This post got rejected by our team at inbound.org for being a little long winded - still, I wanted to publish it somewhere for my own reference. My own personal site and "Notes to Self" blog seemed a fitting place for that, together with a pithier, to-the-point takeaways post on inbound.org itself.

So please forgive the length and benign detail of this post - as with the title of this blog, it's a 'note to self' :)

Recently, we’ve been sharing insights behind the scenes of what the community team does to grow the community here at inbound.org such as our outreach process (and how we scaled up to 100,000s of personal emails) and our community growth model (how many discussions and contributors we need to hit our end of year weekly active users goal).

But how can we drive outsized engagement with less input from our own team? We wanted to figure out how to stimulate lots of activity and discussion discussion without having to scale our own inputs proportionally (i.e. hiring and managing a massive community team).

One way to do this is by figuring out what our most engaging content is - views, comments, comments - and how we can reproduce it. First place to start is our most commented threads...

What are our top commented threads?

If you’ve been around the community for any length of time, you’ve probably seen our “Ask Me Anything” threads in the community with notable marketers and brands in our industry. These typically have a lot of traffic and engagement. They’re also good for branding and new member acquisition.

But, whilst they have a lot of engagement, the engagement per person who see’s it is very low. “AMAs” are a one-to-many channel. The threads can only be answered by one person (or one team) which limits the potential number of comments back-and-forth that can occur.

Secondly, they’re difficult to multiply. AMAs take place over a few days (some days leading up to them for taking questions, and then a day with the experts answering questions). This makes it difficult to run many threads close together without them overlapping and performing worse.

AMAs are helpful and good for the community, but they’re a cherry on top and not a lasting strategy on their own to growing outsized levels of engagement in the community.

What types of threads are likely to get outsized engagement?

To answer that, we first need to define what we mean…

How can we measure “engagement” with content?

An engaged member isn’t just a reader. Contributors create content which creates value for others, so we need thread types that are more susceptible to get contribution.

You could think of this as a conversion rate of views to comments. This would exceptionally give you a number of no more than 5% (in the best ever threads).

Or you could the inverse - how many views do you need to a thread before you get a comment. For instance, Views:Comments and Views:Commenters. In English this means:

“For a given view by a relevant visitor, which threads are more likely to generate a commenter?”

i.e. Where could our content, promotion and outreach effort have the highest return?

Using the inverse 'ratio' form gives us bigger, more meaningful numbers (instead of small % variations of view-to-comment conversion rates) and also helps us calculate our outreach effort more transparently. From our personal email outreach process (we published how we scaled this to 100,000s of members, together with our numbers here), we know that we average about a 10% click rate (“open rate” x “click-through rate”) - 1 in 10 people who we email will click through.

By thinking of engagement in terms of “number of people we need to get a comment” it’s easier to run quick mental maths on the number of people we need to run outreach on.

By querying our database for all our discussions, comments, commenters and views, and then sorting them according to view:comments and view:commenters ratios, we can see a number of interesting insights.

The first of these was that there were very few AMAs (in fact, only one scored below 50:1 view:comments (2% comment conversion rate) - “We Run Marketing Agencies. Ask Us Anything”). This highlighted early how we were finding new, different sorts of threads over looking at top commented threads.

Three threads scored below 10 views per commenter (10% comment conversion rate) - extremely high levels of engagement.

Inbound conference in India? - This thread was posted by a community member who was looking to start a conference in India. This was almost collaborative in nature with lots of activity as members tried to organise elements of an event.

Clearly, the idea of creating a big, standout event “like INBOUND” resonated with our Indian members. Since the views:comments ratios were so strong, I emailed this to all our members in India. We just needed to fuel the fire with eyeballs, and even re-send the thread to people who had previously commented but with an update.
Introduction threads - We’ve experimented with threads for new members to introduce themselves. This quickly introduces the concept of commenting, gives us some material to point them in a direction that might interest them on site, and triggers a cycle of notifications.

However, these threads are hard to repeat and scale. Most of these weren’t very “social” - more a formality of “welcome to inbound.org” than a more authentic, welcoming, back-and-forth discussion like you might have in-person. Part of the problem here is those that comment on the thread have little in common with other members besides their cohort. To make this work in future, we’d need to build thread types by parts of members’ identities they share, like their location or skill.
Private Discussion - The most engaging thread however was in a private group. It discussed a major change for that group (which had financial consequences for all those involved), presented a video explaining the change, and then invited feedback. It took 2.5 views for every comment. This is hard to reproduce - but it illustrates one key principle. Humans are self-interested!

This wasn’t enough for a proper “strategy”

All these threads expressed some element of self-interest - the pride of a great conference in your own country, opportunities from promoting your new membership in the community, and the impact of a program on your revenue…. However, these were only three threads. It’s hard to create a strategy from these three threads alone. None of these were obviously reproducible.

We still had a large dataset of discussions showing high levels of engagement (views:comments) - I wanted to classify and categorise our top 100 engaging threads, and see if there were traits in common with others that we could replicate and scale.

To do this, I wanted to boil it down into a few, simple types - this would be easier to remember, communicate internally and build processes around than having several, overlapping types.

Given the huge variation of engagement different threads, classifying based on language, topics, keywords and so on didn’t appear that useful. This ruled out using a lot of tools or database queries to do this. Instead, I manually classified the top 100 based on the original posters desired output of the thread - what sort of answer were they looking for?.

Whilst there was some overlap, this bucketed 90% of the posts into four (almost) discrete categories - “POLS”. These are thread types which are abundant (repeatable!), and highly engaging. They offer the most potential for growing the community.

A quick caveat if applying to your own content - these classifications may not apply to other communities. Inbound is B2B, and for marketers. Use these as inspiration, but recreate this process to classify and categorise your own discussions and content.

Here’s how the top 100 most engaging threads broke down.

Process - Sharing “how to”
Opinions - sharing subjective, reasoned opinion
Lists - answers could just be bullets
Storytelling - sharing own experiences

These posts solicit instructions and to do lists. They’re looking for highly actionable advice to a defined problem.

They take three forms:

How do I? (my problem)
How should they? (learning about someone else process)
How do you? (as above, but gives space for self-disclosure and sometimes bragging)

Process threads are relatively easy to run outreach on. There’s often a clear match with a skill set - we can ask members whether they’ve any experience or suggestions for a named member. “Can you help with Casey’s PPC problem?”

These threads are easy to spot… see the Process examples on the inbound.org post about this article.

Opinions

If the output is supposed to be more subjective, then it has to be treated very differently. They can take the form of “closed questions” (with reasoning and debate in the comments), polls, why something is what it is, or a thread seeking a set of value judgements.

This works especially well if members of the community can be divided and debate against each other. So long as it doesn’t get personal or nasty, a good debate is good for community - it show’s people care.

So far as outreach, people don’t respond well to being asked for their opinion on something. It often isn’t that important. Conversely, they will respond if you point them to a controversial thread and say there’s a “good debate” that you “as a {interest_group} might find interesting”. It’s also the easy to keep opinion-based threads stoked by emailing members who have seen the thread with key updates from the latest comments.

I've shared ten examples of opinionated threads and debates in the inbound.org thread too.

Lists

If the primary output of a comment is a list (or from the thread is a list), the conversation can take a very different form. Often this limits meaningful interaction between commenters beyond “nice idea!” or “Haven’t seen that before!”. But they’re still powerful for driving engagement, and surfacing a lot of new insights and ideas very quickly.

Lists can take many forms, including sharing recommendations, plans, self-promotional posts (like this #Brexit sale thread) and so on. Arguably AMAs are similar to lists too.

Storytelling

Everyone loves a good story. It’s a form of content we’ve been aware works well in our community and tried to pursue with our original posts. They’re often a highly entertaining form of content.

It’s not a form of content that’s actionable (it’s hard to replicate someone else’s exact experience), and there’s little incentive to share for other people - outreach has to be based around people’s desire to share their past experiences. It also helps to do large-scale outreach after seeding the thread with a few, quality answers to set a good example.

It tends to take the form of either something a person has done, or something closely tied to that person’s identity (for instance, something eventful about the city they live in).

Under “other” we also had a number of content types which drive high levels of engagement, but aren’t abundant or easy to reproduce. Amongst them were critiques and collaborations.

Critiques

Critiques, such as those in The Pit: Landing Page Critiques, invite lots of contributions. It’s always easier to be a critic than a creator - use this to your advantage with peer-to-peer content! The Pit works because there’s a tight community which gets notifications of new threads there. Outside of here though, we’ve found it hard to scale.

Perhaps this something we can release in future with our channels and personalised homepage, and some element of a “private” group. Who really enjoys being critiqued in public? In your workplace, I bet there’s certain “contributors” who have no shortage of comments, critiques and things to say about various different projects. Private channels are likely the best way we can harness the engagement and content between critics and creators.

Collaboration

As with the Indian conference thread, if we can encourage collaborative work to happen on inbound.org (perhaps tied in with critiques?), we could drive a lot of new engagement and also business opportunities amongst members in our community.

One option here is to create private groups for companies to make it easier to share and discuss the latest news and ideas on inbound.org with just their own team. It’s one thing seeing something interesting, but it’s another thing to discuss that in private with trusted individuals. Perhaps in future, we can create private channels and integrate discussion threads on inbound.org with internal email, Slack, wikis and so on.

What threads don’t do well?

Whilst we’re running these numbers, it’s easy to look at what threads don’t do well” - threads which get no comments, low views and few upvotes (though upvotes can be a poor signal of quality with voting rings).

By reversing the order of the comments/views ratio, we could look into thread types that get very low engagement and views. These are discussion types we could advise against posting, give advice on how to tweak for higher engagement, actively ignore and avoid doing outreach.

Five categories stand out

Promotional - no value or purpose besides promoting something.
Obscure - hard to understand, let alone answer
Limiting - only has value to the original poster. Few benefit from participating
Statements - the original poster leaves nothing else to be said
“Do my homework for me” - the original poster has taken little to no initiative to solve their own problem already. Laziness isn’t attractive to the community

What about Articles

By running similar analysis on our articles (sort by comments:view ratios), we can see what sorts of articles drive discussion and engagement.

With articles in particular, there’s a lot of “great post”, “thanks for posting” and agreement with what was posted. Whilst it’s great to have positivity, and it can encourage members to share more quality articles, it doesn’t lead to great comments and debate.

The more interesting comments come where members take an article, and add their own commentary on top. This is hard to prompt for (it’s weird to ask people “what do you think” out of the blue) but ends up emulating discussions too. Sharing processes, disagreeing opinions, listing extra suggestions and sharing stories and related experiences. It’s especially easy to run outreach if the title of the article is a question itself.

How does this shape our community content strategy?

Whilst reviewing threads, we’ll look for threads which show these characteristics and plug them into our our outreach process. A few case studies:

Nurturing Controversial Threads

Knowing that opinionated threads drive lots of engagement, we look for thread titles on site to promote to the top. But, by monitoring social platforms, we can use people and content outside of the community to drive more debates within inbound.org.

This thread taking a ‘controversial’ quote that was dividing opinion on Twitter from Kristina Halvorson @halvorson (who wrote the book on Content Strategy) and asking people if they agreed or disagreed. This was powerful for pulling in a large pool of people - including many influential. When Kristina herself commented (in good grace), I created a list of our contacts who viewed the (already active) thread the day before and sent them an email with an update. 69% open rate and 24% click rate (delivered and clicked through).

We need to balance nurturing threads for activity against the negative perception we built amongst those who are quoted and then accused by our members. That said, we actively look out for negative. Shaped into a process - look for negative and controversial phrases about marketing on social and use those to trigger new discussions in the community.

Creating Long Form Stories

Stories can become truly engaging to our members. It’s also a form of content that can differentiate inbound.org from other publishers in the marketing industry. It’s something we’ve really tried to focus on with our inbound.org Originals. Some of our all time very stories include Confessions of a Google Spammer, A Competitor Bought 200 1-Star Reviews For Our Facebook Page - Here's Our Story and My Content Only Went Viral After My World Collapsed.

Panel Discussions

Topical “Ask Us Anything” threads which take little effort to start another one. Broader pool of engagement, and makes it clearer members can identify with the topic. These threads also play on the ‘availability heuristic’ since members are more likely to believe their question will be answered over posting.

Lazily, these threads are quite easy to setup on our end to. I’ve an HTML template which I can swap out the topic, add in the experts on the panel (via a Google Sheets tool I built to scrape our own site and concatenate the HTML together. Hacky I know… :p), and then share the on inbound.org. To get the first people there, I email out (with our personal email process), then segment the call to join our panel to our known contributors vs. just posting a question. This can all be done in under 2 minutes!

Here’s some examples:

We’ve experimented with the formatting of these threads (here’s a failed attempt) but found that tailoring an existing discussion thread. Though separate threads for each question is better for SEO, it dilutes activity across multiple different threads. We’ve found evidence for more activity generating faster activity. A simple rule of thumb to drive engagement in a community - increase social density! Get everyone with something in common in the same place and the same time

This works so well, it drives one of our key product decisions...

Creating “Never-Ending” Discussion Threads

Currently, members can upvote comments in threads so that the most upvoted comments float to the top. This is great for surfacing a “best comment” - which makes sense for answering questions, but doesn’t for some other formats:

Ongoing AMAs / Clinics shouldn’t have a best question to answer...
Ongoing Welcome / Introduce Yourself threads shouldn’t have a best introduction…
Events threads shouldn’t have a best comment

Rather, these types of threads would benefit from comments being in chronological order. This would mean we could continuously nurture threads like this and get perpetual, ongoing activity. Members would know to go to these threads to find new, relevant discussions, and we could nurture other members who would find them interesting.

Together with our new feed view that (some of you) might have noticed we’ve been A/B/C testing, I mocked up an idea of what the future inbound.org interface might look like. Inspired by Slack, I thought to call these never-ending threads ‘channels’. These would appear in the left hand navigation for members to tab between topics that they care about. The homepage would then be a mashup of all the channels you’re following.

Whilst we’ve successfully created clinics around topics, channels could take other forms too including:

Topics
Events
Locations
Latest Jobs
Companies
Private groups
Editorial picks & featured discussions
AMAs and other specific types of threads
Your conversations and @mentions

Perhaps inbound.org could end up looking something more like this mockup?

Mockup of channels

We can also redesign our sharing submission form to encourage freeform comment and turn article sharing into a source of discussions (which delivers more engagement, contributors and weekly active users). Everything would be a discussion thread, shared with a channel which has followers who have opted-in to that feed of content and discussion.

This is the direction that inbound.org is going - a personal feed of the latest ideas, news and people in marketing that matter most to you. Stay tuned for channels launching on inbound.org soon!

Join the discussion on inbound.org