second Wikimedia Brasil activity continue to edit independently, that is, whether low breadth necessarily implies disengagement from Wikimedia as a whole. The persistence data⁹ offers a partial answer: Among long-term editors, 63,90% continued editing after their first Wikimedia Brasil activity, regardless of whether they returned for another activity or not; among newcomers, 31.78% did so, and among established editors, just 21.26% kept contributing to the Wikimedia ecosystem. This pattern reinforces the picture painted: established editors are the least likely to return for a second Wikimedia Brasil activity and the least likely to continue editing afterwards, suggesting that their engagement with the organization’s activities may be more transactional in nature, either to contribute to a particular topic, complete a task or fulfill an obligation (e.g. students in an university course, institutional edit-a-thons or a singular thematic competition). Long-term editors, by contrast, show both the highest breadth and the highest persistence rates, indicating a deeper and more sustained connection to the Wikimedia ecosystem.

Experience

Experience measures an editor’s wiki skills and understanding of editorial policies and practices, as reflected by their cumulative edit count up until a given date. Editors are classified into four tiers: Editor-0, who have never completed any edit; Editor 1-99, who have completed at least one but fewer than 100 edits in total; Editor 100-999, who have completed between 100 and 999 edits and are assumed to have basic wiki skills and understanding of editorial policies; and Editor 1000+, who have completed at least 1,000 edits and are assumed to have good wiki skills and a solid understanding of editorial policies and practices.

In this analysis all the edits of all editors in the dataset were gathered from the projects in Portuguese (Wikipedia, Wikibooks, Wikinews, Wikiquote, Wikisource, Wikiversity, Wikivoyage and Wiktionary) and the interlanguage projects (Wikimedia Commons, Wikidata, Wikispecies and Meta-Wiki). Some projects and activities have different recommendations to which specific domain the editors should edit on (e.g. a training on Wikipedia might be advisable for the editors to edit first on their sandbox page before transferring to the main domain). All the edits were considered, regardless of which domain on those projects they were made.

Some of the questions that guided this analysis are:

1. What is the distribution of editors with different levels of experience at their first Wikimedia Brasil activity?
2. How has the experience distribution evolved over time?

Overview

This analysis focuses mainly on three points in time: before the start of the editor’s first Wikimedia Brasil activity, at the end of that activity, and as of the date of April 30, 2026. This approach allows us to have a clearer picture of what was the experience of the editors before they engaged with Wikimedia Brasil, what they accumulated during their interaction with us, and how their experience evolved afterwards.

The experience profile of editors before their first Wikimedia Brasil activity began reflects a community of predominantly new and inexperienced contributors. Of the 12,797 editors analysed, 10,830 (84.6%) had never completed any edit at the time of the start of their first Wikimedia Brasil activity². A further 1,452 editors (11.3%) had between 1 and 99 total edits. Editors with between 100 and 999 edits accounted for 224 (1.8%), while 291 editors (2.3%) had already accumulated 1,000 or more edits before the beginning of the Wikimedia Brasil activity they registered for. That is, approximately 4% of editors arrived at their first Wikimedia Brasil activity with substantial editing experience.

By the end of that first activity, the picture had shifted. The Editor-0 cohort fell from 84.6% to 34.4%, as 6,424 editors made their first edits during the duration of the activity itself. The Editor 1-99 cohort grew from 1,452 to 7,523 editors (58.8%), becoming by far the largest tier. Editors with between 100 and 999 edits accounted for 560 (4.4%), while 308 editors (2.4%) had reached 1,000 or more edits by the activity’s end. Taken together, these figures indicate that the overwhelming majority of editors (93.2%) ended their first Wikimedia Brasil activity with fewer than 100 edits, reflecting the organization’s strong orientation towards engaging new and inexperienced contributors.

The migration matrix comparing the experience before the activity started with experience at its end reveals that the activity period was transformative primarily for Editor-0s. Of the 10,830 editors who arrived with no editing experience, 6,148 (56.77%) reached the Editor 1-99 tier by the activity’s end, 271 (2.5%) reached Editor 100-999, and 5 (0.04%) reached Editor 1000+. The remaining 4,406 (40.69%) made no edits at all during the activity.

Editors who already had some experience before the activity showed far less movement. Of the 1,452 editors in the Editor 1-99 tier at entry, 1,375 (94.7%) remained there, with only 75 (5.2%) progressing to Editor 100-999 and 2 reaching Editor 1000+. Among the 224 editors in the Editor 100-999 tier, 214 (95.5%) stayed and 10 (4.5%) crossed into Editor 1000+.

Comparing the experience distribution at first activity’s end with the current status as of April 30, 2026 reveals modest but meaningful progression across all tiers. The Editor-0 cohort fell from 34.4% to 30.6%, as an additional 561 editors made their first edits after the activity ended. The Editor 1-99 cohort grew slightly from 7,523 to 7,762, the Editor 100-999 group from 560 to 702, and the Editor 1000+ group from 308 to 419. These shifts, while modest in proportional terms, represent real editorial growth among a subset of Wikimedia Brasil’s editor community in the months and years following their first activity.

The overall picture reveals a concentration at the lower experience tiers with limited upward movement after the first activity. The large migration of Editor-0s into Editor 1-99 tier occurs primarily during the activity itself, suggesting that Wikimedia Brasil activities are effective at triggering first edits but less effective at sustaining editorial development over time.

Evolution over time

The experience distribution over time reflects the broader growth trajectory of Wikimedia Brasil’s editor community. In the early years of the analysis period (2014 to 2019), the total number of editors per year was relatively small, with Editor-0 and Editor 1-99 more prominent. The year of 2020 is a highlight because it reveals an inflection point: that year Wikimedia Brasil had 1,356 Editor-0s, the largest single-year cohort of editors with no prior editing experience, alongside 486 in the Editor 1-99 tier. This pattern reflects the mass mobilization of new participants during the COVID-19 pandemic, when online activities significantly lowered the barrier to first-time participation.

From 2023 onwards, another shift in the distribution is visible: while Editor-0 remains the largest group in absolute terms, the Editor 1-99 tier has grown substantially, and the Editor 100-999 and Editor 1000+ tiers have reached their highest absolute values, although small in comparison to the other tiers. In 2025, for the first time, the Editor 1-99 group (1,175) surpassed the Editor-0 group (1,004), suggesting that recent editors cohorts are arriving with slightly more prior editing experience than their predecessors.

Persistence

Persistence measures the continuity of an editor’s contributions following their first registered Wikimedia Brasil activity. Editors are classified into two groups: retained editors, who made at least one edit after the end date of their first Wikimedia Brasil activity; and churned editors, who made at least one edit during or before their first Wikimedia Brasil activity but have not made any edit after its end date. Editors who never made any edit (Editor-0s) are excluded from this dimension analysis.

Some of the questions that guided this analysis are:

1. Which activities have the highest number of retained editors?
2. What percentage of editors are retained vs churned overall?
3. How has the retention rate evolved by year?
4. Is there a critical window for retention? If an editor doesn’t edit within X days of their first activity, are they likely to churn? How quickly do retained editors return after their first activity?

Overview

Of the 12,797 editors registered in Wikimedia Brasil activities, 8,885 have at least one edit on record, and of those, 2,767 (31.1%) continued editing after their first Wikimedia Brasil activity and are therefore classified as retained, while 6,118 (68.9%) did not and are classified as churned. This classification is a snapshot, as those who are marked as churned may edit in the future and become retained. Roughly one in three editors who ever made an edit went on to edit again after their first activity.

Evolution over time

Retention rates have varied considerably across cohorts. Early cohorts (2014 and 2015) show the lowest rates, at 9.2% and 10.9% respectively. From 2016 onwards, retention rates increased steadily, reaching 26.2% (166 out of 634 editors) in 2017 and peaking at 46.9% (351 out of 749 editors) in 2023 – the highest retention rate recorded across the entire analysis period. The 2020 cohort, despite being the largest in absolute terms, shows a retention rate of only 31.7%, suggesting that the mass influx of new editors during the pandemic did not convert into engagement.

The most recent cohort (2025) shows a lower retention rate of 30.7% – lower than the 43.4% recorded in 2024. As with the breadth and tenure dimensions, this figure is artificially deflated: editors who first participated in 2025 have had at most one year to return to editing. Their true retention rates are likely to be higher and will only be accurately assessable in a future edition of this report.

Among the 2,767 editors who continued editing after their first activity, the median time to first post-activity edit was just 26 days – meaning half of all retained editors made their first post-activity edit within less than a month of their activity ending. The mean is considerably higher at 147 days, with a standard deviation of 316 days, reflecting a long tail of editors who returned much later.

Activity breakdown

The activities with the largest absolute number of retained editors reflect two distinct profiles. On one hand, the Wiki Loves Monuments Brasil 2024 and Wiki Loves Folklore 2024 stand out with 114 and 110 retained editors respectively. The Wiki Loves Monuments Brasil 2021 shows a perfect 100% retention rate across all 112 of its participants. These photographic contribution campaigns appear particularly effective at attracting editors who continue editing after the activity ends, likely because the act of uploading and documenting photographs builds habits and skills that transfer naturally to ongoing Wikimedia contribution, and due to the attractiveness of the outreach of the competitions. The Wikicontests Brazilian Women in Sport (68.0% retention) and the Every Book Its Reader campaign in 2024 (56.9%) also show strong retention rates, suggesting that thematically focused competitions with clear community relevance sustain engagement more effectively than broader initiatives.

The activities with the highest absolute numbers of churned editors are dominated by the major Wikicontests – the Wikicontest New Ipiranga Museum (341 churned, 20.5% retention), the Wikicontest Brazilian National Archives (217 churned, 26.9% retention), and the Wikicontest Brazilian House (212 churned, 28.4%). These are also among the largest activities, so the high churn counts reflect volume as much as disengagement – a large activity will produce large absolute numbers in both columns. More telling are the education program activities from the Cásper Líbero journalism school, which appear consistently in the high-churn rate list with retention rates between 13% and 20%. These activities draw on captive student audiences fulfilling course requirements, which may explain both the high participation numbers and the low likelihood of continued editing once the academic obligation is met.

Frequency

Frequency measures the number of edits an editor makes in a given month. Editors are classified into four categories: occasional editors, who make between 1 and 4 edits in a given month; active editors, who make between 5 and 99 edits; very active editors, who make at least 100 edits; and lapsed editors, who were previously active or very active but now make no more than 4 edits in a given month. This dimension does not measure the frequency of edits in Wikimedia Brasil activities.

Some of the questions that guided this analysis are:

1. What is the current frequency distribution of editors?
2. How has the frequency distribution evolved over time?

Overview

As of April 2026, of the 12,797 editors analysed, 122 (1.0%) are classified as very active, 191 (1.5%) as active, 4,372 (34.2%) as occasional, and 4,198 (32.8%) as lapsed. The remaining 3,914 editors (30.6%) are Editor-0s who never made any edit (they just created a username and registered to an activity) and are therefore outside the scope of the frequency dimension. Among editors with any edit history, occasional and lapsed editors are roughly equally distributed, together accounting for just over two thirds of the total editor population.

It is important to note that the lapsed classification is broader than might be expected. An editor who was active before or during their first Wikimedia Brasil activity is classified as lapsed if their most recent recorded month showed 4 or fewer edits following a month of active or very active contribution – regardless of whether they edited after their first activity at all. This means lapsed editors include both retained editors who tapered off over time and churned editors who were once active but never returned after their first activity.

Evolution over time

The monthly frequency data reveals several structural patterns that have remained consistent across the entire analysis period.

The active tier is the dominant frequency group in virtually every month from 2014 to 2025, typically accounting for two to four times as many editors as the occasional or lapsed categories in any given month. This reflects a community where those who engage at all tend to do so with meaningful regularity – editors who cross the 5-edit threshold in a month are not ephemeral editors but consistent contributors.

Cross-Dimensional Analysis

The cross-dimensional analysis examines how the five dimensions – breadth, tenure, experience, persistence, and frequency – relate to one another. Rather than treating each dimension in isolation, this section identifies patterns that emerge when dimensions are combined, offering a more nuanced picture of the editor community that Wikimedia Brasil has engaged over the analysis period.

Some of the questions that guided this analysis are:

1. Are long-term editors more likely to be multi-activity?
2. Do newcomers return for a second activity more or less than established editors?
3. Do long-term editors retain more than newcomers?
4. Which tenure class has the highest retention rate?
5. What is the frequency profile of newcomers, established and long-term editors?
6. How does the experience of an editor influence their retention?
7. What is the relation between editors’ experience and frequency of edits? Do very experienced editors tend to edit less frequently?
8. Does registering in more activities predict retention?
9. Are multi-activity editors more likely to be occasional, active, or very active editors?
10. Do returning editors edit more frequently than single-activity editors?

Tenure and persistence

The relationship between tenure and persistence reveals a striking and counterintuitive pattern. Long-term editors – those who had been active on Wikimedia for more than a year before their first Wikimedia Brasil activity – show by far the highest retention rate, with 63.9% continuing to edit after their first activity. Newcomers follow at 31.78%, while established editors show the lowest retention rate at just 21.26%. This finding echoes the breadth analysis, where established editors also showed the lowest returning rate to subsequent WMB activities (6.5%). Together, these results suggest that established editors engage with Wikimedia Brasil activities in a more transactional manner, participating for a specific purpose without developing a lasting connection to the organization’s activities or the broader editing community.

Tenure class	Churned	Retained	Total	Retention rate
Long-term	405	717	1.122	63,90%
Established	3.555	960	4.515	21,26%
Newcomer	2.162	1.007	3.169	31,78%

Tenure and frequency

The frequency profile of each tenure class reinforces this picture. Long-term editors are the most active frequency group by a considerable margin: 107 are currently classified as active and 105 as very active, giving them an active-or-very-active rate of 18.89% – compared to just 1.67% for newcomers and 0.99% for established editors. Long-term editors also show the lowest lapsed rate (36.27%) relative to the other tenure classes. Newcomers and established editors are overwhelmingly lapsed or occasional, reflecting their lower prior engagement with Wikimedia and their tendency to disengage after their first activity.

Tenure class	Very active	Active	Occasional	Lapsed	Total
Long-term	105	107	503	407	1.122
Established	9	36	2.252	2.218	4.515
Newcomer	6	47	1.565	1.549	3.167

Experience and persistence

Editors classified as Editor 1–99 at the end of their first activity show a retention rate of just 20.68% – meaning roughly four in five editors with minimal experience did not continue editing afterwards. This rate rises with experience: editors in the Editor 100–999 tier were retained at 65.0%, while editors with 1,000 or more edits show a retention rate of 96.1%. This near-perfect retention among the most experienced editors suggests that highly experienced Wikimedia contributors who engage with Wikimedia Brasil activities almost invariably continue editing, regardless of participation in future activities.

A notable case appears for Editor-0s, as 494 of them are marked as retained in the persistence analysis. This is a mathematical corollary of how the two dimensions interact: editor-0s are defined as having zero edits before their first activity ended, while persistence requires at least one edit ever. The only editor-0s who appear in the persistence data are therefore those who made their first edit after their first activity ended – and by definition, they are all retained. These editors represent a group engaged in editing by their Wikimedia Brasil activity, and their 100% retention rate reflects this selection effect rather than a true edit pattern.

Experience level	Churned	Retained	Total	Retention rate
Editor 0	0	494	494	100,00%
Editor 1-99	5.967	1.556	7.523	20,68%
Editor 100-999	196	364	560	65,00%
Editor 1000+	12	296	308	96,10%

Experience and frequency

The relationship between experience and frequency follows a clear gradient. Editor 1-99s are overwhelmingly lapsed or occasional (98.94% combined), with only 1.06% currently active or very active. The Editor 100-999 tier shows a more balanced distribution, with 11.1% active or very active. Among Editor 1000+ editors, the pattern reverses itself: 96 (31.2%) are very active and 65 (21.1%) are active, meaning more than half are currently editing at a high frequency. This confirms that editorial experience is a strong predictor of sustained engagement — the more edits an editor has by the end of their first activity, the more likely they are to remain highly active editors.

Experience level	Very active	Active	Occasional	Lapsed	Total
Editor 0	1	9	223	259	492
Editor 1-99	7	73	3.924	3.519	7.523
Editor 100-999	18	44	161	337	560
Editor 1000+	96	65	64	83	308

Breadth and persistence

The relationship between breadth and persistence is pretty strong. Multi-activity editors show a general retention rate of 77.02%, compared to just 22.98% for single-activity editors. More interestingly, the relationship shows direct correlation: editors who participated in two activities show a 65.2% retention rate, rising to 88.69% for three activities, 96.91% for four or five, and 100% for editors who participated in six or more activities. This pattern suggests that repeated engagement with Wikimedia Brasil activities can be either the cause or a consequence of sustained editing. Editors who return for more activities are more likely to keep editing, and editors who keep editing are more likely to return for more activities.

Breadth and frequency

Multi-activity editors also show a markedly different frequency profile from single-activity editors. Among multi-activity editors, 4.9% are currently very active and 5.9% are active – compared to 0.8% and 1.5% respectively among single-activity editors. The proportion of lapsed editors is similar across both groups (45.3% vs 47.6%), but multi-activity editors are more likely to be found in the higher frequency tiers, consistent with their higher retention rates and deeper engagement with the Wikimedia ecosystem.

Conclusion

After analysing 12,797 editors throughout more than a decade, some inferences are very clear. Wikimedia Brasil has been good according to the data at bringing new people to the Wikimedia ecosystem –especially via Wikicontests on Wikipedia and photographic campaigns on Wikimedia Commons. But the majority of those people do not keep engaging afterwards. Nearly 89% register for just one activity, and don’t come back.

What this analysis also reveals is that there are distinct profiles of editors, and each one answers differently to what Wikimedia Brasil has to offer. The long-term editor –those who edited more than a year before registering– is the most engaged: 64% keep editing afterwards, and one in four comes back for other activities. They are the core of the community. The newcomer –those who arrive without experience–, when they find what motivates them, tends to be kept engaged. The established editor –which includes those engaged in long-term activities– is the one that least returns. Their engagement is transactional: they participate or demonstrate interest in a specific theme, register for it, sometimes edit, and leave.

Taking a look into individual activities, a conclusion can be drawn: Wikicontests engage in volumes, but photographic campaigns retain. Big and structured wikicontests bring editors in the hundreds, but the majority do not come back or edit afterwards. They make sense as a strategy for outreach and visibility, especially when there’s an institutional partner behind them, but volume is not engagement. The activities highlighted in the analysis demonstrate what is a hunch: community-centered events, focused on the human connection and low-barrier practical activity and recurrent thematic, engage more.

On that note, one of the most action-centered findings on this analysis: The moment to retain editors is directly after the activity. More than half of all editors who will be retained make their first post-activity edit within a month. After 90 days, seven in ten have already made their decision. This means that the activity itself (the edit-a-thon, the wikicontest, the photography campaign) is only part of the job. What an affiliate does in the weeks that follow, whether that is a follow-up message, a community space to return to, a second low-barrier activity on the horizon, or simply a person who remembers someone’s name and reaches out, is likely to matter more for long-term community health than the design of the activity itself.

But what does that mean for Wikimedia Brasil?

Wikimedia Brasil has reached returning rates up to 15% in the three last years, stable and significantly higher than previous years. That is a real progress, but the analysis suggests two parallel paths to grow, but that would require different approaches.

For converting newcomers into engaged editors, the data suggests that Wikimedia Brasil should focus on sustained investment in community infrastructure and recurring activities –thematic wikicontests with institutional partners, national photographic campaigns and in-person edit-a-thons. These are spaces where they –the newcomers– start to create a sense of belonging to the community.

For the established editors, Wikimedia Brasil should rethink what is expected from this relationship. If the engagement is transactional by nature, the goal needs to shift from engaging this person into participating in more activities to make their experience editing good enough that they keep editing on their own afterwards; That is, the focus should be on the quality of experience in the editorial process.

What the Wikimedia Movement can learn from this analysis, and what it can not yet measure?

This analysis offered something that is rare in the Wikimedia Movement: a longitudinal study of real editors, with context, that crosses internal activity data with public editing data from Wikimedia’s projects infrastructure. The cross-dimensional analysis of the five dimensions –breadth, tenure, experience, persistence and frequency– generates a richer and more actionable picture than any isolated metric could.

But there is a fundamental limitation that any affiliate would face if trying to replicate this: activity data. Without a reliable system registering who participated in which activity, and when, it is impossible to cross the editorial behavior with the engagement context. Knowing that 785 people participated in the Wikicontest New Ipiranga Museum says nothing on its own. What says something is knowing that of those 785, only 20.5% continued editing afterwards – and that the median time to their first post-activity edit was less than a month. One number without the other is just a headcount.

Wikimedia Brasil was able to do this because it invested years into continually improving its activity reporting infrastructure. Even so, the older the activity and the report, the harder it became to recover reliable data – usernames not collected, full names instead of wiki usernames, activity without clear dates.. Without this information, what one gets is editing statistics without narrative or context.

For the Wikimedia Movement to be able to replicate this in a global scale, at minimum three things would need to be implemented:

1. Standardize the way affiliates and user groups register editors and participants. Today each group does this differently –a Google spreadsheet here, an Outreach Dashboard form there, a manual list on a Meta-wiki page somewhere else. For any cross-referencing with editing data to be possible, the minimum needed is a username and activity date. A simple, widely adopted format would already solve most of the problem.

2. Connect those registrations to Wikimedia’s existing data infrastructure. The replica databases exist, are public, and are available to anyone with Toolforge access, but the technical barrier to actually using them is considerable: configuring SSH tunnels, navigating the _p database suffixes, and knowing which projects are available and how to query them is not straightforward. A documented script, an API layer, or a managed service would dramatically lower the barrier for affiliates without their own technical infrastructure.

3. Build distributed technical capacity and common definitions. Even with data and access, someone needs to know what to do with them –and what, exactly, they are measuring. Wikimedia Brasil spent years building this pipeline, iterating through errors, and figuring out what each metric actually measures and how to adapt to local reality and capacity. That knowledge is not documented in any transferable form anywhere in the Movement. But the challenge is not only technical. One of the less visible difficulties this analysis faced was definitional: what counts as a retained editor? What defines an active editor? What is a newcomer? These are not obvious questions, and different affiliates, programs, and even Wikimedia Foundation teams answer them differently today. Without shared formal definitions, results from different contexts cannot be compared, aggregated, or learned from collectively. The Wikimedia Movement would benefit from investing in both: mentorship programs between affiliates, accessible documentation, and dedicated support from the Wikimedia Foundation for groups that want to do this kind of work – alongside a shared basic vocabulary for editor engagement that makes analyses like this one legible and comparable across regions and programs.

None of this produces results overnight. The value of this analysis comes precisely from its time depth — twelve years of consistent, recoverable data. The best time to start building that foundation is now.

Notes

Annex: Methodology

Data sources

This analysis draws on two primary data sources. The first is S.A.R.A. (Sistema de Avaliação de Resultados e Aprendizados), Wikimedia Brasil’s internal activity tracking system, which records the activities promoted or supported by the organization, including activity names, dates, editors, organizers, participants and many other qualitative and quantitative metrics registered for each activity. S.A.R.A. covers the period from 2023 onwards and was supplemented with legacy spreadsheets and reports for activities recorded before the system was adopted. Together, these sources form the definitive list of Wikimedia Brasil activities and the people associated with them.

The second data source is the Wikimedia replica database infrastructure, accessed via direct database connections. Edit data was extracted for all editors in the dataset across twelve Wikimedia projects: Portuguese Wikipedia (ptwiki), Portuguese Wiktionary (ptwiktionary), Portuguese Wikibooks (ptwikibooks), Portuguese Wikiquote (ptwikiquote), Portuguese Wikisource (ptwikisource), Portuguese Wikinews (ptwikinews), Portuguese Wikiversity (ptwikiversity), Portuguese Wikivoyage (ptwikivoyage), Wikidata (wikidatawiki), Wikimedia Commons (commonswiki), Meta-Wiki (metawiki), and Wikispecies (specieswiki). All namespaces were included in the edit counts, reflecting the full scope of each editor’s contribution across all content types on these projects.

Account creation dates and current user group information were retrieved from the Wikimedia Action API, queried against the Global account on Meta-Wiki.

Scope and freeze date

The analysis covers all editors registered in Wikimedia Brasil activities from 2014 to April 30, 2026. April 30, 2026 was established as the freeze date – the point at which all data was considered final for the purposes of this report. Edit data, experience classifications, and frequency calculations all reflect the state of each editor’s contributions up to and including this date.

The unit of analysis is the unique Wikimedia username. Editors who appear in S.A.R.A. under multiple usernames – for example, due to account renames – were matched to their current username where possible through the Wikimedia rename log. Usernames that could not be matched to a Wikimedia account are classified as unknown and excluded from dimensions that require account-level data, such as tenure and experience.

Dimension operationalization

Each of the five dimensions analysed in this report was operationalized as follows.

Breadth was measured as the total number of distinct Wikimedia Brasil activities an editor registered for, based on the dataset. Editors with one activity are classified as single-activity; those with two or more are classified as multi-activity. The returning rate reported for each year reflects the proportion of editors whose first activity occurred in that year who went on to participate in at least one additional activity.

Tenure was measured as the number of days between an editor’s Wikimedia account creation date – retrieved from the Action API – and the start date of their first registered Wikimedia Brasil activity. Editors whose accounts were created up to 30 days before their first activity are classified as newcomers; those with accounts between 31 days and one year old are classified as established; and those with accounts older than one year are classified as long-term.

Experience was measured as the editor’s cumulative edit count across all twelve projects at two points in time: before the minimum start date of their first Wikimedia Brasil activity, and before the minimum end date of their first activity. The difference between these two snapshots captures any editing done during the activity period itself. Monthly tracking of cumulative edits continues from the first activity end date to the freeze date, stopping when an editor reaches the Editor 1000+ tier. Editors with zero recorded edits at the time of their first activity end date are classified as Editor-0.

Persistence was measured as the presence or absence of at least one edit after the end date of an editor’s first Wikimedia Brasil activity. Editors with at least one post-activity edit are classified as retained; those with edits before or during their first activity but none afterwards are classified as churned. Editors with no edits at all are excluded from the persistence dimension.

Frequency was measured as the number of edits an editor made in each calendar month, from their first recorded edit to the freeze date. Editors are classified monthly as occasional (1–4 edits), active (5–99 edits), or very active (100+ edits). An editor is classified as lapsed in any month where they make 4 or fewer edits – including zero edits – provided they were classified as active or very active in the immediately preceding month. The frequency dimension covers all editors with any edit history across the twelve projects, regardless of whether they edited before, during, or after their first Wikimedia Brasil activity.

Limitations

Several limitations should be noted when interpreting the findings of this report.

Recent cohort deflation

Retention rates, returning rates, and other longitudinal metrics are artificially deflated for the most recent cohorts – particularly 2025 and 2026 – because editors from these years have had limited time to return for a second activity or to make post-activity edits. All metrics for these cohorts should be interpreted with caution and are expected to increase in future editions of this report. 2026 was cut from most or all the analysis, because of the truncation of the data.

Unknown usernames

A proportion of editors in the dataset could not be matched to a Wikimedia account, either because their username was recorded incorrectly, or because their full name was recorded instead of their Wikimedia username, or because their account no longer exists, or because they never created a Wikimedia account despite registering for an activity. These editors are excluded from dimensions that require account-level data and are reported as unknown where relevant.

Editor-0 ambiguity

The Editor-0 classification encompasses two distinct groups: editors who genuinely never made any edit across the twelve projects analysed, and editors whose edit data could not be retrieved or matched. The latter group, of approximately 340 editors, is treated as Editor-0 by default, though their true edit history is unknown.

Project scope

Edit data was extracted for twelve Wikimedia projects. Editors who contribute primarily to projects not included in this scope – such as other language editions of Wikipedia – have their edit counts underrepresented. This is most likely to affect the experience and frequency dimensions, where cumulative edit counts are used as proxies for editorial skill and engagement.

Namespace scope

All namespaces were included in edit counts, meaning contributions to talk pages, user pages, and project pages are counted alongside article edits. This may inflate edit counts for editors who are primarily active in non-content namespaces, such as administrators or community coordinators.

Technical pipeline

The analysis pipeline was built in Python, using pandas for data processing. Data was extracted from the Wikimedia replica databases via direct database connections, merged with Wikimedia Brasil activities data loaded from CSV exports, and processed through a series of dimension-specific scripts. The charts and documents were produced using Google Docs and Google Spreadsheets. Analysis of data and cleanup of functions were enhanced using Claude AI, and all code, numbers and interpretations were double checked and human-validated. All code and data will be available in the GitHub of Wikimedia Brasil.