My learnings from a year of handling outreach at Effective Thesis
Context for this post
This post is meant to be valuable for those handling marketing and outreach (with any level of experience) at EA orgs, specifically those who may be building their strategy from scratch and working with smaller budgets ($10-50k/year). It includes my recommendations for thinking about outreach and the results of handling outreach at Effective Thesis while I worked there. My recommendations are definitely not applicable to everyone, though larger EA orgs trying to formalise their marketing may still find parts of this helpful.
This is particularly useful for those trying to reach a target group similar to ours (talented students from undergraduate to PhD levels interested in research careers) but I expect that you’d also find this interesting if you are trying to reach a different audience.
The duration for reporting is a little longer than a year: from 1 May 2022 to 31 May 2023, though in most cases, I did not start marketing activities till the end of May 2022.
Special thanks to David Janku, Silvana Hultsch and Grace Adams for feedback on this post. All mistakes remain my own.
Executive Summary
Note: I use outreach/marketing/communications in this post interchangeably, though they perform different functions. These are my opinions based on my personal and mostly for-profit experience, along with findings from outreach at Effective Thesis.
Align outreach with your Theory of Change and vice versa
Marketing is not an autonomous arm and important findings from marketing should inform your product/services.
Branding doesn’t have to be an afterthought: without a well-defined brand, the organisation may struggle to effectively communicate its key objectives, outcomes and evaluation findings publicly.
Define your target audience sharply through initial audience research, keyword research and continuous feedback
Learnings from defining our target audience:
We found that Academic Twitter is very active and probably a good way to engage with students interested in a research career.
Many people find out about programs such as CERI, CHERI and SERI through their university newsletters.
Using Google Trends and the keyword research planner to figure out popular terminologies helped us understand specific terms our target audience was searching for.
Speaking to EA groups from different countries helped us in creating a (non-exhaustive) database of universities, foundations and student clubs we could reach out to.
Define messaging well
The steps to crafting messaging well can include establishing your org’s tone of voice, tailoring messages to different segments, using emotional appeal (carefully) and testing and refining based on feedback.
We used different messaging strategies centred around inspiration, ambitiousness, solving real-world problems, encouraging academic success and finding a topic that’s ‘the best fit’ for students. Campaigns that performed best were ‘choosing inspiration’ and one that appealed to students’ ambitiousness.
Select the right channel
After defining the target group and learning more about audience behaviour and where they spend their time, the next step would be to select channels based on audience reach, relevance, communication objectives, budget and resources.
Channels we selected and their results
Email outreach—reaching out to university course coordinators, relevant foundations, and student clubs to share what Effective Thesis does. This was time-consuming and results were not as trackable since we weren’t directly sharing the word about Effective Thesis but did result in high conversion rates from student clubs and research foundations sharing the word about Effective Thesis (16.4% and 11.76% respectively).
Social media advertising—we focused on running campaigns on Twitter and Facebook as they seemed promising in terms of audience reach and were inexpensive. We found that campaigns which ran for longer (>1 month) performed better than shorter campaigns. Overall, the results from these were poor compared to other channels (GoogleAds, email, virtual sessions), both in terms of number of conversions and conversion rates. This might have been different if we promoted services like our newsletter instead of coaching.
Google Ads—has extensive reach, is intent-based, and uses keyword targeting, which can be especially useful if the keyword research stage produces promising results. Additionally, for nonprofits, this can be especially cost-effective since Google awards a $10,000 USD/month grant to run these ads. We received a total of 921 conversions in a year of running Google Ads, and utilised 88% of our grant spend overall. A significant portion of our total conversions from all outreach efforts came from Google Ads. We also received a lot of low-quality applications, which resulted in us excluding certain keywords and locations from our campaigns.
Running sessions with EA groups—we ran ~1-hour virtual sessions on “Changing the world with your research” to introduce EA student group members to Effective Thesis and possible ways of thinking about using their thesis as a path to impact. We ran 8 virtual sessions, with a total of ~112 attendees and received ~33 sign-ups for at least one service, making the conversion rate ~30%, the highest of any other outreach activity.
Influencer marketing—we targeted small-medium sized influencers (3-100k followers) on Twitter, along with a podcast, and ran a total of 5 paid campaigns. Our total spend was ~$900, which resulted in a cost per acquisition of $17.3 (average) and a conversion rate between 2-6%. The process of reaching out and getting a response turned out to be extremely time-consuming. In retrospect, I might have worked with an influencer marketing agency instead.
Channels we did not select that seemed promising
SEO: I would recommend new orgs to work on their SEO, though we were unable to do this due to high competition and a very low DA ranking. We still were able to rank on the first page on Google for the term ‘thesis topic’ by making minimal changes to our webpage.
YouTube ads: these seemed really promising and still do, but we didn’t have the capacity to do these well enough and prioritised other channels.
Reddit: I did not select this due to a lack of presence on Reddit and did not have the capacity to provide quality answers on relevant threads.
PR: we primarily did not invest in public relations as we did not have a lot of our own research or publications to promote, though I think this is a really helpful channel to build legitimacy.
LinkedIn ads: seemed most promising because of context relevance compared to other social media platforms, but it’s much more expensive to promote and I was uncertain which service might work best on this platform.
Additional learnings from outreach
Google Analytics: Analytics plays a critical role in understanding user behaviour and website performance. Google Analytics provides valuable insights into popular pages, traffic sources, and conversions. It is recommended to start with simple metrics and set up tracking for all channels.
Results from Google Analytics: Outreach efforts resulted in a significant increase in website traffic (475%) and conversions (235%) compared to the previous year. Referrals and emails were more effective than paid search in generating sign-ups, likely due to higher engagement levels and contextual relevance.
Logo Design: The process of creating a logo helped refine the messaging and target audience. Multiple iterations were needed to clarify messaging. Eventually, working with a freelance designer proved successful in achieving the desired logo.
Switching to the Newsletter: promoting our coaching faced resistance, leading to a strategic shift towards launching the Future Researchers Newsletter. This approach allows students to engage with Effective Thesis gradually and provides thesis topic inspiration, advice, and tools. The newsletter has gained over 350 sign-ups and shows promise in generating interest and eventual sign-ups for coaching.
Preliminary Newsletter Results: The open rates and click rates of the newsletter varied across different editions with the average open rate and click rate 38.9% and 7.05%, respectively.
Remaining uncertainties to consider
A crowded space: Many EA organisations/projects target the same audience. We need better coordination to know where to not duplicate efforts, and how to cross-promote where possible.
Defining audiences through metrics: metrics like targeting students from ‘Top 100 Universities’ may overlook those that need the most support.
Quality vs. quantity: We might want to spend more time optimising for a narrow pipeline (better quality leads) but use automation to reach a broader audience.
Hitting the ground running: I’m concerned there is an expectation to ‘execute’ marketing strategies before taking enough time to scope and research.
Organic social media: Allocate minimal resources here unless content creation is central to your organisation’s strategy.
Recommendations for outreach
These are my recommendations for outreach that most orgs should work on. Many of these might not work for all types of projects, but I think it’s quite important to consider each of these steps to know what might work for you. This is not an exhaustive list and is based on the experiments we conducted at Effective Thesis, along with the lessons learned from those experiments.
Aligning outreach with your theory of change
Before starting to work on your marketing strategy, a (somewhat obvious) task is to identify your organisational goals and align them with your marketing goals. This means that your organisational goal informs your marketing goals, but findings from market research also influence your theory of change.
At Effective Thesis, this looked like taking learnings from our market research, i.e. by surveying students who we thought were the right fit for our services, and finding whether there was a gap we could fill.
The importance of branding for nonprofits[1] is often overlooked and is sometimes done mid-operations or after monitoring and evaluation rather than as a tool to flesh out the larger theory of change. Since EA orgs are growing and there’s a chance of multiple organisations reaching similar kinds of people, it’s important to know whether there’s a need/gap or message you’re filling or you’re doing similar things to reach the same people.
Branding may include
a) Visual identity -
Logo, colour palette and typography
Messaging and positioning
Mission statement, tagline and tone of voice
Crafting a compelling mission statement helps articulate the organisation’s theory of change and communicate its goals and intended impact. The elaboration likelihood model suggests that persuasive messages, including mission statements, can influence attitudes and behaviours when they are clear, relevant, and emotionally appealing.
b) Stakeholder Engagement -
Identifying and segmenting the range of target audiences
Establishing communication channels like social media, newsletters or events
c) Impact -
Transparent reporting: Communicating impact and outcomes through annual reports, impact assessments, and success stories
Brand reputation: Building a positive brand reputation through consistent delivery of high-quality programs and services.
Clear and transparent communication of an organisation’s impact and mistakes might aid with accountability internally and help its reputation
Why I think branding is important
It helps the organisation articulate its theory of change, goals, and intended impact in a compelling and coherent manner. Theoretically, without a well-defined brand, the organisation may struggle to effectively communicate its program outcomes and evaluation findings to stakeholders.
Branding can sometimes help clarify key messages and big-picture goals of the org. At Effective Thesis, the process of creating our logo helped us in clarifying our audience and key messages.
Brand strength is proven[2] to have both conative outcomes (e.g., the intention to donate) and affective outcomes (e.g., trust)
Defining your target audience sharply
Having a clear and detailed understanding of the target group helps you relay what strategy and channels to undertake. There may be several methods to do this: user interviews, focus groups, researching the sector (not just the audience you’re trying to reach but finding out more about the ecosystem), keyword research and desk research to find reports around the cause area you operate in.
Initial audience research (surveying)
Other than knowing we wanted to reach out to students who were thinking about their thesis topic, we also wanted to learn how to reach talented students across the world, what their challenges and motivations were and their initial impressions of Effective Thesis. We decided to have paid interviews with students (top rejected applicants) who applied to CERI, SERI and CHERI Fellowships to understand:
Social media usage—to understand where we would be likely to have ad campaigns
Motivations to apply for these programs—to understand behaviours
Impressions about ET—whether our messaging was being understood
Why we selected applicants from these programs
Aligned motivation: we could safely assume most students applying to these fellowships were interested in doing research work.
Cause alignment: similarly, we could also assume that students applying were aware of and were interested in similar cause areas to our recommended research directions.
Talent: we reached out to the ‘top’ rejected applicants in 2 out of 3 of these fellowships, which would allow us to speak to students we ideally want to reach.
What we learned
Twitter was the primary engagement platform for most Masters and PhD students. Facebook for several, but only to keep in touch with friends. Undergraduates preferred Instagram (or even Discord, but sparsely). YouTube was common among most for informational/educational purposes.
A common theme for motivation to apply for these programs was topic relevance (an interest in X-risks).
Many students (~40%) found out about the fellowship through their University newsletter and thought it would be interesting to apply.
Some mentioned testing their fit for research and it would be a good way to develop their skills with others who have similar interests.
Some also mentioned that collaborating in person was very attractive.
Students who were already highly engaged with EA found our website easier to understand and go through. Most students wanted to apply after speaking to us, but were unsure they would from the website alone—this signalled that we needed more clarity in our communications.
We also learned what sort of content, resources and workshops students might be interested in that they don’t have access to otherwise.
Keyword research
For more granular information, researching beyond behaviours led to using tools like keyword planner and google trends to view which keywords were widely adopted and which results we could use for intent-based searching. We found this to be most helpful in using specific keywords for our website and somewhat helpful for our campaigns. For more live data, we used social listening to gather what terminology (especially those in graduate school) were using.
Eg. We found that ‘thesis topics’ was a much more relevant term than ‘research projects’ for the students we wanted to reach. We also found that ‘early career researchers’ and ‘aspiring researchers’ were terms people in our target group were likely to use and recognise.
Feedback
This is something best done continuously. We had feedback loops every 6 months from mid-2022 onwards on our content, all services and any external outreach we did, mainly through feedback forms. This was extremely helpful in learning how to communicate expectations around our services, and knowing how to better improve them.
Eg. Collecting feedback around virtual sessions with newer EA groups helped us learn that they were effective in students signing up for coaching.
Defining messaging well
Once you have a clear understanding of your target audience and their needs, the next step is to define your messaging effectively. Well-defined messaging helps communicate your organisation’s goals, values, and intended impact in a compelling and coherent manner. It ensures that your outreach efforts resonate with your target audience and motivate them to take action.
Here are some key considerations for defining your messaging well:
Core Message: Craft a Clear Value Proposition
Your core message should succinctly convey the unique value your org offers to your target group. Consider prompts like—“Why do we exist?” or “What problem are we solving?” and “Who should care, and why?”
For Effective Thesis, our value proposition (different from our org mission and vision) was to provide personalised support and inspiration to students writing their thesis.
An alternate, perhaps more effective value proposition could be—to provide support to students interested in pursuing an impactful research career—which clarifies motivations and somewhat narrows who we’re trying to help instead of when we’re trying to help, but we found that influencing thesis topic choice may have additional value, theoretically in:
Inspiring students to learn about EA causes in an applied manner
Testing their fit for research
Writing a thesis that may be more beneficial than selecting a less impactful topic
Tone of Voice: Establishing the Right Tone
Consider the tone that aligns with your organisation’s values and resonates with your target group. It could be authoritative, compassionate, inspirational, or conversational, depending on your org’s personality and the nature of the message.
In the process of refining our tone of voice, we analysed the tone of voice of other organisations doing similar work or targeting similar audiences to ours to better understand how our voice might be similar or unique. I don’t think we have been able to figure out our tone of voice perfectly yet, but our aim was to present a tone that was informative, transparent, professional and inspirational.
Key Messages: Tailor Messages to Different Segments
Your target audience may consist of different segments with varying needs and interests. Tailoring your messages to specific segments allows for more personalised and impactful communication. Identify the key messages that resonate with each segment and address their specific pain points or aspirations.
For instance, we tried a reverse approach to figure out our messaging strategy with the intent of receiving applications from our ideal target audience, i.e. - ‘What don’t we want people to perceive about us?’ We then decided we didn’t want to highlight that we are a ‘free service’ as a key message. We also did not think it would be wise to have messaging that competes with university resources.
Emotional Appeal: Connect on an Emotional Level (carefully)
Appealing to emotions can significantly influence attitudes and behaviours. Consider how your messaging can evoke emotions that align with your organisation’s mission and resonate with your target audience. Use storytelling techniques to create narratives that evoke emotions and illustrate the positive impact of your organisation’s work. Using this approach sometimes seems risky or somewhat manipulative to me, so it is wise to present this in a way that isn’t inauthentic or ‘hides’ important information and actually addresses concerns your audience may have.
For Effective Thesis, we wanted to invoke a sense of inspiration for our messaging to be more memorable and persuasive. We used inspiration from our own content pieces that we thought would promote that messaging, like this one.
Test and Refine: Iterate Based on Feedback
Regularly gather feedback on your messaging to understand its effectiveness and make improvements. Conduct surveys, collect user feedback, and analyse the response to your messaging efforts. This feedback loop allows you to iterate and refine your messaging based on real-world data and insights. Continuously monitor and adjust your messaging to ensure it remains relevant and resonates with your target audience.
How we defined our target audience better
At Effective Thesis, an important consideration for us was to segment and prioritise our audience since our services were applicable to undergraduate, master’s and PhD students worldwide. Some initial ideas about our ideal audience:
We wanted to reach students who would become good researchers/were considering a research career
Since we’re also reaching out to undergraduate students, looking for research foundations students apply to was a starting point, along with knowing where considering research is a popular career path.
We wanted to reach talented students
These may not always be at prestigious universities, but it’s more likely we find students who are talented at top 100 universities, so our focus was to reach top research-based universities across the world.
We wanted to reach the students we could help most (those who would have otherwise not considered a thesis or research in one of our recommended topics)
Being members of the EA community is anti-correlated with this because people in EA are much more likely to hear about these problems from other sources.
Due to some of these ideas, we decided it would be good to answer:
What are the traits of good researchers
To do this, we reached out to promising students we coached previously, along with senior researchers to study and interview them to find out more about these traits. We were only able to speak to people who mildly/considerably engaged with EA, so the results of these are quite skewed.
Where do we find talented students globally and how do we reach them (specifically non-EA students)
We reached out to knowledgeable individuals and researchers from ~23 different countries to find out about top universities, what programs existed for talented students there and any communities we could reach out to and created the country diversity database. This is a non-exhaustive list and we’re unable to share any contact information for outreach, but it could be extremely useful for others with similar target audiences to take a look at.
Messaging strategies we used
Choosing inspiration: We ran a set of campaigns focused on messaging that encouraged students to find a thesis topic that ‘inspired’ them. The idea behind this was because most undergrad/masters students think of their thesis as just another project (anecdotal evidence from talking to many students and university group leaders) and not as something very valuable for their career. Since we want students to get and give more value out of their thesis, we think an avenue that would excite them is to be inspired by the results of their own work.
Solving real-world problems: Similar to inspiration, some students get value out of not just stimulating their curiosity and building skills but also by working on issues that solve real-world problems. Since our ideal target group consists of students that are altruistic-leaning, we hoped this message would speak to them specifically.
Academic success: We also ran a single campaign in parallel to ‘choosing inspiration’ and ‘solving real-world problems’. This focused on students who care about academic success. We imagined that academically successful students were likely to make good researchers, and a percentage of them would also care about well-done research on their thesis that might help them later in their careers.
Direct, smaller subset of PhD students—calling out to ambitious PhD students:
We also experimented with a set of ads for PhD coaching made for students who self-identify as ambitious. This messaging had the risk of PhD candidates who may not think they’re ambitious enough, or work on problems that seemed too difficult for them to pursue. This did not seem like a large risk due to the fact that PhD candidates are generally very committed to their work and already work on hard problems.General coaching for all students—finding your fit/best topic for you:
A more general approach to talking about selecting a thesis and coaching, this messaging strategy leaned heavily on the fact that students want what is ‘best suited’ or ‘well tailored’ to their interests. By personalising our coaching, we’re able to find impactful topics that are best for them.
Results
Looking at the click-through and conversion rates of these campaigns across Twitter and Facebook, we find that the most effective messages are likely to be those that encourage students to find ‘inspiration’ in their topics and help them self-select themselves as ‘ambitious’. We found the least promising avenue to be ‘finding your fit’, as it is too general, though this may just have a low click rate because the design was not as clean or good quality as some of the others.
Note: From these campaigns, other than the ones tailored to PhD students, we didn’t send users to landing pages and sent them to our website instead, which isn’t optimised for conversions. This is a large reason why the click and conversion rates are not as high as they could have been.
Full keyword and messaging report here.
Selecting the right channels
After defining the target group and learning more about audience behaviour and where they spend their time, the next step would be to select channels based on audience reach, relevance, communication objectives, budget and resources.
Through the audience research phase and learnings from previous outreach attempts by Effective Thesis, I selected a mix of channels for marketing to reach students that seemed most promising.
Email outreach
From our audience research, we created an outreach database with contact information and categories that could potentially help us promote Effective Thesis to students. These included: research coordinators at Universities, scholarship and research foundations, student clubs, EA groups and influencers.
Why this seemed promising
Instead of only reaching out to individuals, we thought to maximise our outreach by reaching out to ‘nodes’ within (and outside of) the university environments. Our main goal was to reach out to individuals with high credibility and/or trust within these systems in a way that would be a low effort for them to do and would encourage groups of students to seriously consider applying for Effective Thesis services.
Eg. Theoretically, students would be much more likely to regard an email forwarded from their program or research coordinator than they would if they were sent an email by a nonprofit they have never heard of before, assuming they open it at all.
Process
A large part of this was reaching out by cold-emailing to ask contacts to share our digital flyer, highlighting coaching as our primary service. I would send out ~3 personalised emails per contact (including follow-ups). We maintained a comprehensive AirTable with information about each contact and dates they were reached out to.
For research coordinators, I tried different strategies promoting Effective Thesis as a resource to support talented students that included highlighting either 1-1 coaching, our research topics, workshops (we were running a reasoning transparency workshop at the time) or specifically, the Exceptional Research Award.
For student clubs other than EA groups, I focused on general societies across top research universities (AISEC, students interested in social work etc.) along with cause-specific clubs (like AI societies, biological sciences, and philosophy clubs) that I was able to find.
For all other types of outreach emails, we would send out a summary of what Effective Thesis does and ask to share an email or ask for a shoutout in exchange for cross-promotion. This process was iterated several times based on results from experimenting with messaging, subject lines, the product/service and the content of the email.
Results
I sent out 323 emails, focusing on research coordinators at Universities, student groups (both EA and non-EA societies) and influencers. Out of these, the highest results for emails shared came from student groups (EA) and student societies.
Student clubs were happiest to share what Effective Thesis did and generally responded quicker than all other categories of contacts we tried to reach out to.
It is worth noting that the scholarship and research foundations we reached out to were mostly those we had contacted previously, some of whom had agreed to share our emails in the past. We sent similar messaging that worked previously, however, this time we didn’t receive any responses.
Channel Type | Number of contacts approached | Conversion rate (%age that agreed to share our advert) |
Foundations/Scholarships | 14 | 0 |
Research Opportunity | 17 | 11.76% |
Student Group/Society | 189 | 16.4% |
Specific Course/University | 52 | 11.53% |
Influencer | 47 | 4.25% |
Other | 3 | 0 |
Channels used for Effective Thesis outreach and their conversion rates
Learnings
If you’re asking for a favour virtually from someone who isn’t aware of your org, it’s probably best to make it extremely easy for them to say yes. Sharing email and social media templates that people could use (instead of asking them to reach out to people where they would need to add their own message) resulted in more people sharing our advert.
Emails can be automatically forwarded to the ‘promotions’ tab. At some point in the process, someone pointed out that my Effective Thesis emails were shown in their promotions instead of the primary tab (in Gmail). After some research, I found out that may be due to sending emails with a large number of links and using words like ‘apply’ or ‘sign up’.
Networking in academia helped. In reaching out to a specific university coordinator, only those through a personal connection agreed to share our advert.
Reaching out to professors we didn’t know seemed risky – on one hand, if they were sympathetic to EA causes, they would share the word about Effective Thesis with their researchers. On the other, we could cause some harm if we reached out to academics who weren’t sympathetic to EA and wouldn’t appreciate us influencing their students on their thesis topics. Owing to this, I only reached out to University academics directly less than a handful of times and they were happy to share our message (in one case we received a very negative response).
Using tracking links: One large error was not using tracking links on many of the initial emails or digital flyers. That made measuring the effectiveness of these emails nearly impossible.
Email outreach had higher conversion rates compared to most other channels, however, we eventually decided not to pursue it as:
We didn’t have a CRM and it was extremely time-consuming to research who the right person to reach out to was, personalise emails and follow up.
Tracking is hard. In many cases, people agreed to share the email and confirmed they would, but we didn’t find out whether that happened. We also don’t know how many students ended up seeing our emails.
We received a few emails (<10%) from research coordinators that they would not share what Effective Thesis does with students, as it conflicts with their faculties’ responsibilities.
Recommendations
I am fairly confident about recommending email outreach (whether it’s cold outreach or newsletter marketing) for most orgs who want their audience to take some sort of action. For us, cold emails took a fair amount of time, however, were more effective in terms of conversion rates so I would recommend automating as much of this as possible.
As for reaching out to influencers, I’d highly recommend using an agency instead if you have the funds.
Social media—Twitter and Facebook (ads)
Why these seemed promising
Academic Twitter is large, active and not a very crowded space for advertisements. Twitter also has features that allow you to target people based on who they follow. Through our audience research, we learned more about the influencers some of our audience is interested in.
Facebook also has an extremely large reach and was a common answer to the ‘most used social media platform’ question in our initial audience research.
Facebook and Twitter are cheaper than other large platforms. For context, here’s the average cost per click for each of the four major social media advertising platforms:
Twitter: $0.38
Facebook: $0.97
Instagram: $3.56
LinkedIn: $5.26
Process
We created campaigns based on location, targeting, and A/B tested with design and messaging. We focused primarily on promoting our primary service coaching, along with the more recently launched Future Researchers Newsletter.
Results
We ran short campaigns (between 2 weeks − 2 months), and had a small budget to run these. We spent a total of $2370 on social media ads that resulted in 132 conversions, making the cost per acquisition ~$18. The results below are from long-running ads (not including promotion of a post or tweet). These weren’t strong results overall, looking at conversion rates alone compared to other campaigns, especially for coaching. I think this might have gone differently if we used a different approach or service to promote.
| Campaign | Average CPC | Average cost per acquisition | Average conversion rate |
| Facebook—newsletter | $0.11 | $8 | 4.36% |
| Facebook—coaching | $0.48 | $17 | 0.29% |
| Twitter—coaching | $0.32 | $31.70 | 1.30% |
| Twitter—PhD coaching | $0.17 | $31.25 | 0.53% |
Learnings
On both Twitter and Facebook, when not defining a specific location, the campaigns would automatically reach more people in APAC than Europe and USA.
We found the campaigns we ran for longer periods (1 month+) performed significantly better than shorter campaigns.
The results from these campaigns did not seem promising enough to promote coaching—we observed that through ads, people were hesitant to sign up for coaching but more inclined to sign up for our newsletter or something low-effort, since coaching requires them to research and fill out a form. This pattern emerged across different channels, and we decided instead to take a longer approach to familiarise students with Effective Thesis and launched the Future Researchers (now Topic Discovery) Newsletter.
Recommendations
I am not confident that Facebook/Twitter ads are the best channels for reaching a ‘quality’ audience for a high-fidelity service (like coaching) since the space is extremely crowded and attention spans are short. These channels seem better for quick sign-up services (like a newsletter) or awareness if you have a large budget that exists to share valuable information and encourage nudges to engage with other parts of your org.
Additionally, if you’re running ads on these channels, you should probably invest a little time in organically growing your account.
GoogleAds Grant
Why this seemed promising
GoogleAds has extensive reach, is intent-based and uses keyword targeting, which can be especially useful if the keyword research stage produces promising results. Additionally, for nonprofits this is especially cost-effective since Google awards a $10,000 USD/month grant to run these ads.
Read: Digital marketing is under-utilized in EA
Process
Since setting up a GoogleAds grant, we promoted our home page, research directions, advice, funding database and potential supervisors database.
Since (other than our research directions) these were not our primary services, I decided to promote only coaching and our opportunities newsletter—which seemed most promising as it has the lowest bar to sign up and has the potential for impact through open opportunities.
I generally preferred to use keywords with low to medium competition and high volume for short-tail keywords.
I used short-tail keywords with lots of search volume like ‘thesis topics’ and long-tail keywords like ‘thesis to improve the world’, which had less volume but a narrower intent.
I checked in on my campaigns each week to add or remove keywords that were performing well.
I bid keywords based on results from the keyword planner and initially bid on keywords slightly higher than the average bid. After a week or two of learning, I automatically switched my campaign to ‘Maximise Conversions’.
Based on the results, I started to exclude popular keywords related to keywords like ‘thesis writing service’, which were automatically added by Google but were not intent-relevant. I also excluded locations where poor-quality applications of our services were coming from.
Results
We received a total of 921 conversions in a year (officially for 9-10 months with the new setup, see process above) of running GoogleAds, and utilised 88% of our grant spend overall. A significant portion of our total conversions came from GoogleAds from all outreach efforts. The lowest conversions from these ads, as previously expected, was applying for coaching. We received the highest conversion rate and highest number of people who signed up for our newsletter, however, the impact and quality of those sign-ups is uncertain.
| Campaign | CTR | Conversion rate |
| Coaching | 10.68% | 0.30% |
| Newsletter | 8.57% | 5.90% |
| Content | 6.43% | 3.71% |
Learnings
Running worldwide campaigns (not targeted to a specific location) may pose the risk of ‘low quality’ customers, the more specific the targeting, mixed in with high volume and high intent keywords, the better. I had to exclude keywords and locations where low-quality applications were coming from.
GoogleAds comes with limitations that were not extremely clear when setting campaigns up. There are several restrictions on bidding, ad type and daily spends.
Recommendation: GoogleAds Grants is a good experiment to invest in for 3-6 months, and should give you a clear picture of whether the results look promising. However, it might not work for all types of campaigns or products/services that require action that has a higher bar than signing up. Test for quick signing up and content (where you have built a good pipeline for further engagement).
Online sessions with EA groups—introduction to impactful research
We ran ~1 hour sessions on ‘Changing the world with your research’ which consisted of introducing EA student group members to Effective Thesis and possible ways of thinking about using your thesis as a path to impact.
Why this seemed promising
Students attending these virtual sessions would likely be interested in or familiar with our recommended research directions. We also presumed that reaching out to EA group leaders and setting up these sessions would contribute to a part of their outreach efforts, as a way to encourage students to do applied work in EA cause areas.
Process
This part was simple. I reached out to EA group leaders, met to find out about when students typically start thinking about their thesis and coordinated a date and time. Initially, the process was rough but we created a structured communications pipeline that allowed us to track applications better and encourage attendees to take action.
Results
We ran 8 virtual sessions, with a total of ~112 attendees. There is a non-insignificant margin of error on the number of attendees (likely -/+10-15%) as we ran a portion of these sessions with students attending group meetings together.
Out of the 112 attendees, we received ~33 sign ups for at least one service, making the conversion rate ~30%, the highest of any other outreach activity. We also did not track initial events diligently, so this rate is likely higher.
Learnings
Have a robust communications plan around events. We learned it is important to have everyone register before the event, create a sharable digital poster, content templates for group leaders, reminder emails, calendar invites and clear next steps sent soon after the event (within 24 hours).
We did not receive many sign-ups from the 2 events which were scheduled around student exams; we also did not share emails for next steps directly with students for these 2 events which made it difficult to track whether attendees signed up.
Influencer Marketing
Why this seemed promising
High trust: Similar to email outreach for university coordinators, influencers with relevant, high-quality content and an engaged audience can reach their followers through one campaign for which trust is already built.
Content for most formats (videos, blogs and podcasts) usually helps SEO through backlinking.
We received a shoutout from a video by a YouTube content creator with high relevance to our content and received a steady, though small stream of applications from this video over 7-8 months.
Authenticity and relatability: By integrating brands into their content organically, influencer marketing allows for a more authentic and relatable promotion. This approach helps overcome ad fatigue to some extent.
Market insights and feedback: collaborating with influencers can provide valuable insights into consumer preferences, trends, and feedback on your product or service. This feedback loop can help refine your campaigns and overall strategy.
Results
We targeted small-medium sized influencers (3-100k followers) on Twitter, along with a podcast and ran a total of 5 paid campaigns. Our total spend was ~$900, which resulted in a cost per acquisition of $17.3 and a conversion rate between 2-6%.
There is a significant error of margin on these results as 2 out of 5 campaigns ran on Twitter which had a glitch that showed inaccurate numbers for these campaigns.
Learnings
Influencer agencies: Without using an agency that could connect us to influencers directly, this was an extremely long and arduous process. I would recommend looking into this if influencers are likely to be a part of your marketing strategy.
Progressive overload: Running longer/bigger subsequent campaigns with the same influencers resulted in more traffic, clicks and conversions.
Micro-influencers might be happy to promote your organisation for cross-promotion. Look for relevant influencers (for us, these were students who would share study and research tips). In our case, it usually took ~3 months to get a response to our original email, so accounting for time while planning this is important.
Sponsorship and collaboration are two different things—we attempted to co-create content with medium-sized influencers which led to very long feedback loops and resulted in the collaboration never happening.
Channels we did not select that seemed promising
SEO
SEO (search engine optimization) is the process of optimising a website and its content to improve its visibility and rankings in search engine results pages (SERPs). The goal of SEO is to increase organic traffic to a website by making it more relevant and authoritative for search engines. This is usually done by creating relevant content, good UI/UX (including page speed) and link popularity which tells search engines how many other relevant websites with higher domain authority have linked to this one.
Why it’s promising
Increase in organic traffic: an increase in organic traffic has a higher probability of converting, as users tend to find you vs you reaching them.
Long-term visibility: unlike paid ads which stop being visible once the budget runs out, SEO is a more sustainable solution that produces results after initially working on optimization.
Cost Effective: SEO actually adds value to your stakeholders and brand, even if initial setup and content strategy costs more, it’s likely much more cost-effective in the long-term.
At Effective Thesis, we observed that users who were visiting our website were mostly one of the top sources for those who spent longest on the website. This indicated that users who visited our website from other web pages found our content relevant.
Why we (mostly) decided against it
Note: we did invest a little time in improving the UI/UX of our website, creating better meta-descriptions for pages that did rank. Like our thesis topics page, which now ranks #3 on Google search (as of July 2023).
Highly competitive keywords: our relevant keywords were mostly rated against universities and content-heavy educational websites. We didn’t think there was a high chance of ranking above more resource-intensive sites.
Our domain authority (DA) was already too low: our DA score was already pretty low, and ranking for highly competitive keywords needed significantly more resources and a larger focus on content.
Didn’t align with short-term goals.
YouTube/video ads
I have low confidence that it was a good strategy to not have YouTube ad campaigns even if we didn’t have our own YouTube channel or original video content.
Why it’s promising
Very refined targeting
Higher chance of brand recall through video
Students who may not be active on social media channels we target will still likely use YouTube
There is lot of educational content on YouTube
Video ads can be more engaging than text/image-only ads
Why we decided against it
Limited time and resources
Lack of presence or video content resources on YouTube
Reddit
Why it’s promising
Target group: Advertising on Reddit could be valuable due to subreddits related to Academia that are closely related to the kind of services we offer.
Active engagement: Users on these subreddits are generally quite active and engage with the platform for longer periods of time. The average engagement time for the US is over 8 minutes, and being a primarily text-based platform (though gifs and images are common), it is better placed than Instagram or Facebook for reading.
Cost-effective: Reddit ads are some of the cheapest compared to other large platforms.
Why we decided against it
Reddit users are known to dislike ads—I’m not 100% confident about this, but I have seen across many subreddits that self-promotion (or any kind of promotion) is banned.
Limited resources—in order to advertise on Reddit, it seemed important to also grow an organic following by answering questions acting as a thesis or academic expert. This takes up significantly more time than simply advertising and would require coaches input who were otherwise engaged in 1:1 coaching, which seemed more important.
Brand dilution—I was unsure how the brand’s reputation might come across by advertising on reddit, since it is a more casual platform with much less regulation than other platforms.
PR (Public Relations)
Why it’s promising
Builds credibility and awareness: PR efforts could help establish us as a trusted authority in academia by securing media coverage, guest speaking opportunities, or byline articles in relevant publications.
Targeted outreach: By targeting specific educational publications, or online platforms that cater to students or research coordinators, we may reach a broader audience to be aware of Effective Thesis.
Human storytelling: this gives an opportunity to share the stories and experiences of students who may have benefited from coaching or other services that are more difficult to capture while advertising on a limited budget.
Why we decided against it
Note: Consequently, I am of the opinion that we should have spent more time building out a PR strategy as high credibility is important when reaching out to coordinators within academia and building awareness.
Initial focus on students in the EA community: we were initially focused on reaching EA channels that seemed more accessible.
We didn’t have events or our own research to share, so producing a continuous stream of content for PR might have had to be paid.
We did a cheap test that didn’t go well: We ran an article in an online educational news channel and though we received a lot of traffic, the conversion rate was low and the CPA was the highest out of all the channels we tried.
Uncertain results: PR outcomes can be more difficult to measure and quantify compared to advertising or marketing campaigns. While media coverage and exposure are valuable, it can be challenging to directly attribute specific outcomes, such as an increase in coaching, solely to PR efforts.
Resource intensive: PR requires time, effort, and sometimes financial resources to develop compelling stories, build relationships with media outlets, and pitch ideas.
LinkedIn (advertising)
Why it’s promising
Context relevance. Though we can find our target group on other platforms, LinkedIn shares more relevance with regard to career, education and impact than others. When browsing LinkedIn, users seek something more valuable or purpose-driven to them than they would on, say, Instagram.
Targeting specific Universities was possible. We could target Masters, Undergrad and PhD students specifically according to their programs.
Higher conversion rates: From having previously run ads on LinkedIn, I found that conversion rates were better than on Facebook and Twitter.
Why we decided against it
Note: if given more time or we were running more virtual events, I would have likely run ads on LinkedIn as they have performed much better for similar campaigns I have run in the past.
LinkedIn ads are more expensive than Facebook or Twitter, though having run ads previously for larger organisations, they can be more effective.
Our page on LinkedIn wasn’t yet as active or built out as Facebook or Twitter—we felt we needed more visibility and content on our LinkedIn before advertising.
Additional learnings from outreach and marketing
Google Analytics
Analytics plays a critical part in understanding user behaviour. Google Analytics delivers insights not only into these behaviours but also into website performance and audience characteristics. It can be somewhat overwhelming to use at first, but simple enough to learn. I highly recommend Analytics Mania to learn more about GA4 if you’re new to it.
Our reports usually contained what our most popular pages were, where a lot of the users were coming from (paid ads, social media, emails, referrals etc.), how long they were engaging with the website, and which one of them was likely to sign up for a service.
Some ways to use Google Analytics 4 (GA4) well:
Set up tracking for all channels: tracking involves the use of tracking codes, pixels or other technologies to gather information about user actions, such as page views, clicks, conversions, or interactions with specific elements on a website.
I usually use this simple tool for UTM tracking.Start with simple metrics: views, users, sources of traffic, engagement time and popular pages is a good place to start. Depending on your goals and website, you can build on these and create user flows.
Make sure your conversions are set up: If you’re using a tool for people to submit their information, like signing up for a newsletter, ensure it’s being accounted for by doing a live test in Google Tag Manager. We initially faced difficulties setting up conversions, as we used Jotform and GA4 wouldn’t count it as a conversion directly. Using Google Tag Manager, we set up conversions to be counted when people submitted a form that would redirect them to a ‘Thank you’ page.
Using exploration: Using free forms in Google Analytics, we were able to find out how many people spent over 5 and 15 minutes on our website as we consider them to be highly engaged. We were also able to track the flow of user behaviour using path exploration, i.e. which pages were visited in which order once people visited a specific page on the site. This helped us to better develop the website to ‘nudge’ users in the right direction. Eg. placing a ‘where next’ section after people read our research profiles.
Block your organisation’s traffic: To get more accurate results, you can block internal traffic on analytics by blocking your team’s IP address on GA4 or using this add-on.
Results from Google Analytics
Through the process of outreach, we found that traffic (counted by new users visiting the website) increased by 475%(~5.75x) from the previous year and conversions increased by 235%(~3.5x) compared to last year.
If we exclude paid search traffic, which can largely be seen as counting vanity metrics, the increase from the previous year (2021-2022) is approximately 119% (2.2x).
We also found that though the most amount of traffic was coming through paid search, it wasn’t as effective as referrals (other trusted websites) or emails (like group newsletters) at getting people to sign up for our services. This is likely because many of these referrals and emails were EA-related, where we generally find higher levels of engagement due to value alignment and context relevance.
| Channel | Users | Conversion rate |
| Paid Search | 81322 | 0.7574826 |
| Direct | 12333 | 2.156815049 |
| Organic Search | 10081 | 2.63862712 |
| Unassigned | 4462 | 2.39802779 |
| Referral | 2327 | 8.122045552 |
| Organic Social | 1993 | 2.358253889 |
1051 | 7.040913416 | |
| Organic Video | 98 | 10.20408163 |
| Display | 12 | 16.66666667 |
Creating our logo
We learned to hone our messaging and our audience through this process. Earlier on, our logo was closely EA-related (lightbulb + graduation hat). We didn’t think it was needed any longer, especially now that we were reaching out to students unfamiliar with EA. It was extremely helpful to find logo design long and challenging, as questions about what we wanted to be perceived as, also helped us ask questions about who we are.
The process of creating a logo for Effective Thesis was, again, long and we tried different things:
Having a contest for our logo: This didn’t work. I received feedback that designers didn’t appreciate taking time to submit their work for relatively small prizes (unlike writing submissions)
Worked with a design agency: with multiple rounds of feedback and short timelines (we wanted to launch our logo at an EA conference), we realised we didn’t really like the work and more revisions were very expensive.
Finally, we worked with a freelance designer whose work we liked, and we would also get revisions and feedback incorporated fairly quickly.
Why did it take us so many tries to figure out the ‘right’ logo?
We hadn’t defined our value proposition or target group sharply yet. Though I still don’t think our logo is perfect, since we’ve somewhat iterated our strategy a few times in the past year; but the process of having a logo designed helped us dive deeper into our value proposition, with some prompts like:
What do we want our logo to convey?
How important is our logo vs our name (which is self-explanatory)?
What are we trying to achieve with our logo? (memorability, legitimacy etc.)
What are some key emotions, thoughts or associations we want it to invoke?
What are some key emotions we already invoke? How can we tie that in with the aspirational brand we want to be?
If you have been through the process of logo design and found it frustrating, it’s likely you need to work through initial branding questions more.
Switching from promoting coaching to launching the Future Researchers Newsletter (now called the Topic Discovery digest)
We saw from running ads that there was higher resistance to people signing up for our primary service i.e. coaching than there was to newsletters, workshops and virtual information sessions. In theory, that made a lot of sense—applying for coaching is a more cognitively demanding and time-consuming process. It’s also likely that a student won’t sign up if they’re not already aware of any of our research directions and want to go through the website first. Additionally, we were likely reaching students even when they weren’t thinking about their thesis; we needed an introduction or ‘warm up’ to Effective Thesis while staying salient.
We decided to bridge that gap by directing people to the Future Researchers Newsletter for students to sign up and receive thesis topic inspiration, advice and tools they might find useful for academic (or other) work.
Preliminary Results from the Future Researchers newsletter
Level of engagement of subscribers | Open Rate | Click Rate |
No to mild engagement with EA | 41% | 5.10% |
Considerably engaged with EA | 75% | 16% |
We launched this in April, with subscribers receiving 4 weekly emails to ‘onboard’ them and ~2 newsletters a month. At this stage, the newsletter is still new and we plan to iterate it further, but it has received over 400 sign-ups since we launched and seems like a more promising path for people to learn about our research directions and eventually sign up for coaching. (May #1 edition had an error and many emails were sent to spam)
Newsletter edition | Open Rate | Click Rate |
June #1 | 42.7 | 8.5 |
May #2 | 53.2 | 10.5 |
May #1 | 17.2 | 4.6 |
April | 42.5 | 4.6 |
Average | 38.9 | 7.05 |
Remaining uncertainties to consider
A crowded space—are many orgs trying to reach the same people?
Confidence level: high
Currently, it seems like many EA organisations are trying to reach a similar audience → talented students in prestigious (top ~100) universities. That may be oversimplifying for the differences between causes, timing or objectives but it’s likely that there is a large overlap in marketing efforts.
Some uncertainties I have are around the counterfactual impact of one org reaching these students than the other and that it doesn’t seem very efficient to be competing. It also doesn’t seem like an answer as simple as – the org with the most resources and historically positive impact is the best choice to promote their service.
Some ways to solve for this overlap can be to:
Coordinate marketing strategies
Clarify specialisation and differentiation
Encourage cross-promotion
Evaluate the impact of outreach/marketing publicly (eg. the forum)
It’s likely that there will be more EA orgs reaching out to the same people, and it’s probably not bad if people are more aware of them, but for smaller budgets and more efficiency, coordination and cross-promotion seem like good things to do.
Defining audiences through targeted metrics
Confidence level: low
While it’s a good bet that talented students from Top 100 Universities around the world are more likely to become impactful researchers, I’m concerned that only using these (and related) metrics to reach out to students from prestigious programs is not the most impactful. We may miss reaching out to those students who need the most support in writing their thesis on a pressing problem. On the other hand, I agree that if we’re optimising for the most talented students, the probability of finding them at these universities is quite high.
As a way to diversify outreach, we ran virtual sessions across newer and LMIC groups for visibility, though this was done on a small scale and thinking about this more intentionally could be valuable.
Quality vs. quantity: is a broad funnel the best thing?
Confidence level: medium
Initially, it made sense to broaden our outreach across locations, degree levels and channels but by applying the power law, we have enough information from previous outreach that allows us to reach a smaller, better-suited audience. I’m uncertain whether we will saturate this space if we optimise for a narrow pipeline vs a broad funnel over the long term.
However, if we were to look at our previous high-quality applications, most of them came from within EA and through word of mouth, which has diminishing returns since they are more likely to work on EA causes anyway and our counterfactual impact decreases.
My best guess is to set up channels (like GoogleAds for us, Influencer outreach) that somewhat automate the broad funnel and focus the rest of the outreach capacity on more targeted, narrow audiences.
Hitting the ground running
Confidence level: medium
For a small org with YoY funding rounds, you’re encouraged to hit the ground running by starting marketing activities right away. This can be both good and bad. In retrospect, I might have spent more time carefully scoping our previous data, reaching out to coordinators at Universities and selecting only 2 to 3 most impactful channels.
I’m still uncertain as I see some benefits to hitting the ground running for others and myself:
Short-term impact: By quickly immersing yourself in marketing activities, you have the potential to make an immediate positive impact on the organisation.
Faster learning curve: Engaging from the start allows you to learn the intricacies of the job and the organisation more rapidly. It enables you to gain hands-on experience and understand the nuances of your target audience and industry.
‘Visible’ impact: with limited funding and needing to see quick results, the right experiments lead to more visible impact as you have data you can showcase. Scoping work is not as ‘visibly’ impactful in the short-term.
Building relationships: Being proactive and diving into marketing early on can help you establish relationships with colleagues, stakeholders, and clients. This can lead to better collaboration, increased trust, and a stronger support network as you progress in your role.
Time-sensitive opportunities: Taking immediate action allows you to capitalise on time-sensitive opportunities, such as upcoming events or seasonal campaigns.
Even with these benefits, there are some considerations for caution:
Understanding context: While it’s essential to be proactive, it’s equally crucial to grasp the organisation’s theory of change and mission before making significant marketing decisions. Taking time to assess the existing strategies, gather insights, and aligning with the orgs goals ensures a more informed and effective approach.
Building thoughtful relationships: While building relationships is important and is probably easier to do within the EA community, it’s essential to approach it with care. Rushing into collaborations without understanding the dynamics can be ineffective.
Balancing longterm strategy and short-term wins: While hitting the ground running can yield quick wins, it’s crucial to maintain a balance between short-term tactics and long-term strategic planning. Focusing solely on immediate results may sacrifice the development of a robust marketing strategy that aligns with long-term objectives.
I think people starting out new outreach should invest more time scoping the landscape, and if it’s also important to start executing quickly, start small with the most promising (and previously proven) strategies if possible.
Organic social media effectiveness
Confidence level: high
I’ll keep this brief—I’m unconvinced by organic social media (not ads or collaborations) being an effective tool for outreach, and recommend spending 5-10% of an outreach manager’s time or budget to invest in building social media unless a large output of the org rests on building and sharing content. It still makes sense to have a presence on social media to build legitimacy and for SEO, but it’s very unlikely much of your impact comes from posting on social media alone.
- ^
A study by Fogel and Nut (2017) found that clarity of mission positively influences an organization’s reputation and funding outcomes
- ^
Wymer, Walter & Gross, Hellen & Helmig, Bernd. (2015). Nonprofit Brand Strength: What Is It? How Is It Measured? What Are Its Outcomes?. VOLUNTAS: International Journal of Voluntary and Nonprofit Organizations. 27. 10.1007/s11266-015-9641-8.
Thank you for sharing your outreach strategy at ET with such detail! I bookmarked this post earlier and finally found a good moment to read it :)
For EA Finland my main takeaways from the post are:
- We should take some time to develop/unify the branding of EA Finland further and especially our tone of voice.
- Identify more broadly which our most promising target audiences could be and organise discussions/ 1-1s with people from the reference groups to get insights to qhat kind of communication appeals to them and if they are interested in EA in the first place. Now founder effect and existing networks are impacting our community building efforts quite a lot.
- Create Google Analytics for altruismi.fi so we can develop our website based on more detailed data.
- I believe your results from the messaging strategy are interesting and we could try to emphasise the inspirational part as well in our communications. (I acknowledge you also had a section on diversifying outreach strategies between EA groups but this is quite a general approach)
- I had some initial ideas about creating an influencer-focused effective giving campaign. It was good to get some more information on how time consuming it can be and your experience updated me against its cost-effectiveness.