I used AI to fix transcription errors, rerrarange the ideas, and suggest tweaks to the title and some sentences.
Three of the most exciting projects to come out of EA in recent years are, in a vague sense, CEA spinouts:
* Kairos is directly a spinout of CEA and now handles most support for university AI safety groups. Basically everyone I've found who knows them is really excited about what they do
* NEST is an opinionated ideas-fi...
TLDR; To help the effective animal advocacy movement cost-effectively absorb greater amounts of funding in the near future, we are seeking expressions of interest from people who could found a new organization focused on:
* Highly neglected animals: insects, wild animals, shrimp, fish, etc, or
* AI and animals: AI alignment and governance for animal welfare, strategic actions considering transformative AI, AI for wild animals, etc.
* ...
EA Giving Tuesday is a project of Rethink Charity. Rebecca Baron, Gina Stuessy, William Kiely, Angelina Li, Nix Goldowsky-Dill, Megan Jamer, and Avi Norowitz were on the organizer team in 2020, and many others also provided valuable contributions.
We abbreviate EA Giving Tuesday Donation Matching Initiative as “EA Giving Tuesday.”
“The match” refers to Facebook’s Giving Tuesday matching program.
We call the organizations we coordinate with and make fundraiser pages for “EA nonprofits” or “EA-aligned organizations.”
We call the people that prepared with our instructions, pledged via our form, and/or donated to these organizations on Giving Tuesday “EA donors,” or simply “donors.”
Summary
Since 2016, Facebook has offered a match on donations made through its payment platform on Giving Tuesday. This first-come, first-served match has represented an unusual opportunity for EA donors to get their donations to EA nonprofits counterfactually matched.
EA Giving Tuesday was started in 2017 to help these donors have the best chance of directing the greatest amount of Facebook’s matching funds to EA nonprofits. This match has been particularly valuable for donors in the United States, because of lower donation limits and tax considerations affecting donors in other countries. We feel it is unfortunate that despite effective altruism’s global reach, donors outside the U.S. face several restrictions that limit the Facebook match’s potential upside for them.
On November 11, 2020, Facebook released the Terms & Conditions for its $7 million matching program, to take place on Giving Tuesday 2020 (December 1). As in past years, the match would begin at 8:00 a.m. EST. Unlike in 2018 and 2019, only the first $2 million in donations were eligible for 100% matching in 2020. Thereafter, donated amounts would receive a 10% match, presumably until the remaining $5 million in matching funds were exhausted (details of this “10% match” were a bit unclear at the time).
We had 446 donors give through the Facebook fundraiser pages we’d set up for 111 EA-aligned organizations. We focused on preparing donors to make donations as fast as possible, and to submit donations during the optimal seconds. These recommendations were based on information we had collected and analyzed from the match in past years.
These donors gave $1.6 million to EA nonprofits during the match. These donated amounts received an estimated $411,693 (25%) in matching funds. The 100% match ran out in 2 seconds. Because the 100% match was extremely competitive in 2020, many donors who followed our instructions correctly were still only eligible for the 10% match. The 10% match lasted considerably longer, more than 3 hours.
We conducted follow-up work to ensure that donated and matched amounts were appropriately restricted or regranted (if applicable), and that EA nonprofits were receiving donated and matched amounts consistent with our estimates.
We conducted two follow-up surveys with donors. The first was sent 25 minutes after the match, and key findings included a reduction in reported payment problems (key takeaways from this survey are shared in the Giving Tuesday follow-up survey section below). Our second survey, sent on February 9, 2021 (after match results were announced), investigated when EA donors submitted each of their donations, and what the corresponding match outcome was. Key findings from this second survey were that only donations processed in seconds :00 or :01 of the match were eligible for 100% matching, and that of the 221 donations we have complete data for, 99% received either a 100% or a 10% match.
We faced several challenges in 2020, including the late announcement of the match (i.e., 20 days before Giving Tuesday), the reduction in the 100% match from $7 million to $2 million, and unpredictable and variable delays in donation processing times. We were able to mitigate these challenges and others to some degree through strategies built from experiences and data collected in past years. The outcome of these challenges was that EA Giving Tuesday saw a 25% overall match rate on donated amounts, compared to 52% in 2019.
EA Giving Tuesday had several successes in 2020: More donors participated, and record amounts were donated. Our instructions seemed to reduce payment problems, and effectively prepared donors for last-minute changes. We collected more data about which of our outreach efforts people found helpful. Donations to a record number of EA-aligned organizations and projects (111) were eligible for matching, made possible by regranting assistance from several organizations. (For example, the EA Funds donor lotteries were included this year.) Through collaboration with Rethink Charity and grant support recommended by the EA Infrastructure Fund, we were able to increase team bandwidth.
We estimate the resources that were contributed to EA Giving Tuesday in 2020, and consider factors relevant to estimating our impact.
We consider a few questions for 2021: What should we expect Facebook’s 2021 matching program to look like? And stemming from this, what level of organizer effort and promotion in the EA community would make sense, and what would our optimal donation submission recommendations be?
Background on Facebook's Giving Tuesday matching program
Giving Tuesday and Facebook’s matching program
Giving Tuesday is a “global day of giving.” Although it was started in the United States and falls on the Tuesday after American Thanksgiving each year, people and nonprofits around the world participate in Giving Tuesday by fundraising and donating.
Since 2016, Facebook has run a matching program on donations made through its payment platform on Giving Tuesday. Although donors from a number of countries can participate, only donations to U.S.-registered 501(c)(3) nonprofit organizations are eligible to receive Facebook’s matching funds. (Donors outside the U.S. face a number of other limitations during Facebook’s match, which we outline in 2020 considerations for donors based on geography).
Each year the match has been first-come, first-served, continuing until the matching funds are exhausted. Each year the match has started at 8:00 a.m. EST (5:00 a.m. PST). The amount of matching funds (USD) available has varied: in 2016 there was $900,000, in 2017 $2 million, and in 2018 and 2019 $7 million.
In recent years, donors and nonprofits did not find out which donations were matched until the following January, when Facebook would release this information.
The history of EA Giving Tuesday
EA Giving Tuesday was started in 2017 to organize EA donors around a shared goal of maximizing matching funds directed to EA nonprofits. To direct matching funds, a donor has to make a donation that is processed within the eligible time window. To help hundreds of people prepare to do this for effective nonprofits requires significant effort — hence, the EA Giving Tuesday initiative.
Although we try to clearly communicate to EA donors the large potential upside of participating in Facebook’s match, the primary purpose of EA Giving Tuesday is not to be a marketing program for effective altruism, effective giving, or for particular nonprofits. This initiative exists because of the opportunity for a donor to direct Facebook’s matching funds to nonprofits that would otherwise go elsewhere (see The opportunity for EA donors). If Facebook were to suspend the match or make particular changes to the rules, and if no other comparable matching opportunities were to take its place, then EA Giving Tuesday would likely be suspended.
Here is what donors have accomplished through EA Giving Tuesday’s preparation and coordination efforts:
Our attempt in 2017 was not very successful, with $48k matched out of $379k donated. However, Facebook did feature our efforts in their whitepaper on Giving Tuesday (search the whitepaper for "effective altruism").
In 2018, donors gave $719k, directing a counterfactual $469k (65% of donated amounts) in matching funds to EA nonprofits, despite the matching funds running out in 16 seconds. For more information, see our 2018 retrospective.
In 2019, donors gave $1.1 million, directing a counterfactual $563k (52% of donated amounts) in matching funds to EA nonprofits, despite the matching funds running out in 14 seconds. For more information, see our 2019 retrospective.
In 2020, donors gave $1.6 million, directing a counterfactual $411k (25% of donated amounts) in matching funds to EA nonprofits, despite the 100% matching funds running out in 2 seconds and despite 99% of reported donations being matched at least 10%.
2020 timeline for release of match details and rules
Throughout October we monitored various Facebook and Giving Tuesday social media channels to detect details of the match as soon as possible, to maximize the time available to evaluate the match rules, prepare strategy and instructions, and to promote the opportunity. Each year, it is unclear when Facebook will release the rules, and it did so 32 days before the match in 2017, 48 days before the match in 2018, and 32 days before the match in 2019.
On October 21, we discovered an informal mention of the 2020 match, 54 minutes into a September 29 video from Facebook’s Social Impact team. On November 11, 20 days before the match, Facebook published the match Terms & Conditions. On November 19, the Facebook Social Impact team shared a Help article that included the following information about the match:
How matching works: Matches are made on a first-come, first-served basis and will continue until the $7M USD fund has been met.
First: Matching 100% of the first $2M: For the first $2 million, we’ll match 100% of donations made to qualifying fundraisers. For example, if you donate $10, Facebook will also donate $10.
Second: Matching 10% after: For the remaining $5 million, we'll match 10% of donations made to qualifying fundraisers. For example, if you donate $10, Facebook will donate $1.
Understand match caps: We'll match donations to qualifying nonprofits up to:
$100,000 USD total donations matched per nonprofit
$20,000 USD in qualifying donations per donor
When matching will start: Matching begins on Tuesday, December 1, 2020 at 8am Eastern Time. We’ll continue matching qualifying donations to nonprofits until the $7M match runs out.
Who is eligible for matching: Any US-based 501(c)(3) nonprofit organization is eligible to receive donations on Facebook may be matched.
2020 considerations for donors based on geography
As in past years, the 2020 matching program was most worthwhile for U.S.-based donors, who were eligible to direct up to $20,000 in matching funds. Furthermore, U.S.-based donors were unlikely to forfeit their donation’s potential tax benefits by participating in the match.
Donors from several other countries were able to participate in the match, but were generally subject to a much lower donation limit, in the hundreds of USD equivalent. This lower donation limit reduced the potential upside for these donors to invest time in preparing for the match by practicing and reading our instructions. These non-U.S. donors would also generally be ineligible for tax benefits associated with their donation.
The opportunity for EA donors
Facebook’s annual donation matching program on Giving Tuesday continues to represent a rare opportunity to get donations to EA nonprofits counterfactually matched. This opportunity is different than most matching opportunities in the following respects:
The matching funds are available to nearly any U.S. nonprofit organization registered as a 501(c)(3). This permits donations to the majority of EA nonprofits, and reduces the risk that donors are incentivized to give to less-effective nonprofits because of the availability of donation matching. (For example, if Facebook only matched donations to a preselected number of nonprofits, we expect that some people might prefer to donate to the nonprofits where matching opportunities are available, regardless of the nonprofits’ effectiveness.)
In the Facebook match, matching funds that are not directed to EA nonprofits will otherwise go to organizations of more average effectiveness. This contrasts with most donation matching opportunities, where it’s likely that the matching partner would have made the same donation to the same nonprofits anyway.
We continue to see this donation-matching opportunity as one that avoids most of the pitfalls described by this GiveWell blog post on donation matching.
Our strategy to direct the matching funds
Our strategy in 2020 had six components, each described in a section below:
Preparing donors to submit donations quickly
Making new donation submission recommendations
Developing more robust day-of communication plans
Facilitating eligibility of EA-aligned organizations
Optimizing data collection for future improvements
Starting preparation early
Preparing donors to submit donations quickly
Why we focus on speed
Facebook’s matching funds have been exhausted in fewer seconds each year: in 86 seconds in 2017 and in 16 seconds in 2018; in 2019, the match ended in 14 seconds, but because of long processing delays, it was typically necessary to donate significantly earlier than 14 seconds to get matched. From 2018 onwards, our instructions have strongly emphasized the importance of submitting donations as quickly as possible, for the best chance of getting donations processed before the matching funds are exhausted.
Once Facebook indicated that the 100% match was for the first $2 million in matching funds, rather than for $7 million as in 2019 and 2018, it seemed probable that the 100% matching funds would be exhausted in very few seconds. Although a 10% counterfactual match would still be valuable, it was more worthwhile in expectation for donors to focus on the 100% match, which is where we focused our efforts.
Our computer environment document helped donors sidestep issues like mouse problems or browser issues. Our instructions gave recommendations on how to mitigate potential payment problems, and helped donors understand two aspects of Facebook’s payment platform that could slow or potentially even prevent their donations: The hard donation limit and the “Confirm Your Donation” dialogue, both of which presented last-minute challenges for our team in 2019.
The hard donation limit
The hard donation limit is why in 2020, we cautioned U.S.-based donors to never attempt to donate more than exactly $2,499 in a single donation window.
When it’s not Giving Tuesday, U.S. donors can sometimes submit up to $100,000 in a single donation window, provided they first complete a one-time “Confirm Your Identity” process that is triggered for amounts of $2,500 and above. In 2019, we had advised donors to simply complete this “Confirm Your Identity” process beforehand. We assumed that they’d then be able to submit a much larger donation (e.g., $9,999) in a single window. But just 30 minutes before the match, Facebook reduced the hard donation limit to $2,499—rendering our “Confirm Your Identity” advice irrelevant. With this change, donors planning to give more than $2,499 had to quickly prepare to submit in multiple donation windows, spread across multiple browser tabs, as fast as possible. This last-minute challenge likely partially explains why in 2019 our pledged donations were higher than what was ultimately donated.
Based on this experience and our monitoring, we anticipated that Facebook would again reduce the hard donation limit before the match. Our instructions to U.S. donors planning to donate any more than $2,499 (even a dollar more) were to prepare multiple donation windows. In 2020, 30 minutes before the match our testers detected that the hard limit per donation had been reduced from $100,000 to $2,500.
This hard donation limit made the 100% match especially challenging for large donors: for example, a person planning to donate $20,000 needs to maneuver between eight browser tabs!
Confirm Your Donation
The “Confirm Your Donation” dialogue presents different challenges than the hard donation limit. The former is triggered at a lower amount ($500), and, significantly, there is no way to reliably predict in advance whether this dialogue will be present or absent during the match. Furthermore, it could be added or removed right before the match. The best we can do is to closely monitor Facebook to detect the presence or absence of this dialogue.
When present, this dialogue (with a button to click) pops up after the normal button is clicked on the donation window. If donors plan for this second button and it’s not there, donors may accidentally submit donations too early to be eligible for matching. If donors don’t plan for the button and it is there, donation submission is greatly slowed. Given how quickly the 100% matching funds are exhausted, if we gave donors the wrong prediction, we’d effectively prevent their donations from being eligible for the 100% match.
In 2020, we intentionally prepared donors to be uncertain and to know what to do in either the presence or absence of the dialogue. See Developing more robust day-of communication plans below for more information.
Tools and practice opportunities for donors
We strongly encouraged EA donors to read the version of our instructions that reflected their location and their planned donation amount. We had 6 versions of instructions:
U.S., up to $499 USD
U.S., $500 to $2,499 USD
U.S., $2,500 to $20,000 USD
Canada, donating $5 to $499 USD
Canada, donating $500 USD or more
Other countries, donating €5 EUR to €250 EUR
This year we put more emphasis (in our instructions and outreach) on why and how to practice making quick and precise donations, as many donors would have to contend with not only the “Confirm Your Donation” dialogue (its presence or absence), but also the challenge of submitting donations across multiple tabs.
We recommended making $5 practice donations on Facebook and to use our donation simulator; the latter was new for EA Giving Tuesday 2020. With both methods, we encouraged donors to mimic match conditions while practicing.
Although the donation simulator did not supplant the benefits of making practice donations on Facebook, it had several important features: it required no money or activity on Facebook, it provided click-speed feedback, and, significantly, it could simulate the presence or absence of “Confirm Your Donation.” We wish to especially thank Angelina Li and Nix Goldowsky-Dill for their work on the donation simulator.
The recommendation to submit donations as quickly as possible, detailed above, was straightforward. But our team also analyzed the optimal timing of donation submissions (i.e., when to start donating). We had researched this question in previous years, but had not made recommendations. We wish to especially thank William Kiely and Avi Norowitz for this work.
The timing recommendations we made in 2020 were based on past years’ data and research, and on new analysis and on the particulars of the match rules (i.e., that the 100% match was for $2 million, not $7 million). Our recommendations were based on
synthesis of multiple years of data,
forecasting that was conducted before the match rules were released and updated after their release,
reasoning and evidence made available to EA donors (reach out to [email protected] to view the whitepaper), and
the amount a donor planned to donate.
Developing more robust day-of communication plans
Donors were instructed to not participate in the match until they had checked our homepage (https://www.eagivingtuesday.org) shortly beforehand for essential, anticipated updates and for any surprise, last-minute changes made by Facebook to its payment platform. We developed a day-of strategy so that we could effectively coordinate the monitoring and sharing of any new information donors would need.
We automated our planned communications as much as possible: We scheduled SMS and email campaigns in advance, designed multiple versions of an updated homepage, and practiced changing between live homepages versions. We built unscheduled SMS and email campaigns for surprise, unanticipated changes, and designed our planned day-of homepages with language that could be quickly updated in a variety of scenarios.
Our day-of testing and communications team was in close communication on match morning, with testing team members monitoring Facebook’s behavior with increasing frequency as the match start time approached.
Facilitating eligibility of EA-aligned organizations
Coordinating regranting arrangements
Only U.S.-based 501(c)(3) nonprofit organizations are eligible to receive Facebook matching funds, but many EA-aligned donation opportunities are not 501(c)(3)s. Each year we face the risk that these opportunities will be inaccessible to EA donors participating in the Facebook match. Two negative consequences could arise if this was unaddressed:
This could reduce the amount of EA donations eligible for Facebook matching.
This could influence EA donors to donate less effectively for the match.
We ensure that as many EA-aligned donation opportunities as possible are eligible for Facebook matching funds by coordinating restriction and/or regranting arrangements with several nonprofits. Particular thanks in 2020, as in past years, is due to several 501(c)(3) nonprofits including CEA, ACE, and GiveWell. (The Rethink Charity infrastructure was also used to facilitate regranting.)
Mitigating risk of hitting $100,000 per-organization match limits
Like 2019, in 2020 the match limit per organization was $100,000. The organizations we coordinate regranting with are at a higher risk of hitting this limit. This could have limited the amount of matching funds a nonprofit received, or it could have encouraged some donors to donate less effectively in order to avoid hitting this limit.
To mitigate this risk, in October and November we coordinated with many EA-aligned organizations to help them receive donated and matched amounts directly or through their fiscal sponsor, rather than through Rethink Charity, CEA, ACE, or GiveWell.
Optimizing data collection for future improvements
Data collected from donors, nonprofits, our team, and other contributors inform our strategy and recommendations. The data we collected for Giving Tuesday 2020 and in past years will play a significant role in our plans for 2021.
Donors supplied us with several datasets for Giving Tuesday 2020 through the forms and surveys we asked them to complete:
A community survey in October (132 respondents)
A pledge form in November (250 respondents)
A donation follow-up survey in December (189 respondents)
A match follow-up survey in February (62 respondents)
We also received the following data from the EA nonprofits we coordinated with, as appropriate:
The final donated and matched amounts they received, and on what timeframe
Any challenges they encountered with receiving funds
Feedback and ideas about EA Giving Tuesday promotion and target audience
Starting preparation early
We began early preparations in October. Early preparation allowed us to plan our strategy, give nonprofits advance notice and help them prepare, build the team, arrange to get donations and matches to difficult-to-reach organizations, and to prepare quality instructions and content.
We nonetheless faced a time crunch in late November: Certain key projects, like our recommendations and instructions, had been blocked until Facebook released the 2020 match Terms & Conditions on November 11 (i.e., 20 days before the match). We also learned that October was too late to start certain activities, such as for collecting and acting on community feedback.
Pledged amounts
By the evening before Giving Tuesday (November 30, 2020, at 8:30 p.m. EST), $880,764 had been pledged to EA nonprofits. By the morning of the match, pledges had increased to $1,491,558 from 250 pledgers, with a median pledge of $2,499.
Donated amounts
After our fundraisers ended on Wednesday, December 2, 2020, at 11:59 p.m. EST, we had $1,618,724 in donations, from 466 donors. Of the $1,618,724 donated, 81% ($1,309,138) of these donations were made within the first minute of the match.
(Note: The total donated amount excludes some donations to our fundraisers which we know were made without the intention of being matched.)
You can see a more complete breakdown of pledges and donations on our EA Giving Tuesday 2020 website.
Nonprofits signed up for Facebook Payments should have received donated amounts in early January 2021. Other nonprofits should have received donated amounts in late February 2021, from Network for Good in the form of a check or an electronic funds transfer.
Matched amounts
On January 29, 2021, Facebook reported the match amounts. The donations made through our fundraisers received in total $411,693 (25%) in matching funds.
You can find a more complete breakdown of donations and matches on our website.
Nonprofits should have received match amounts in late February 2021, from Network For Good in the form of a check or an electronic funds transfer.
Match end time
There were in effect two matches this year: the 100% match on the first $2 million donated, and the 10% match on the following $50 million donated.
The 100% match was extremely competitive, even more so than in past years. Based on our analysis, only donations processed within the first two seconds of the match start time (i.e., seconds :00 and :01) were eligible for 100% matching.
Due to unpredictable and variable processing delays, some donors had zero donations matched at 100% even if they closely followed our instructions.
Based on the available data, we believe that donations processed at :02 and later were eligible for the 10% match, which we think ended within the time period 11:10 a.m. EST to 11:23 a.m. EST, over three hours after the official match start time.
Follow-up with nonprofits
From December 2020 to May 2021, we conducted follow-up work with EA nonprofits. We wish to especially thank Gina Stuessy and Megan Jamer for this work.
Because we administer each EA Giving Tuesday Facebook fundraiser page, we have access to certain data that the nonprofits do not. The first part of our follow-up work was to save, organize, and analyze all fundraiser data and distribute as appropriate to the relevant nonprofits in private spreadsheets.
Whenever possible we confirm the donated and matched amounts received by EA nonprofits, to ensure that our data and estimates are accurate, and to investigate any discrepancies (i.e., nonprofits receiving less than our fundraiser data suggests they should). With few exceptions, amounts reported by nonprofits have been consistent with our expectations. For more details, see the EA Giving Tuesday 2020 website.
With all restriction or regranting arrangements, we confirm to the best of our ability that these arrangements have been carried out, and we try to make this work as easy as possible for all stakeholders involved (i.e., the receiving and regranting organizations).
There were some instances where nonprofits were delayed in receiving payouts from Network for Good. In these cases, we worked with all stakeholders to resolve these issues.
Follow-up with donors
Giving Tuesday follow-up survey
At 8:25 a.m. EST on Giving Tuesday (December 1), we emailed donors asking them to complete a survey about their donations.
The survey captured responses from 189 people, out of 250 people that pledged a donation and 466 people that donated through the fundraiser pages we set up. The donated amounts represented by the survey respondents account for 71% of all donated amounts in EA Giving Tuesday ($1.1 million out of $1.6 million).
We performed an analysis on the survey results, led by Gina Stuessy. Out of 189 responses, 187 were sufficiently complete to be fully analyzed. The following are key takeaways from the survey analysis:
The survey respondents reported successfully donating 94% of the amounts they planned to donate. This is a notable improvement from 2019, when our survey analysis revealed that only 73% of respondents’ planned donation amounts could be submitted, mostly due to last-minute changes to payment limits or other problems.
About 60% of all donors spent 30 minutes or less preparing. Of people who planned to donate more than $2,500, 55% spent at least an hour preparing; among those who planned to donate more than $10,000, 46% spent an hour and 28% spent 2 hours or longer.
Over 95% of survey respondents reported reading our instructions. Most respondents planning to donate over $2,500 also took additional preparation steps (i.e., practicing on the donation simulator, Facebook, or both).
80% of donated amounts recorded in the survey were likely submitted in the timeframe we had recommended. This is evidence that most donors were successful at following our advice to donate quickly.
The 18% of survey respondents that were outside the U.S. donated only 2% of the total amount captured in the survey. Unfortunately, donors outside the U.S. were generally not eligible for tax benefits and generally had far lower donation amount limits.
69% of survey respondents (representing 74% of donated amounts recorded in the survey) reported donating to the same organization that they would have donated to otherwise. Among the donors who reported donating to a different nonprofit, 2% (representing 1% of donated amounts recorded in the survey) reported donating to a nonprofit they thought was less effective.
5% of survey respondents (representing 6% of donated amounts recorded in the survey) reported losing the ability to donate appreciated shares. There were few other reports of lost tax benefits because of the matching opportunity.
44% of survey respondents (as weighted by donated amounts recorded in the survey) reported that without our work, they would have “definitely” or “probably” donated on Facebook with the intention to get matched. However, these respondents expected to submit only a modest amount of donations (~$125k) within the timeframe we used as a proxy for eligibility of 100% matching. (Put another way, most donations that these respondents expected to submit without our work would not have been eligible for 100% matching.)
The most common ways survey respondents heard about or were convinced to participate in EA Giving Tuesday 2020 were emails to our mailing list, Facebook posts, and the EA Newsletter. Medium and large donors were more likely than smaller donors to hear about the initiative from a friend and/or attend our overview and Q&A event.
For more details, see our full survey analysis here.
Follow-up with donors on match results
Facebook began reporting match amounts on January 29, 2021. We emailed donors to inform them of EA Giving Tuesday total amounts matched, and how to check whether their own donations were matched.
In our first follow-up survey, detailed above, we had included a request to estimate when respondents had clicked the final donate button to submit each of their donations.
On February 9, 2021, we sent a second follow-up survey to those who had responded to the first survey (thanks especially to Avi Norowitz). Respondents were asked to report the receipt time of each donation, and whether the donation was matched at 100%, at 10%, or not matched. We had 62 unique donors who completed both surveys.
Our analysis aggregated the responses from both the December 1 and February 9 surveys to help us understand Facebook’s donation processing delays and when donors needed to submit their donations to get matched at 100%. The survey findings strengthened our confidence that only donations processed in seconds :00 and :01 of the match were eligible for 100% matching. The results will help inform our recommendations for 2021.
Challenges
In this section, we review some of the unexpected challenges we faced.
Late announcement of matching program
Each year we don't know when (or if) Facebook will announce a matching program or its details. Facebook announced its 2020 Giving Tuesday matching program only 20 days in advance, giving us less time than the 32 days given in 2019, and far less than the 48 days given in 2018.
Our general approach has been to start planning and doing outreach for the matching program a few months in advance, working on the assumption that the details will be roughly similar to the details of the previous year. Although we risk wasting time and building publicity for a matching program that doesn't materialize, we believe the benefits of advance planning make this worth it.
In 2020, the match details were significantly different from those of previous years. Our team had to quickly analyze the new aspects of the match and to revise our recommendations, instructions, promotion, and strategy in light of these changes.
Reduction in matching funds for 100% match
In 2020 the 100% match was for the first $2 million donated, whereas in 2018 and 2019 the 100% match was on the first $7 million donated. This change affected the value of EA Giving Tuesday overall, as the decrease made the 100% match significantly more competitive.
This affected how we communicated and promoted the opportunity to donors, and how we advised others (e.g., nonprofits) to communicate the opportunity. It seemed likely that the overall match percentage directed by EA donors could be far lower than the 52% directed in 2019, and the 65% directed in 2018.
The overall match percentage in 2020 was 25%. The larger amounts donated partially compensated for this, and more matching funds were directed than might be inferred: The $411k in matching funds directed in 2020 is 73% of the $563k directed in 2019, and 88% of the $469k directed in 2018.
Ambiguous terms of matching program
Based on Facebook’s wording around the 10% match, we had uncertainty about whether the $20,000 limit for U.S. donors was on the amount donated, or the amount of matching funds received. This ambiguity affected a minority of donors and did not affect the 100% match. It turned out that the $20,000 limit was on the matching funds received, not on the amount donated.
Non-U.S. donors continue to face restrictions
Effective altruism and effective giving are ideas that from their beginning have been international. It’s unfortunate that donors outside of the U.S. continue to face very low donation limits during Facebook’s Giving Tuesday match. This limits the value of their participation in the match, given the time needed to prepare, and the loss of tax benefits (in some but not all cases).
Mailchimp issues
We primarily use Mailchimp mass email campaigns to communicate important information and updates to donors. It’s important that donors see the emails we send. We design and test each campaign thoroughly before sending, and it took significantly more work and troubleshooting than expected to craft emails that we were confident wouldn’t be caught by Gmail’s promotion filters. We may investigate alternatives to Mailchimp as part of our preparation for the 2021 match.
Isolated reports of PIN issues
Facebook Payments allows for optional PIN setup. Our instructions intentionally omitted mention of the PIN, because we did not want donors to enable it, as they would then need to disable it before the match. An enabled PIN would slow or prevent donation submission.
Most donors had not enabled a PIN, so they didn’t need to know or think about it. However, we did hear from a few donors for whom an enabled PIN had caused problems for them during the match.
Unpredictable delays in donation processing
As discussed earlier in Our strategy to direct the matching funds, many donors experienced unpredictable processing delays. These delays were one reason that many donors that correctly followed our instructions did not receive 100% matches on any of their donations. (In virtually all cases these donors would have still received 10% matches on these donations.)
Challenges in nonprofits receiving amounts
Nonprofits receive donated amounts from Facebook Payments (in January 2021) or Network for Good (in February or March 2021), and matched amounts from Network for Good in all cases (in February or March 2021). After Giving Tuesday 2019, CEA only received these payouts after a delay of several months (i.e., in May 2020). This was particularly significant because we were expecting CEA to receive a large amount ($229k).
After Giving Tuesday 2020, Rethink Charity experienced a significant delay in receiving donated and matched amounts from Network for Good, and the amounts were large (i.e., over $200k). However, the reason for delay was less concerning than what caused the challenge for CEA in 2019. The issue with amounts for Rethink Charity was resolved in April 2021.
Challenges like these highlight the risk of amounts never arriving to some nonprofits. It remains very important to follow up with Network for Good if amounts are not received in the expected timeframe.
Areas of improvement since 2019
Each year we try to learn from our experience to increase our performance. This section reviews some of the areas we've improved in 2020 compared to 2019.
Leading up to the match
The project received more funding
EA Giving Tuesday received grant funding based on a recommendation from the EA Infrastructure Fund. This funding was used to help build the project’s resilience by training more team members to be able to run this initiative, and to attract more team bandwidth for work before and after the match.
Inclusion of the EA Funds donor lotteries
Donations to the EA Funds donor lotteries were eligible for Facebook matching this year. This improvement was facilitated by CEA, our team (thanks especially to Rebecca Baron), and participating EA Funds donors, who had to do a little extra work reporting donations and matches.
The donor lotteries ($20k, $100k, and $500k) received $131,280 in donations and $33,620 in matches. We had 18 EA donors participate, and the large majority of donated amounts (76%) were for the $500k donor lottery. Matched amounts were allocated, at the discretion of each donor, to any organization on the EA Funds platform.
More nonprofits, funds, and programs available
We created Facebook fundraiser pages for 111 EA-aligned organizations and projects, an increase from 81 in 2019. As in past years, we coordinated with a large number of nonprofits and organizations to ensure that more hard-to-reach EA donation opportunities could be eligible, and to ensure that all donated and matched amounts were allocated accurately and received.
This year we also worked to ensure that certain organizations recommended by Open Philanthropy for their work related to the COVID-19 pandemic (e.g., Johns Hopkins) or criminal justice reform work (e.g., National Council) were eligible for Facebook matching and included in EA Giving Tuesday.
New recommendations for donation submission timing
Were it not for these recommendations, we predict that far less matching funds would have been directed to EA nonprofits in 2020, given the exceedingly fleeting processing window eligible for 100% matching (i.e., seconds :00 and :01). That said, we learned through our match follow-up survey that these timing recommendations could be further improved, which we will address in 2021.
Improved tools for practice and preparation
As noted in Our strategy to direct the matching funds, our new donation simulator was a primary tool for practice and preparation. On our website and during our virtual talk, we promoted preparation tools (i.e., the simulator, our instructions, Facebook practice donations) and explained the rationale for using them. Thanks to William Kiely, we also had new training videos available.
Strengthened day-of monitoring and communication plan
As noted in Our strategy to direct the matching funds, our Giving Tuesday monitoring and communication plan was more robust and comprehensive than in past years (thanks especially to Megan Jamer).
Specifically, we had improved plans for how, where, and when to communicate last-minute updates and changes to EA donors. We maintained clear messaging about where to check for last-minute updates before the match, and why it was essential for donors to do this.
Shortly before the match, our testing team maintained a peak testing frequency of every three minutes to maximize our chances of detecting last-minute changes.
Although we automated what we could, our day-of communications and monitoring strategy still required attention and execution from our team.
Improved communications and engagement with stakeholders
We developed a more comprehensive plan for communicating with donors and nonprofits:
We added or significantly updated several pages to the EA Giving Tuesday 2020 website, including “New in 2020,” “Team,” “Nonprofits eligible,” “Testimonials,” and our FAQ.
We significantly increased the number of EA newsletters that featured EA Giving Tuesday (thanks especially to Rebecca Baron).
Responses from the community feedback survey we conducted influenced what we included in our virtual talk and our FAQ.
As not everyone who supports the nonprofits we coordinate with may identify as an effective altruist, we commonly opted for “highly effective nonprofits” instead of “EA nonprofits,” and “people who support effective nonprofits” instead of “EA donors.”
The nonprofit feedback survey we conducted helped us better understand how nonprofits may best promote this event to their donors. (Feedback from nonprofits also suggested that EA Giving Tuesday probably is best promoted to people already quite involved in the effective altruism community.)
Improved use of automation
We used Zapier to automate form signups for Mailchimp and SimpleTexting. (Special thanks to Gina Stuessy.) As noted in Our strategy to direct the matching funds, where possible we automated our email and SMS campaigns in order to preserve our team’s attention for monitoring and updating on match morning.
The match itself
More donors
Although the number of pledgers was the same (250) in 2019 and 2020, we saw a significant increase in people donating through our fundraiser pages: 466 in 2020, up from 330 in 2019.
Record amounts donated
In 2020 $1.6 million was donated through EA Giving Tuesday, up significantly from $1.1 million in 2019. (Note that 2019’s donated amounts likely didn’t fully reflect EA donors’ intentions; our 2019 retrospective explores this.)
We infer that 2020’s record amount of donations was the result of several factors: increased participation in EA Giving Tuesday (i.e., more donors), reduced payment problems (based on follow-up survey responses), and better preparation for last-minute changes.
Higher proportion of EA Giving Tuesday donors got a match
The 2020 match structure had negative and positive effects on match proportions. To the negative, fewer donated amounts got a 100% match. To the positive, a higher proportion of donations got some match.
In past years a proportion of Giving Tuesday donations received a 100% match and the rest of donations received no match. In 2020, of the 221 individual donations for which we have detailed data (from survey responses), 99% were matched at either 10% or 100%. The match percentage breakdown for these 221 donations was 78.8% of donations matched at 10% and 20.4% of donations matched at 100% (0.9% of donations received no match).
Potentially higher percentage of 100% match directed
In 2019, donors directed 8% ($563k) of the $7 million, 100% matching funds to EA nonprofits. In 2020, there is uncertainty about how much of the $411k in matching funds directed by EA donors were from the 100% match versus the 10% match. However, our best estimate is that $275k of these matching funds were from the 100% match. This would imply that in 2020, EA donors directed 14% of Facebook’s 100% matching funds, a large increase from the 8% in 2019.
Reduced payment problems
We saw decreases in payment problems among every group of donors relative to previous years. The most notable decrease was among U.S. donors attempting to donate more than $10,000. Among these donors, the percent reporting payment problems were 68% in 2018, 59% in 2019, but only 32% in 2020.
Match follow-up work
Less significant problems with nonprofits receiving funds
In 2020, we did outreach work with EA nonprofits before and after Giving Tuesday to help them streamline the payout process, or troubleshoot any issues as they arose. We believe this work may have contributed to reducing the number of nonprofits experiencing substantial delays or challenges in 2020 (but see Challenges, above, for related discussion).
More donation-specific information about match results
As mentioned above in Follow-up with donors, we conducted two follow-up surveys. Our match follow-up survey revealed useful insights about the characteristics of donations that were matched at 100%. We expect these insights to help inform our recommendations for donors in 2021.
Lessons learned in 2020
Our new recommendations were necessary for the 100% match
Only donations processed at seconds :00 and :01 of the match were eligible for 100% matching. For the best chances of having donations be processed in this window, it was necessary to follow our updated donation submission recommendations.
Our project seems best for specific target audiences
In 2020 we supported (i.e., made Facebook fundraiser pages for and in many cases did outreach to) 111 EA-aligned nonprofits and projects, some of whom have substantial donor bases outside of EA. Some of the most useful feedback we got from the nonprofits we communicated with was that this opportunity was likely best for people who already have some prior knowledge and involvement with effective altruism.
Relatedly, some of our last-minute emails sharing important updates before the match were written assuming all recipients would have certain background knowledge about the match. This was an oversight on our part: We continued to promote the opportunity through late November, but did not fully think through the experience of a first-time participant learning about EA Giving Tuesday mere days before the match.
Strategy and reflection work should have happened earlier
In October, we spent significant time building a community survey to learn more about people’s reasons for participating (or hesitating to participate) in EA Giving Tuesday. We also spent time analyzing the responses. However, this timeframe was too short to meaningfully act on the results, with a few exceptions. Similarly, we did not have enough time to act on some aspects of our promotion strategy
Reasons for reluctance to participate in EA Giving Tuesday
In our community survey we learned more about common reasons for reluctance to participate in EA Giving Tuesday. Hesitation commonly had to do with
concern about how quickly the match would end (and this was even before the match rules were released), or
reluctance or unwillingness to use Facebook.
Estimating our impact
Matching funds directed
In 2020, the EA Giving Tuesday initiative directed an estimated $411,693 in matching funds to highly effective nonprofits.
The resources contributed to EA Giving Tuesday in 2020
Paid labor costs
Approximately 297 paid hours were contributed.
These paid hours cost approximately $7,988.
Note: There is some uncertainty due to averaging out hourly rates and calculating costs across currencies. In addition, other Rethink Charity staff did work on EA Giving Tuesday (approximately 15-30 hours) that is not reflected in the above cost and hour estimates.
Voluntary contributions
Approximately 466 unpaid hours were contributed (conservative estimate).
Note: This estimate does not include additional unpaid efforts made by some of our testers, or by people at the nonprofits we coordinated with to regrant donated and matched amounts.
Non-labor costs
Technology services cost $845.
Test donations cost $160.
Other non-labor expenses cost approximately $100.
Note: Some team members did not expense their test donations, and these amounts are not included above.
Factors relevant to estimating the impact of EA Giving Tuesday 2020
Factors that could increase impact
Currently we focus on the matched amounts. If we were to consider some portion of the $1.6 million in donated amounts as counterfactual, the impact of EA Giving Tuesday could increase significantly.
Based on our follow-up surveys after EA Giving Tuesday 2018, potentially 12% of donated amounts could be considered counterfactual (with some caveats). See the EA Giving Tuesday 2018 donation follow-up survey analysis for more discussion.
In 2020, we did not ask EA donors in our follow-up surveys whether the initiative motivated them to donate more.
Factors that could decrease impact
There are a number of factors that could have potentially reduced our counterfactual impact. We collected data on some of these factors in our 2020 donation follow-up survey, and we discuss three of them below. (Visit the “Our Impact” section of our donation follow-up survey analysis for more information.)
Would EA donations have been matched without our work?
Our best guess is that $275k of the $411k in matching funds came from the 100% match. We think that very few, if any, of these 100% matching funds would have likely been directed without our efforts.
Our best guess is that $136k of the $411k in matching funds came from the 10% match. We think that some amount of the 10% matching funds might have been directed without our efforts. The 10% match lasted far longer (i.e., over three hours), and 44% of survey respondents (weighted by donated amounts) reported they would have “definitely” or “probably” participated without our efforts. However, without our efforts, it is not clear whether these donors would have donated the same amounts or less.
Did donors donate less effectively?
We think this effect is minor: Approximately 2% of survey respondents reported donating to a nonprofit they thought was less effective. These respondents represent 1.3% (approximately $15k) of the $1.1 million in donated amounts accounted for in the survey.
Did donors lose tax benefits?
Approximately 17% of survey respondents reported losing the ability to claim one or more type(s) of tax benefit (i.e., related to Donor-Advised Funds, appreciated shares, Gift Aid, and tax deductions or tax credits). All together, these respondents represent 9.3% of the $1.1 million in donated amounts accounted for in the survey. Of this 9.3%, 8.8% concern Donor-Advised Funds or appreciated shares.
We have not estimated the value of this potential loss of tax benefits. The vast majority concern Donor-Advised Funds or appreciated shares, so it could be reasonable to estimate the value of these lost benefits at around 10% of the donated amounts they applied to. This would imply a very small overall loss of tax benefits.
Visit the “Our Impact” section of our donation follow-up survey analysis for more information.
Questions for 2021
If the Facebook donation matching opportunity is available again in 2021, our experience in 2020 raises some questions for 2021.
What should we expect Facebook’s matching program to look like?
Although the amount of matching funds available had changed over time, the match structure had remained consistent (i.e., all matching funds offered at 100%) in 2016, 2017, 2018, and 2019. This changed when the 2020 match had a significantly different structure. We see several potential options for the structure of the 2021 match, including the following:
Same details as 2020 (combined 10% and 100% match)
10% match only
100% match only (as in 2016 to 2019)
Facebook increases the amount of matching funds available
Facebook discontinues its Giving Tuesday matching program
What should our promotion strategy look like?
In 2020, we made some changes to our promotion strategy in response to the match structure. We still did a moderate amount of promotion (e.g., through a virtual talk, our email newsletter, our Facebook group, and a variety of EA newsletters), but we curtailed our one-on-one donor outreach.
If the 2021 match has the same details as 2020, we expect that the 100% match will be extremely competitive, even more than in 2020. But the 10% match may be achievable for most people willing to do modest amounts of preparation (we infer this because the 10% match lasted approximately 3 hours in 2020). Given this contrast, we may need to revisit which audiences we promote the match to, how to promote it to different audiences, or to what degree promotion is expected to be valuable.
The 2021 match may lack a 100% match entirely, with only a 10% match offered. In this scenario, EA donors could weigh the reduced potential upside of 10% matching in the context of the time required to prepare (discussed above), the planned donation amount, and the likelihood of achieving a 10% match (this likelihood could be quite high). These considerations would also be weighed by our team as we built our strategy for promotion.
Facebook could also increase the value of the 2021 match compared to 2020 by increasing the amount of 100% matching or by making other changes. In any case, our promotion strategy would depend on the value of the match, its structure, how competitive we expect it to be, and our team’s capacity and prioritization.
How can we increase the amount of donations matched at 100%?
We expect that donors will need to submit donations with even more timing precision than they did in 2020 to maximize their chances of directing the most matching funds. This consideration will be baked into our recommendations and instructions for 2021.
What level of organizer effort is worthwhile?
It takes significant time to actively prepare donors and nonprofits for EA Giving Tuesday. It is true that our initiative is more established and that the initiative has built up better workflows, systems, and experience among team members. Nonetheless, facilitating regranting and restriction arrangements, building instructions, and conducting outreach are particularly resource-intensive activities for our team.
If the 2021 match is structured as it was in 2020, it may be worth considering curtailing the amount of effort the organizing team puts into the match. There may be particular activities that could be put on hold that would help limit the initiative’s time requirements.
If the 2021 match is structured as a 10% match only (i.e., no 100% match), it will likely make sense to significantly curtail the amount of effort the organizing team puts into the match.
It’s also possible that Facebook will increase the matching funds available in 2021, or otherwise restructure the match so that the expected potential upside is higher than it was in 2020. In that case, it is likely that the amount of organizer effort should remain as it was in 2020.
A challenge in making these decisions is that we don’t expect to know what the 2021 match details will look like until a few weeks before the match begins.
Thank you!
EA Giving Tuesday brings together hundreds of people around the shared goal of maximizing the amount of donations and matches directed to highly effective nonprofits. This would not be possible without so much help, and many thanks are due:
We thank the generous donor who gave $13,000 towards EA Giving Tuesday’s operating costs, based on a recommendation from the EA Infrastructure Fund.
Thanks to the 446 donors that participated in the match by donating to EA nonprofits.
We appreciate everyone at the EA nonprofits we liaised with for your help with coordinating, especially if you helped us arrange a restriction or regranting plan.
EA Giving Tuesday has significantly improved over the years because of the effective altruism community’s feedback, ideas, and critique, and we greatly appreciate the community’s efforts this year as well.
Thanks to everyone on the EA Giving Tuesday 2020 organizing team: Gina Stuessy, Will Kiely, Rebecca Baron, Megan Jamer, Avi Norowitz, Angelina Li, and Nix Goldowsky-Dill. (Marisa Jurczyk, a previous year’s organizer, gave valuable feedback on our communications and did many other tasks to ensure a smooth transition between team members.)
The initiative also benefited from the work of several additional volunteers that provided high-quality help on a wide variety of tasks.
