How to Be a Wolf in a World of Sheep: #2 Mind = Blown!
In "How to Be a Wolf in a World of Sheep" part 1, I asked for input on questions that were bugging me. The questions were mostly about the growth-hacking side of the Web-App-With-No-Name (yet).
You guys did not disappoint.
Some opinions surprised me. Some didn't.
One conversation was a game changer. Oh boy, was it ever. Mind = Blown is not click-bait. I think this App of ours is probably the most powerful Twitter app in existence. It certainly has the best cost to functionality ratio.
It was a simple yet mind-blowing thing.
Let's look at the other stuff first.
Following based on hashtag
I was surprised that people actually like the idea of following others based on hashtags they use. I don't consider it particularly useful.
Anyone can use any hashtag. I can post a picture of my cats under #ContentMarketing. That doesn't make Abby and Bruce relevant to content marketing.
Still, many people want the option. Who am I to argue? I added it.
I made it a little more useful. I set it up so that you can scan the accounts before you decide if they are follow-worthy.
That’s like sponge-worthy but not as funny.
I used it after the beBee Tweet Chat and found some interesting potential connections. I guess hash tags aren’t all that bad, after all.
More on tweet chats in a future post. The short version is that we will add a Chat moderation/promotion tool soon. Send me your wish lists.
Tweeting from your Library of stored tweets
It's important to keep a steady flow of tweets. I was so busy coding, I neglected my regular Twitter activities.
I ran dry. No worries.
You can now grab a bunch of stored tweets and schedule them to fill the day. Just tell the system how many and it will pick ‘em. That feature is on the Tweeting Today page.
Following those who follow you
As a rule, I don't follow people just because they follow me. I used to, but I got fed up of porn, scams, and spam.
Anyone can follow anyone. The problem isn't really about that. The problem is that it isn't easy to decide whether a tweep is follow-worthy short of visiting every single profile.
Like that's gonna happen.
Still, many felt that we should look more closely. So we did.
We added scripts that show you info, including profile pictures, profile by profile. You can see how many tweets they posted, when they last tweeted, how many followers they have, how many they follow, how many lists they're added to, and their profile description.
You can even see your follow/unfollow history with that tweep.
I'm still not a huge fan of following just because they follow. At least now, you have the tool to check them out quickly and easily.
Less surprising was the fact that few people liked the idea of subsidizing costs with promotional tweets. I wasn't thrilled with it either. I'm not even sure if it's okay under the Twitter developer's agreement.
Let's just say that there won't be any advertising in the foreseeable future. I don’t think we’ll need it.
That's because of the HUGE implications of the conversation I spoke about.
All pages are now fully responsive. They adjust for desktops, laptops, tablets, and phones. That involves some compromises but I think it worked out well. A true mobile-first version will have to wait until V2.0. That’s a seriously pricey proposition.
I also cleaned up the design. What is about programmers and bold type?
The Game Changer
I had three long conversations with someone who wishes to remain anonymous. She is not a Bee. She does not test the Web-App-With-No-Name(yet).
She can't do either.
I understand why. I don't push it.
The Web-App-With-No-Name(yet) is a Lite version of the Beast I wrote for clients' managed accounts. That Beast applies Big Data concepts to the full collected timelines of major players. It also collects the timelines of whoever interacts with those players.
That's about 3,000,000 tweets every month. It also costs about U$450 a month.
The Beast does far more than the Web-App-With-No-Name(yet) needs to do. For us, it's just important to know who followed/unfollowed and when, but not why.
We don't need a massive set of custom reports either. The reports Twitter supplies are ample.
Still, we need to target.
Still, we need to filter.
I kept thinking Big Data.
She asked, "Why?"
I asked, "Why what?"
She said, "The thing that gets in the way is all that data you need to mine. The Beast works because users choose whose timelines to mine for their specific needs. You can't do that for a few thousand users. You'd need a thousand servers. You wanted to create generic tribes, but that goes against the targeting thing. You can't be generic and targeted at the same time."
When she's right, she's right.
Epiphany: I don't need to collect millions of tweets to study who needs to be followed, unfollowed, when and why. I just need to know who.
Enter Twitter Lists
Twitter lists are lists people create and others subscribe to. It’s a deeper form of connection with a topic and a Tweep.
What if you could follow people who add you to a list?
What if you could follow people who add someone specific to a list?
What if you could follow people who are members or subscribers to a specific list?
What if you could review all lists a user subscribes to or is a member of, then follow members and subscribers to those?
You’d have a helluva lot of potential follows that’s what.
I pulled out 32,213 in twenty minutes.
They have an interest and/or expertise in the topic.
But . . .
You don’t know if they are active enough for you. You don’t know if they’re interested in you.
So, filter them.
The 32,213 dropped down to 16,387.
We swap out artificial intelligence for good ol’ human intelligence and solve the provisioning vs. cost issue.
Sure, the data mining approach is more accurate. The Beast maintained a 27% average follow-back rate over 11 months.
The Web App With No Name(yet) is running at about 19% for its first week.
I figure a 30% drop in efficiency compared to a 99% drop in associated costs is a good trade-off.
The final trade-off will be much less than that. The Web App With No Name(yet) isn’t taking retweets and mentions into account yet. The Beast does.
Mentions and Retweets
I haven’t forgotten about mentions and retweets. People who mention/retweet you, or a set number of major players, are people you want to follow. They are also much more likely to be tightly targeted and willing to engage.
Since we will not data mine timelines, they will be tougher to collect. Gimme a couple of weeks to figure it out.
You have more than enough to get you started.
How to use this thing
Beta-testers can access the Account GRRowth portal from their dashboard.
The first thing you need to do is hit Growth Settings and set up what you want us to do.
There may be people you want to continue following even if they don’t follow you. Add them with the Manage WhiteList button.
Then hit the Refresh All Relationships button.
The system will pull out followers and friends for all your linked accounts. (Friends are what Twitter calls people you follow.) It compares the two lists and reports back.
Be patient, it can take a while for big accounts or if you have many connected accounts.
Next, clean up your account. The system will tell you how many people you follow who don’t follow you back. Hit Process them all.
It’s a good idea to let the system run long enough to clean up completely before you start following anyone new. The system will gradually unfollow up to 700 people a day. Before unfollowing, it does a final check to make sure they still aren't following back.
Use that clean-up time to build up your potential follows list.
Following is a multi-step thing. First, you collect possible follows en masse. Then, you filter them. When you’re happy with the filtering, you schedule the follows from that filtered list.
Th etwo step process allows us to grab far more possible follows than we need. Filtering drops them down to a more targeted, manageable number.
It’s pretty easy. If you need more details, let me know.
I’ll be moving on to the how-to videos next.