While part of the impetus for the changes is the “fake news” scandal, much of it has to do with the rise of “clickbait” publishers which profit from low-quality, sensational or salacious stories which get high visibility and are highly addictive but do not result in an enjoyable or informative experience overall. The battle between content discovery platforms (i.e. Facebook) and publishers who try to game the rules is currently playing out on Facebook just as it did on Google in the early 2010s.
Facebook’s 2017 – 2018 changes can be summarized as:
The visibility of activity from friends and a user’s community is prioritized relative to online publishers. Also, content which sparks conversation and reactions from friends is prioritized.
“High quality” stories are prioritized over “clickbait.” Quality stories include longer videos, comments, web pages with substantive content and other criteria.
Many publishers who formerly received a large percentage of traffic from Facebook have seen significant drops in traffic. FEE was especially affected as we received nearly 43% of our traffic from Facebook in early 2017 and subsequently saw that drop to 23%.
Are these changes bad for publishers?
While sites like FEE.org have been somewhat unfairly punished as a publisher of “quality” content rather than “clickbait,” it is overly simplistic to see Facebook as the “bad guy” taking away traffic from publishers. Facebook has made a business decision to prioritize content from friends and family. They want to be a social network, not a news discovery platform. It is up to the market to decide whether Facebook becomes more or less valuable as a platform as a result.
Furthermore, in general, publishers of “quality” content will tend to compete in a meritocratic fashion with publishers for users’ organic reach. There are a lot of disclaimers here, but people generally accept that Google’s algorithm is mostly fair (and I believe it is) despite many claims of bias and ruined business models when Google had to adjust to “black-hat” SEO tactics. I believe that Facebook News Feed algorithms will probably come to be seen the same away some years from now.
The impact of social media on FEE.org
FEE.org saw substantial and highly volatile growth in social traffic starting in early 2016. This growth continued until late 2017 when it fell to 2015 levels. Facebook’s algorithm seems to have been responsible for nearly doubling social traffic, before falling back to the 2015 baseline. In other words, Facebook giveth and Facebook taketh away. FEE.org’s success with Facebook was largely due to its popularity with younger audiences that tended to push stories to go viral by repeated sharing. However, the “quality” of this engagement was low: for example, FEE.org’s bounce rate for Facebook referrals was 90%. As of October 2017, the bounce rate has been under 4%. It’s likely that many people who shared or liked our stories did not actually read them.
3-year sessions vs bounce rate for Facebook.com referrals:
(Note: The changes are not as stark as this graph shows due to measurement changes but the overall trend is valid.)
If we look at sessions which came from Facebook.com, we saw fee.org sessions grow from 159K in December 2015 to a peak of 386K in February 2016, before falling back under 100K in September 2017. We speculate that this drop is in part because Facebook lowered the value of sharing posts relative to the value of genuine reactions (likes/comments). (Note: Facebook implements strong and ever-adapting anti-cheat measures to ensure the integrity of “genuine” reactions.)
FEE.org saw a long-term lift of about 300K sessions per month from all sources from 2015 – 2018 — such as external referrals, organic search, and email.
FEE.org traffic follows a strong seasonal pattern (much of our traffic is from students performing research during the school year).
Illustration: 3-year performance — social versus all traffic:
How FEE is Responding to Facebook Rule Changes
The loss of FEE’s #1 traffic source requires adapting our strategy to maintain our business goals. Our primary means of doing so is to replace lost organic reach with paid reach (ads). The team at FEE has learned to do this with some effectiveness, despite ongoing algorithm changes:
Organic vs Paid Reach: January 2017 – April, 2018:
We believe that paid reach should have different business goals than organic reach. In other words, we do not believe in merely paying to replace organic readership with paid readership. This is both expensive and does not generate returning traffic. Instead, FEE’s marketing team had focused our paid reach on specific business goals that serve our organizational priorities, such as email leads, event registrations, market research, and new audience outreach.
Finally, while Facebook has moved away from a focus on publishers, there are other platforms which are focused on story discovery. FEE has experimented with Apple News, Quora, Flipboard, Medium, Reddit and other platforms which may surpass Facebook as content discovery mechanisms.
Reacting to Facebook Ad Strategy Changes
We’ve also had to adapt to ongoing changes in Facebook’s ad platform in response to events such as ads by alt-right extremists, the Cambridge Analytica leak, and intense competition from other businesses who were “exiled” from organic reach by algorithm changes in 2018.
We’ve noticed a significant change in the effectiveness of our paid lead generation ads. We ran the same ad for our Essential Guide to Health Care Reform in November and April and had wildly different results. In November 2017, we paid on average $0.72 per email address. In April 2018, the cost per email rose to $5.07.
Our current strategies include:
Experimenting with landing page view ads instead of lead gen ads.
Relying more on popups than Facebook ads to collect email addresses.
Boosting content that gets great organic reach (>80% usual reach) to try to “ride the wave” of the organic traffic.
Focusing on Quality of Interactions vs Quantity
As explained above, in the last two years, both Facebook and Google have made efforts to limit content reach to the subset of users with “genuine” engagement or interest. The downside for us is that fewer people see our content. The upside is that the people who do see our content, are more likely to pay attention to it. Furthermore, we have a chance to be rewarded for more meaningful and useful content.
To measure meaningful engagement, we will try to prioritize meaningful engagement: active time on page (provided by Parse.ly analytics), longer video views, email captures on FEE.org, Facebook Ads, and landing pages, and Facebook/YouTube subscriptions. We will also try to improve the quality of our content by sharing fewer stories from other sites on FEE.org and writing more in-depth analysis and practical career/life guidance.
The other day I mentioned on LinkedIn that I was looking for a freelancer web developer. A USA-based agency messaged me to express interest. Their rate — $150/hour. This a common rate for an experienced US-based web developer. The project I need help with has a budget of about $1000 per month, so this rate would give me under six hours of work, once rounding and communications overhead (or project manager) is factored in. I ended up going to UpWork.com and hiring two developers: one from Ukraine for $20/hour, and one from India for $10 per hour. The Indian developer will do close to 100 hours of work for the same cost. So that’s the basic case for sourcing your own freelancers. The details are considerably more complicated: sometimes a $150/hour developer is a better value than a $10/hour one: a good programmer can be far more productive than an average one, and a bad one will just waste your time and money. The trouble is that if you don’t know what you’re doing, you may end up paying high-end prices for shoddy or useless work. Here is how I approach hiring freelancers:
Good people are hard to find
Whether you’re hiring app developers, artists, video editors, or a project manager, there are usually huge differences between the productivity and quality of freelancers. The average quality of the people who will apply or be recommended for your freelance job posts will be quite low. Most highly-skilled engineers are not interested in freelancing or busy with existing projects, while the bad apples bounce from one project to another.
UpWork makes an effort to recommend the most qualified people for the job, but the candidates often engage in deceptive tactics boost their ratings. Freelancers will steal the information from good profiles, lie about their skills and experience, promise to take on projects they don’t have skills for, or hold your work hostage in exchange for good ratings (yes, this has happened to me).
If you add more rigorous qualifications or required experience to your post, they will use them against you: the frauds will promise to be experts, but honest freelancers (people are rarely a perfect fit) will move on. Many will try to circumvent UpWork: they will email and call you, your boss, your coworkers, and anyone else associated with you on your LinkedIn profile.
It’s possible to screen out the worst applicants (see below), but the only way to identify the 1% who are affordable and qualified is to test them.
How I do freelancer screening
A typical job, assuming a decent description will get about 100 applicants. If you get significantly less than that, your job needs a more detailed description of your project.
You can screen out 90% of candidates by filtering for a 90% job success rate, 1000+ hours worked, good reviews, a credible portfolio, and 10+ high test scores. However, this will usually raise the hourly rate.
1 Screening questions:
I typically do minimal profile screening simply because good profiles are easy to fake. Many individual profiles are secretly teams or agencies, and 1000+ hours on UpWork doesn’t mean anything when they put a brand new guy on your job. I also just don’t have time to screen 100+ applications for each job:
My screening process has three parts:
I ask simple questions which require job-specific experience to answer.
For a web designer: What’s your favorite CSS3 feature?
WordPress e-commerce developer: What version of WooCommerce have you worked with? (Candidates who are not paying attention usually give a WordPress version instead.)
Graphic artist: What navigation pattern would you use for this design?
2 Setup process: With developers, I ask them to follow a specific process which includes
(1) checkout out the code,
(2) messaging me on Slack for a test DB,
(3) sending me a screenshot of their configured dev environment
3 Trial task
(4) suggesting a trial task from the issue list.
90% of applicants will not complete this process correctly. Some tell me “I don’t work for free.” I respect this, and instead ask how they would solve a specific task.
For a complex software development job, 100 applicants followed by a 90% rejection rate still leaves 10 candidates. Each of these 10 receives a trial task. I try to simple, self-contained, well-described tasks with a $100–200 budget. Typically, about 5 candidates will complete the task, and 1 or 2 will deliver quality work. Sometimes none will, and I have to find more applicants. Getting through this process can be frustrating: you will be charged for work that isn’t delivered or must be discarded, will terminate nice guys who can’t deliver on time, and probably find that the best candidate(s) are not the favorites you bet on when you started.
Things are easier for non-developer positions. Artists and graphic designers are easier to screen, but tend to be more flaky with their time estimates. Being patient and understanding, yet severe and final with your judgment is essential in this process.
Establish a good onboarding process
Early in my freelancer management experience, I struggled with managing the cost of onboarding. Some agencies charge thousands of dollars to setup a dev environment and learn your systems. Since then, I’ve developed guides for new developers which communicate three things: expectations, system design/architecture, and development process (“definition of done”). This allows qualified engineers to complete setup without my help — and disqualifies those who struggle.
Use collaboration tools to coordinate distributed teams
It’s important to establish a mature development environment before you begin work. The development environment allows you to clearly communicate requirements and task assignments to freelancers, allows you to monitor the progress of code and functionality, and allows you to perform releases to end users in a deliberate manner.
Issue tracking tools for developers and stakeholders: JIRA or Bitbucket issues
Source code control: BitBucket or GitHub
Continuous integration: I use TeamCity or a PHP deploy script
Development environment: I use AWS-hosted servers
Collaboration tools: Slack for chat & Skype for team status calls
Documentation and specifications: Google Docs, JIRA Confluence
Establish working relationships before starting a big project
A good software team is a complex system: it requires a lot of momentum (aka money and time) to start up and keep going. If you’re planning to kick off a big project, it’s essential to get your team lined up first. Even after you complete a screening process and trial tasks, some turnover is to be expected. Various startup challenges will need to be overcome. You need to get momentum going with a small project before you begin one that is essential to your business.
Sometimes your business needs will require a pause in development, so you will have to invent work for your freelancers to keep the team together, lest they drift away, and you have to rebuild a new team all over again.
If this process sounds too risky, stressful, and complex for you, paying $150/hour+ for a reputable US-based developer or professional agency (with a 30% management overhead) is probably a better deal for you. Alternatively, you can skip the headache and hire me to do all of the above 🙂
(Guest post by By Jeffrey A. Tucker, Director of Content for FEE)
Introduction: a bold experiment in reaching mainstream culture
Two years ago, FEE embarked on an experiment born of frustration. FEE’s website was not a product of its own but rather a kind of information board for advertising the institution. It had low performance. Brand recognition of the institution was not increasing.
And yet there was clearly opportunity. When you look at the venues considered to be mainstream distributors of ideas, they all trend toward the progressive and social democratic, i.e., statist. They fill up the smartphone feeds of millennials. They speak to them on all their social media platforms. Their websites enjoy millions of hits a week. They are profoundly affecting culture – not through political activism, policy study, or academic work, but rather through public commentary on the passing scene. They define what is fashionable.
What is preventing the ideas of liberty from entering this space? Truly nothing but talent, cleverness, and dedication. FEE aspired to apply these traits to our work. We pursued dramatic technological, distribution, and content changes designed to enter into the realm of public culture in a way that directly competes with mainstream venues. We deployed a series of objective measurements to assess our progress. We also set out to be adaptive in all these areas so we could continue to achieve this goal as the tools grew and adapted themselves.
The remainder of this post addresses the content aspect of the strategy.
Fixes and Changes
We first set out to perform a number of technical fixes: in-sourcing the website code for full control and fast development, tagging thousands of articles according to topic, cleaning up legacy cruft, adding missing metadata to articles, and so on. It was a huge job.
We set out to stop the traffic leakage we were experiencing (people hitting the site and leaving) with a series of strategies to capture email addresses. We then built up our daily sending list, from 3,000 two years ago to 45,000 today. As time went on, we added browser notifications for new articles, a new web design to enhance site credibility, and infinite scroll on content display.
We then turned to new content itself. As we looked back at FEE material over the decades, we found a contrast with the way it appeared in the 1950s and 1960s. Back then, the material was directed toward a general audience. It did not use in-group language. It didn’t presume that people were already on board with the libertarian vision. But as time marched forward, there was a subtle change. The language became ever more insular and in-group focused, aimed at movement edification rather than culture-wide influence. The editors presumed, probably rightly, that they were speaking to a marginal group about a narrow topic.
It was renewed interesting in growth among new audiences that motivated a move toward achieving a broader reach. Today, the purpose of FEE’s content is to describe and explain current events, history, policy, and social and economic theory in light of the ideas of liberty, as articulated by the liberal tradition and exemplified by FEE’s 70-year history.
The purpose is to expand the network of users and broaden the base of people who are exposed to a liberty perspective. The metric for us is summed up in one word: traffic, which is a fair proxy for audience. Without this essential component, not much else matters. You can have the perfect product, the perfect prose, the best analysis, the most wonderful presentation, the most correct doctrine. But if no one sees it, there is a problem. The element of traffic also intensifies the commitment to quality work. As Peter Drucker said, “What gets measured gets managed.” You have some accountability. You can begin to craft your product in line with consumer preferences, and thereby replicate the essential dynamic and driving force of the market itself. After that, we look at demographics, granulated data about types of content, what they do, and how sticky their traffic is.
For traffic metrics, we depend on Google Analytics for granulated data and Alexa for ordinal ranking of our site relative to others. In two years, we’ve moved from the 130,000th most popular site on the web up to a stable 22,000. Our institutional goal is always up, with the hope that we will eventually stabilize in the range of 1,000. We have no estimate for when or even if this will happen.
Growth depends on content sharing – not just a first-round of readership but a second, third, and fourth. The content has to spread, not hit the wall of in-group consumption. Why do people share? There are many reasons. Sometimes a hard-edged ideological piece can work for what we call in-reach. But too much ideological vernacular can also discourage sharing, simply because what appears on people’s social timelines becomes a reflection of how they want to present themselves to a wide range of people within their friend networks.
Because social media is a main source of news today, FEE set out to present content that didn’t so much preach the doctrine but illustrate it in a mainstream and credible voice, and provide excitement about how liberal ideas can provide a better and more fulfilling understanding of the world around us. This content should not only feed our fans, but reach outside our existing audience.
Inreach vs. Outreach
For moment-by-moment analytics (and the team truly does follow traffic patterns all day and through the evening), we use Parse.ly, a platform specifically developed for editorial use. It logs on an ongoing basis what percentage of users are new or returning, a metric we use to determine whether a piece has in-reach power or out-reach power.
These two examples come from the day I’m writing this post. A piece on Google’s new translation algorithm posts these results from today.
This is successful outreach. For each new user, we try our best to harvest email, obtain approval for browser notifications, and keep people on the site by pushing more material along the same lines, dropping cookies that are capable of machine learning according to a user’s browsing habits.
On the other hand, here is a much more complex piece on libertarian strategy and the role of ideas – a piece we had intentionally decided for in-reach in order to deepen reader’s relationship with the liberal idea. And sure enough, the reader results are very different.
To be sure, every piece we publish starts out with nearly 100% returning visitors. It is not possible to bypass this group but for paid and targeted advertising. Given that people mostly reach our content via social media, it would be expected that the people who see it first are the fan base on FEE (and this fan base has grown by a factor of 10 in two years). We depend on them to share further outside our network and into theirs. But the path the content takes following that initial release depends heavily on the topic and approach we take with the content itself.
FEE, then, faces a dual obligation: reach new people and feed an existing fan base to further inspire them toward a deeper commitment. Each is important. But given the desire to grow our audience, FEE takes seriously the obligation to seek and explain to all willing listeners of goodwill. We adhere to principle but don’t necessarily wear our ideology on our sleeve or throw labels around, any more than The Atlantic explicitly advocates social democratic ideology in its articles. We show more than tell, in a way that reflects confidence. To put it another way, our philosophy is our musical scale but our literary output is our song.
We’ve discerned that the path to success must be discovered day-to-day through trial-and-error by a creative and venturesome team. It’s a matter of balance: energy with dignity, boldness with class, accessibility with substance, always striving for impact, excellence, improvement, and growth, while modeling the spirit of freedom.
Below are rules of the road we’ve established that reflect the mission and spirit of FEE.
Illustrate the social, moral, and practical merit of liberty as a principle of human association, and present this radical idea in a mainstream voice;
Achieve a balance of news, think pieces, long form and short form, inreach and outreach, classics, reprints from partner organizations, history, law, economics, cultural criticism, personal advice, biography, and so on;
Transcend the left-right paradigm, with roots in “mainline” intellectual traditions;
Strive to be engaging and interesting, with a harmony of graphics, title, and content with a premium on good writing and not just on taking the right position;
Contain no profanity and avoid tacky and vulgar expressions and images;
Eschew overly technical jargon or esoteric topics;
Avoid overly inflammatory rhetoric that panders to ideological biases or otherwise deploys capricious anger, ridicule, name-calling, and invective;
Avoid in-group, insular language and buzz phrases that can only be understood by our most learned fans while making new readers feel unwelcome;
Avoid appearing to push vendettas against individuals or groups or to attack the person, as opposed to the person’s ideas;
Avoid anything that smacks of partisan politicking.
Social media provides the most referrals to FEE.org. Among the platforms, Facebook is the referral engine for 60% of traffic. The next highest known source is Twitter with 5%, then follows Reddit, StumbleUpon, HackerNews, LinkedIn, Youtube, and Blogger. Instagram refers no traffic. It might provide some brand recognition value, but the ROI is unknown. The low level of traffic from Twitter is a bit misleading because these are high-level influencers who then post to their own pages and to Facebook itself. So Twitter works more as a spark than a flame.
For Facebook, we use the Instant Articles application, which speeds up viewing on digital devices. We do almost no paid boosting of articles because we’ve found that if an article is going to do well, it does so without boosting, and if it is going to fail, it will fail regardless of boosting. We do pay for targeted impressions of particular content on individuals’ news feeds, based on carefully selected demographics. Using emails drawn from Salesforce data, we are in the position to place content on donor pages and others based on web-viewing habits.
Intriguingly, 30% of our social referrals qualify as dark social, that is via private messages, private groups, private forums, SMS, and so on. This is a powerful source of traffic but it is neither traceable nor influenceable. As for Reddit, we’ve discovered what others have found: there is no viable way to focus on feeding this source, for the platform is extremely averse to perceived gaming. All we can really do is provide a Reddit link at the top of articles and invite readers to use it.
We automate as much of our social as we can. Every new article (we publish eight per weekday and three to five on both Saturday and Sunday) is automatically posted to Facebook and Twitter using the service Zapier. This massively reduces the chance of error. We post every 45 minutes during the day and every two hours following close of business. Other postings are done by hand by a specialist who follows trending topics and posts relevant legacy content. We have not typically reposted material from other places on the web, but have opted for the publishing strategy described below. In addition, we use Facebook to post institutional news and media.
We use two additional publication venues: Medium.com and Flipboard.com. Using both hand and automated tools, we attempt to keep a solid lineup of articles published at those distribution channels. While Flipboard provides direct traffic to FEE.org, Medium does not – that is, articles “live” on Medium. However, it does account for some referrals, and it also increases brand awareness and realizes certain mission goals with very low cost.
Email as a Product
As implausible as it might sound, email remains one of the most valuable digital products, and, hence a solid infrastructure of email contacts is essential. Because we are producing daily content, we put a high premium on the number of people who received daily emails. Two years ago, we sent to 300 but today send to 45,000, in part by defaulting our email signups to become a daily subscription. We’ve used third-party popups via AddThis but our signups increased 5-10 times by creating our own internal version. Of course popups tend to annoy people and, for this reason, nonprofits might try to avoid them. This is a mistake, in our view, for one reason: they work. We need this infrastructure for our operations.
Of course this also places an extra burden on FEE to turn its daily email into a valuable commodity, something not just for promotion but that also provides delight on its own. Subject lines are chosen carefully to be engaging, and they are different each day. Each send – and we send every day at the noon hour – includes a charming and witty moving gif that is related to the article. The idea here is to create a sense of drama for each day: what gif will I get to see today? As a result, though our numbers of gone up dramatically, our open rates remain very steady. It is a product that people consume on a daily basis, thus increasing brand awareness and gratitude that translates to donor support.
Republishing and Author Payments
FEE has attempted to foster a culture of content sharing within the movement generally. We first put all our content in the Creative Commons, choosing the license Attribution 4.0, which allows for any kind of republishing on any basis provided the source is credited. We negotiated a number of agreements with partner organizations to re-publish their material. We estimate the ratio of “original” to “republished” to be around 40-60%. In terms of traffic, we can discern no trends to predict the reach of either type. Much depends on title, image, trending topic, and compelling content.
Our RSS feeds are set to retrieve the full article content so it can be republished on any site in full. In addition, there is a separate feed automatically created for every author on the site (more than 2,000).
We ended author payments – a break with a 70-year practice, so far as we can tell – because we saw no relationship between the quality of submission and financial compensation. As a result of ending all author payments, we lost perhaps 3 of our stable of 75 or so contributors of original content. In contrast, we have added cash prizes to incentivize authors as a way of broadcasting that we do value writing talent. What authors do value is speed of response, carefulness of editing, and a quick turnaround time.
Most outside contributors receive a personalized response within an hour of submission during the work day, and accepted articles appear on the site within 24 hours. This speed and responsiveness is the best way that FEE can show its appreciation to those who choose our venue as their preferred outlet. And here is another case for keeping traffic as high as we can: FEE can get the word out.
The Team and the Division of Labor
The content team is made up of five people currently: editor, managing editor, associate editor, and two content interns. The skill set required: high-level literacy, proofing skills, basic html, speed, low-level image manipulation, facility with the content management system (Umbraco, which is an open-source management system designed for Dot Net), creativity with titles, and a willingness to work all hours including nights, weekends, and holidays since digital media has no hours of operation. We try to maintain a content mix of original and republished material on a full range of topics. We consult with each other throughout the day and otherwise, in person and on Slack, which is FEE’s internal communication system.
Our work flow attempts to stay one day ahead of the publication schedule. All progress is logged on Trello, a collaboration board that allows for attachments and conversations as material moves through the production structure.
We use Feedly for aggregating content for possible republication. This permits us to navigate content quickly, so that we are not wasting time with random web browsing. Everything we publish is flexible within minutes of going live. Once the publication lineup is ready for the next day, all staff are encouraged to work on writing original content.
FEE is committed to retaining its existing user base and growing it in every way possible, as quickly as possible. We want to contribute to making the ideas of freedom familiar and credible for the rising generation, thus fulfilling FEE’s historic mission and serving as a beacon of excellence in digital publishing. We are confident that we can achieve this with a continued outward focus on customer needs, adherence to the metrics as a main indicator of success and/or failure, and an unrelenting willingness to adapt to changing conditions in the world’s fast-moving market for information.
FEE.org gets about 30,000 clicks from our primary Google AdWords account per month. This is not a lot of traffic – about one day of organic visitors – but the vast majority of Google traffic comes from new users, so it is an important source of new users.
The best part is that this traffic is totally free: FEE get’s a $40,000 per month AdWords credit from Google’s Grants program. That’s $480,000 of free advertising per year!
A daily spend limit of 1/30th of your monthly total
In practice, these conditions make it difficult to hit your daily spend limit. Even if you do hit your limit (for example, by targeting hundreds of low-traffic keywords), it can be difficult to send your visitors to fresh content because someone has to keep creating new ads for new articles on your site.
Google AdWords offers an ingenious solution to this challenge: Dynamic Search Ads. This campaign type uses Google Search’s index of your site to automatically create ads matching the keywords that people are searching for. It’s as if someone creates a new ad every time new content is posted on your site! Using Dynamic Search, FEE.org is able to spend about 50% of our $40K budget on ads that send people to the latest FEE.org articles.
As social media has become the primary way for young people to discover brands and ideas, the effectiveness of traditional television/print/radio has fallen. Reaching people interested in your ideas on social media is hard — there are so many competing brands vying for their attention. Remarketing (aka retargeting) to audiences known to have an interest or relationship with you is one of the most effective ways to reach people who are most likely to engage with your brand.
While FEE’s marketing has historically been focused on personal referrals and “organic” growth, in 2016, we decided to focus on reaching new audiences and developing new marketing channels. Key to our plan is reaching new audiences on social media, using some tricks used by top ad agencies that are new to the nonprofit world. Typically, when organizations run ad campaigns – whether web, print, or radio – they target broad demographics. What FEE has been doing with great success is targeting individual people.
If you are interested in reaching young people who care about free markets and liberty, would you consider trading Facebook audiences with FEE? I’m not asking for your audience’s email or any personally identifiable information. Custom audiences are the aggregate collection of social media profiles which can be used to show Facebook posts and ads to individual users based on emails or website activity. This data is created by creating a custom audience in Facebook from an email list and/or adding a tracking code to your site. We would be sharing Custom Audiences as described by Facebook here.
Why would you want to share Facebook audiences?
It doesn’t cost you anything, and your audience is never individually identified.
You can share both email list (FEE has 100K active emails) and web audiences (about 700K Facebook users per month.)
FEE’s mission is ultimately one of marketing. FEE does not do basic research or write academic treatises. We do not contribute to academic journals or conduct economic research. The intellectual foundation in support of a free society and sound economic principles is well established and our partner organizations do a great job of building on it. FEE’s mission is to popularize a set of ideas: the values of a free market and individual liberty. We must make effective marketing a core competency of FEE.
Our marketing strategy has three elements:
Create great content that engages, informs, persuades, and delights users.
Distribute our content using the platforms that young people use.
Accelerate our audience’s engagement with the liberty movement by promoting events, online courses, and program partners.
In order to further this mission, FEE is developing a world-class technology and marketing platform based on the practices and tools used by top marketing agencies:
Inbound marketing: we leverage our vast catalog of articles, books, and media in new formats to draw audiences to our site and create more opportunities to create customers. We have developed the capability for cost-effective social media advertising that pushes our content to audiences proven to best respond to our message.
Marketing automation: using tools such as HubSpot, we are building an automated conversion funnel, which automates personalized messaging workflows. We infer our visitor’s goals based on their behavior, then send them personalized messaging. For example, we capture visitors who are interested in events but have not applied, and keep them up informed about our programs.
Targeted messaging with data science: tools such as Parse.ly analyze all our content as well as the behavior of visitors to FEE.org. We use data science analysis to create a personalized experience for each visitor to FEE.org. Additionally, our editors see real-time feedback on how their content performs with each audience.
The goal of FEE’s marketing efforts is to be the most effective advocate of liberty ideas in the world, offer a compelling alternative to anti-market orthodoxy, and develop effective pathways to a deeper and life long commitment to the liberty movement.
A key part of FEE’s advertising strategy is retargeting our customers across multiple channels. Retargeting, also known as remarketing or behavioral targeting, is the practice in online advertising of showing ads to people based on their previous online activity. Here is how we do it:
1. Define Custom Audience
FEE’s retargeting is based on three data sources:
Website visits to specific domains or web pages. For this example, we are tracking visits to http://www.feecon.org/ via a Facebook Pixel tracking cookie. Some of our website audience originates on other organizations’ websites and are anonymously shared with us.
Membership in email lists, including FEE Daily or audiences shared with us by partners. (Facebook allows audience sharing without sharing any individual email addresses.)
Online and offline transactions. If you purchase something from our store, register for an event, or make a donation, you’ll eventually be imported as a Custom Audience for retargeting.
We’ve created a custom audience of FEEcon.org visitors, but we don’t necessarily want every visitor in the audience to see our ads. In this step, we’ll exclude people who have already registered for FEEcon, and set an age range so we can create different ad sets based on age.
Below, we’ve included all FEE.org visitors, and excluded everyone who’s already registered and does not match the specified age range.
3. Use Saved Audience in Ad Set
Now that we have the target audience defined, we can use it in ads sets. I suggest that you create audiences before you create ads. You don’t want to spent a lot of time creating an ad only to learn that Facebook (or Twitter, etc) either cannot target the audience, or that it is too small.
The strategy below is the same as a traditional essay outline:
Show an ad that tells the customer what they’re going to see
Show them the product on the landing page (and capture them via a Call to Action)
Retarget them with an ad reminding them about what they saw.
This screenshot shows the four audience that we’re targeting:
Young professionals (the audience definition actually has 12+ “secret sauce” criteria)
Young people who already visited our landing page
Current FEE supporters who might be interested in FEEcon
Current supporters who have already visited FEEcon.org
Conclusion: Why retarget?
What are we trying to achieve with this strategy? Two things:
First, retargeting optimizes the return of our ad spend. Given a target audience, only 5% might be interested in the product we are trying to sell. We want to identify that 5% and spend much more effort converting them than the 95% who are not interested.
Second, retargeting allows for creating a conversion funnel. People who are seeing our product for the first time should get a different message than those who are already familiar with it. In the ad for the first time visitor, I can direct them to more information, while the retargeted audience is already familiar with the product and can be asked to purchase it or offered a discount (since I can also exclude people who have already bought tickets).
The goal of this project was to see if we could effectively market our content to an audience outside our core demographic in a data-driven manner. We picked this video as the test subject. The success of this experiment validated our strategy for the YEAR project.
Identify which of the target personas our list resonates with the most.
2. Brainstorm audience personas
We reviewed the video and discussed what kinds of audiences would engage with it.
We identified four audience candidates:
Black Entrepreneurship Fans
Inventors and Makers
Mike Rowe Fans who like Entrepreneurship
We tested four additional placements to display the content:
YouTube link from the FEE page
FEE.org article on the FEE page
Facebook add (content specified within ad)
Facebook Video Upload
We excluded Donors from the additional placement test, this left us with creating 12 ad sets (4 placements x 3 audiences, each containing 1 ad).
3. Create saved audiences
Each audience was iterated, so the below represents the final combination of criteria. We identified a number of people, hobbies, and shows which our personas might have in their Facebook profiles, but only the below were available as targeting options:
4. Measure test results:
In terms of placement, we saw that ads with the YouTube link, FEE.org article, and in-ad content were not responding at all. The per-engagement maximum was set at 15 cents, and there were only 1-4 engagements out of 40+ reach. Cost per engagement was 3-14 cents. Once we changed format to Facebook-hosted video, reach and results immediately took off, at 1 cent per engagement.
Out of the three audience we tested, the “Black Entrepreneurship” audience responded the strongest. Here are the video engagement stats for the first day:
Across all metrics, “Black Entrepreneurship” outperformed other audiences. Donors also performed well, as expected.
5. Boost winning ad set
Based on the first day, we decided to boost “Black Entrepreneurship” to $100 per day and “FEE Donors” to $30 per day. All other ad sets (and audiences) were stopped.
6: Measure results
$440 was spent boosting this video, with $38 used for the first day of calibration, and the rest directly on the selected audiences.
A total of 322,000 people were reached. 178,000 of those were from direct paid reach, while the rest were organic. However, once advertising stopped, video views dropped for 30,000+ per day to well under 1,000. The majority of organic views originated from people who saw the ad.
After the ad campaign ended, organic views dropped rapidly from over 10,000 to a few hundred (see below). This video is not effective with our existing audience, and did not have a lot of organic momentum.
Return on Investment
There were a total of 111,000 video views, and 8,000 reactions. Each dollar thus generated 252 views and 18 likes.
Cost per 1000 impressions was $1.50 for the “Black Entrepreneurship” audience and $8.03 for the donor audience.
The FEE Store link was clicked 301 times. There were 28 Real Heroes sold during the ad period vs 26 for the several months prior, so the ad likely drove virtually all of those sales. Facebook matched 7 purchases via Offline Action tracking (each purchase may involve more than 1 book.)
How many of the users who engaged with the ad will become frequent readers? We’re not sure how to track this. (Google Analytics? Facebook Insights?)
How to measure the value of paid ad exposure? Book sales? Shares? Organic uplift? Are paid views as such worthwhile?
Given that our existing audience was not the most responsive to this video, and given that we were attempting to target groups outside of our core user base, is it possible that we need to reach a larger number of people with paid views in order to reach some unknown saturation point that will spark more organic growth? In other words, do we need to spend a lot more money before network effects will take over?
Once we’ve effectively identified an audience target through testing, would different means to reach those audiences be more effective than paid social media ads? For example, should we have planned an advertising strategy that included direct outreach to major audience influencers around Black Entrepreneurship, such as Daymond John?
Targeting segments effectively requires testing different audiences.
We can effectively boost content to new demographics and even get them to buy products, but converting them to returning visitors will require a new strategy.
We need a clear call to action for ads to measure effectiveness and extract value.
The motto of FEE’s marketing process is “Always Be Closing.” Everything we do as marketers is designed to move our customer further toward closing the deal. Every message we make, be it a landing page, email, SMS, facebook ad, flier, whatever, asks the customer to take an action that moves them further down the conversion funnel. Each marketing communication has to have a large, prominent request to take a single action which will give the customer some sort of value.
This doesn’t mean that every marketing message has to ask for a deep commitment. We have a conversion funnel for each customer person which consists of a series of small messages. If we want students to apply for a three-day seminar, we first ask for their email so we can send them a free book or guide. The goal of each communication is to deepen awareness of our products and lower the barriers they have to the next step.
3. Make messaging personal
The essence of our communications strategy is to make every message we send feel like it was written just for you by a real human being who cares about your concerns and is eagerly awaiting a reply, then use marketing automation to scale up that personal feel to thousands of people.
There are a few ways we do this: A. Every email comes from a real human being. We don’t use any [email protected], [email protected] or [email protected] emails. This includes transactional emails such as payments, registrations, reminders, etc. B. Emails use a first person informal tone. We never say “we.” Messages take the tone of “I would really appreciate if you could do x.” C. We sign all emails with our name just as we do with our personal mail. D. We use plain text format whenever possible, especially if we expect a reply. When we use CRM tools to target messages, we export the names into Gmail or send a plain text message whenever we can. Marketing messages sent in plain text from Gmail avoid both the junk mail and the “Promotions” folder. E. We make messages short and to the point. Because we can speak directly to the recipients values, we can offer something that we know they’ll care about and don’t have to waste space addressing everyone. F. We talk like a normal conversion. We experiment with short, informal, lower-case subject lines such as “quick question:” or “you’re missing out.”
Marketing email from our CRM tool sent via Gmail
4: Focus messages on specific customer personas
It’s easy to say that every message should push the customer down the sales funnel. The hard part is to track who each customer is, and where they are in their journey.
At FEE, we use HubSpot to split our users into a lifecycle stage funnel (lead, subscriber, opportunity, customer) and a customer persona (college student, parent, interested donor, casual reader, etc). HubSpot tracks every visitor’s web and email interactions, and dozens of workflows use specific triggers, e.g., visiting the donate page recently, to classify people into personas based on recent behavior.
This allows us to tailor messages to the specific customer profile and offer them a product that we think they are most likely to be interested in. This minimizes our unsubscribe rates and keeps followers interested in our content.
Additionally, we make heavy use of retargeting for our advertising. We show Facebook and Twitter ads based on specific pages people visit. If you’ve visited FEEcon.org recently, you’ll start seeing more FEEcon ads in your Facebook feed until you register, which will switch you from our “promotion” (please register) to the “nurture” campaign (please share this with your friends). If you’re not on our daily email list, you’ll see a lead form in your feed, otherwise, you might see donor messaging if you’re a heavy user of the site.
Partial snapshot of personas and lifecycle
5: Experiment to identify the best strategy, then automate it
We don’t plan a grand marketing strategy for each product. The fact is, we have no idea what kind of message will resonate with our audience. Our marketing strategy is basically this:
A. Define audience B. Define value proposition C. Experiment with campaigns based on A and B at a small scale until we find something that works D. Scale up C.
We build workflows which capture the most effective strategies and automate them for each customer journey. We use HubSpot to build one or more workflow for each product which contain a series of calls to action, triggers, messages and rules. First we capture leads with a CTA on a website or ad, then we enroll customers in a workflow and nurture them until they convert (buy the product, register for an event, donate to us). This enrolls them in a new workflow which is designed to deepen their commitment and cross-sell other products, starting the process over again.