How to Be a Wolf in a World of Sheep
Don't get me wrong. There's nothing wrong with sheep. They're cute. They're cuddly. Their wool makes great sweaters.
But . . .
Their young are mighty tasty. Some people even like to eat the adults. I find them a little too gamey, though.
Nobody eats wolf cubs. Nobody even thinks of stewing a mature wolf.
All in all, I'd rather be a wolf. (tweet)
When it comes to Twitter, I would much, much, rather be a wolf disguised as a sheep.
Let me explain.
As many know, I'm working on a web app affectionately known as the Web-App-With-No-Name (yet). It's a crowd-built app that 26 Bees are directly helping me with. All the other Bees toss in their ideas as they see fit. This is the web app built by Bees . . . 306 of them so far.
Recently, Graham Edwards asked about account growth strategies.
If we embrace the notion that Twiter is a promotional broadcast medium, it stands to reason that the larger, the more targeted the audience the better.
Graham's question was a damned good one. The idea of an integrated targeted growth system is a damned good idea.
This web app is built on everybody's great ideas. (tweet)
- It started when Lisa Gallagher and Pam Williams asked for a simple way to build Click-to-Tweet links with an image like this.
- Then Gary Sharpe asked how I managed to tweet so often. It became a powerful scheduling platform.
- Jared Wiese asked if we can do the same thing with other platforms. Now, it easily integrates with IFTTT or Zapier for multi-platform posting.
- I had the idea that we should be able to pick our favorite members and retweet what they tweet in a selective way. Supportive scheduled retweets for Clans (favorites list) and Tribes (topic) were born.
- Gloria Ochoa wanted to force accounts she controlled to retweet each other. Ok.
Yes, this app was built on everybody's great ideas. Now, Graham has a great idea!
It brings up the question of how
Anybody can follow anybody on Twitter. That is both blessing and curse. (tweet)
It's a blessing because you can follow anyone you deem worthy.
It's a curse because it isn't all that easy to determine who may be worthy. Let's leave that aside for now. We'll get to how to tell who is a good follow in a minute.
The basic idea is "follow / unfollow-if-not-followed-back." That means you follow someone and give them a chance to follow you back within a given time. If they do not follow-back within that time you unfollow them.
Sounds easy enough.
The trick is in who do you follow in the first place. (tweet)
Why unfollow at all?
Unfollowing is not a punishment. We want a targeted, active following, not just any following. (tweet) If they don't follow-back within a reasonable time they are either not active enough or not interested/targeted enough.
Either way, they don't fit our goal. That's fine. This ain't a democracy. It's your following!
Some accounts follow many more people than follow them. That screams out, "Spammer!" If you look like a spammer, I know I won't follow back.
Twitter has limits as to how many people you can follow. Those limits can get pretty complex. Everyone is allowed to follow 5000 people max.
Okay, but I follow about 14,000. How does that work?
Twitter calculates the maximum number of follows as the greater of, number-following PLUS a percentage, or 5000. No one outside of Twitter really knows what that percentage is, but my best guess is 8-10%. Since I have about 16,000 followers, I can follow about 17,500 following.
Twitter also imposes a maximum of 1000 follows per day. That's fine. It isn't the greatest idea to follow that many anyway.
Twitter's limits will get weird at around 4500 followers. You are allowed 1000 a day, but you can't go over 5000 total. I call that "no-man's-land." Easy does it there. Drop down to following 100-150 a day, unfollow every second day. Stay there until you pass 7500 followers.
Unfollowing is an integral part of building a targeted following. (tweet)
More colloquially put, Ya gotta cull the herd.
How The Web-App-With-No-Name (yet) treats unfollowing
Fully automated unfollows are a BIG Twitter no-no. There must be user involvement. We ask you to click on a link to refresh your followers and unfollowers lists.
Yes, it is possible to automate that.
No, we won't automate it.
Twitter sets the rules. We just follow them.
Once you click the link, the app pulls in all your followers and who you follow (Twitter calls them "friends" so that's what we'll call them from now on).
You must then click "Process Unfollows." (I won't tell you again about allowable automation.)
The system compares your friends list (who you follow) with your followers list. Anybody who isn't on both lists doesn't follow you back. Within limits, they get scheduled for unfollowing. We dribble out unfollows. No matter how many people don't follow you back, the system only unfollows a maximum of 720 any given day.
Why "within limits?"
There are two reasons. Both are user set which helps fill the user involvement criteria.
- We need to respect a certain grace period for people to follow-back. Otherwise, if I follow you right now, and run my unfollows immediately after, I would unfollow you before you had a hope of following back. That's not good. Users set that grace period on a Growth Settings screen.
- There are some accounts that people want to follow regardless of whether they follow-back. We created a whitelist function to exempt them from the unfollow scripts. The whitelist is user set.
Two caveats. . .
Since unfollows are scheduled, there is a risk that a tweep follows you after you scheduled the unfollow, but before the system unfollows him. To that tweep, it looks like a bait-and-switch.
That;s not a nice thing to do.
We fix that by making a final check before actually processing the unfollow. That also protects against accidentally unfollowing the same person twice and pissing Twitter off.
The second caveat is one of timing. It's the question of when did you start following someone. It's possible to get the exact date and time when that happened. But, that takes a ton of processing power to no real benefit.
We use the time when the follow was entered into the system by clicking the refresh.
An example is in order. Say you started following me last year, but I never followed back. Say also you ran the refresh today for the first time. All your followers and friends, including me, would look like they all started following or being followed today.
Subsequent refreshes do not affect the creation date.
Continuing the example, remember that I haven't been following you for a whole year. As far as the system knows, you've only been following me for a day.
When you try to process unfollows with, say, a 3-day grace, the system won't find anyone that fits. There's "nobody" that you followed 3 days ago. You followed everybody today.
This is no big deal if you have less than 700 people to unfollow. It only becomes an issue if you have more than 700. I had 1700+!
Three ways to get around this issue if you do have more than 700 unfollows to do. . .
- One, stop following people for whatever number of days grace you set. Refresh your lists every day. Once the grace period ends, your unfollows will start scheduling. FYI: Refreshing daily is a good idea, once you're done with the initial clean-up.
- Two, don't follow anybody for a whole day. Run the refresh. Set your grace period to 1 day. Refresh again tomorrow and run the Unfollow script. Put your grace period back to whatever you want and DON'T run the Unfollow script again for that number of days.
- If you're sure you didn't follow anybody today, drop me a line. I can backdate everything so the unfollow works.
Each time you run the script, any unposted scheduled unfollows get erased and replaced with a new batch.
Following the RIGHT Tweeps
This is the big question, the one Graham brought up. How can you target your tweets? The short answer is you can't. But, you can target to whom you tweet. You do that by targeting your followers, and to a much lesser extent, by using hashtags.
So how can you target your following?
"Conventional" wisdom has it that you pick a big busy account that is in your field. Follow their followers. After all, they are "obviously" interested in the topic.
Remember, anyone can follow anyone. That's not good enough.
Let's go back to the goal.
We want an active following who is interested in the stuff we offer.
The follow-back grace period handles the active part. We need to figure out who is interested in our stuff.
Some people say to follow people who searched or tweeted a particular hashtag. Better, at least in theory, but still no cigar. Anyone can use any hashtag. Anyone can search a hashtag.
Spammers and scammers know this. Using a specific hashtag is no proof of anything.
Here's Our Way
We start with a list of big, busy accounts who are active in a specific topic. For the app, we originally planned to use 10 such lists. I may have to knock it down to 5. We'll see.
Then we start amassing data.
First, we pull out ALL their tweets and store them. We go back as far as Twitter will let us. Just how far that is, varies quite a bit. Sometimes I think Twitter's API is sentient and has moods.
We analyze the leaders' tweets to see who they mention and who they retweet. We pull out those tweeps and set them aside in a possible follows list.
Then we pull out any tweet from anybody, that mentions the leaders. That's a lot of tweets, but we ain't done, not by a long shot. We store the tweeps that wrote those tweets in the possibles list too.
Next, we pull out anyone who retweets the leaders. They too go into the possible follows list.
The now humongous possible follows list gets a run through. There will always be people who show up several times. We want to know who is on that list and how many times they made the list. We don't want to waste our time or annoy Twitter by following the same person several times. This is where we fix that.
Any user who is a member of that Tribe (topic) shares this data.
Now we turn our attention to the user's account. Anyone who mentions or retweets you gets added to a follow list. Ditto for anyone who adds you to a Twitter list.
Conspicuous in its absence
You may notice that there is no mention of following those who follow you. I didn't forget. Often, it's not in your best interest to do so. Thanks go out to Robert Bacall who convinced me of this fact.
It seems counter-productive.
Anybody can follow anybody. If they make the possible follows list that's one thing. We move them to the follows list.
Here's where it gets weird.
Not making the list only means that they don't fit our goal today. It does not necessarily mean that they don't fit our goal at all. Yet, they followed us.
The situation calls for subjective judgment. No algorithm can make a subjective value judgment. (hint hint, LI) Sometimes, it takes a human set of eyes.
This is one of those times.
We are working on a function to emulate a Twitter page. It will show, one at a time, who followed you but that you aren't following. It will have their profile picture, description, stats, and the last few tweets.
Note to self: While we're at it, we may as well include the option to input a Twitter handle. That way, you can check other tweeps out within the app.
You will have the information you need to decide if you should follow or ignore. One or two clicks and you're done. Trust me on this one, more often than not, you will ignore.
You can't stop someone from following you.
You can decide if you follow them back.
So what do we have?
We have a list of people who engaged with you. They proved their interest in you. We follow them for you if you don't already follow them. If they follow back in time, they are both interested and active enough. Goal reached!
Knowing how many of those there are, we add tweeps from the possible follow list until we make up your daily follow goal.
That possible follow list is a list of people who have proven their interest by engaging with the topic leaders. It remains to be seen if they are interested in you or if they are active enough. Following them will answer both questions.
Whoever ends up following you back has proven their interest and their level of activity about as well as that is possible to prove. I mean, for all we know, they may be active once a month and that day fell in your grace period.
No worries, over time they will likely wander off with none the wiser. The system will unfollow them at that point.
Closing (Finally, Right ?!)
You need to use the Web-App-With-No-Name (yet) or something similar to do Twitter growth right. It just isn't possible to do manually. You don't have the tools to act intelligently.
You can build a following over time. You will build it slower. You will have many false starts. It will never be targeted.
Most of the available tools aren't all that pricey.
On the subject of priciness: My initial idea was to keep the final Web-App-With-No-Name (yet) under $5 a month. This following stuff scares me because that just may not be possible. Following takes a lot of juice.
So, I have a question for you all. I could offer the growth-hacking as a premium version. It would be about $5/month more than the regular version. Or, I can bundle both at about $8. Those would be per account costs.
Let me know what you think. After all, it's true that this is my app from a legal standpoint. I wrote most of the code. I pay all the bills.
I own it. . . legally.
But. . . That isn't the whole story, is it?
This app, as it stands, is the net result of some 419 comments and suggestions from 306 Bees.
In a larger sense, it "belongs" to us all.
I'm looking forward to your input.