Follower Data Mining With Twitter Tools

One perplexing problem I had with Twitter was "how do you find people who will Follow You Back?" We all know there are many bulk-follow tools; but they rarely produce good results (even with a 33-50% Follow Back ratio) and eventually Twitter may suspend your account or reduce your Follower ratio.

I tried all of the tools including those for sale and those for free in Beta testing or Alpha development.

We follow people on Twitter because: we are interested in their content; we hope they will follow us back (and possibly buy something and/or spread the word), or both (interest and follow backs). I fall into the last category - but my main focus, for my new web venture, is Marketing Via Twitter.

Now, it takes considerable time to find, digest and analyze a Twitter account's Followers or those they Follow - if you want to build a large base. Certainly, you can search on keywords or use some similar tools to help you, and that will help get you started. But, if your offering relies on a large Twitter Follower base with specific characteristics, you need to analyze and Follow fast - especially if you want to be ready for key Twitter on-times (11-3 or so weekdays + heavy weekends) (and those late night or different time zone people.

A. I found and use Tweepi.com. This free app enables you to look at 40 Followers at a time and display specific data about each one including ratios. For example, I look for: (mostly) US based, 800+ Followers, a good Follower/Following ratio, RTs and number of updates. Armed with this information I Follow. I track results each day and we are receiving a 50-60% Follow-back ratio with spikes of 70%/80% and infrequent (Wednesday) lows of 25-30%.

The approach enables a great Twitter Follow-back ratio of quality or hopefully-quality Followers.

B. Finding a Twitter account or accounts to mine is also a challenge and that requires searching skill, luck, intuition, determination, and testing. For raw follow-back fast tests I use Tweepi + other tools such as Buzzcom, etc.

You need to analyze an account at the beginning, mid-section and end pages for: obvious spammers, porno, location, RTs, Updates, etc. - and often that leads to L2 and L3 (supply chain data mining talk for deeper data levels) look-see's such as who does a Follower's Follower's Follow. (supply chain theory works here!)

Sometimes I actually converse with a hi-potential Follower and ask about their Follower-universe. For example, I noticed that several Twitter accounts had Followers mostly in the 1000 - 5000 range with 8-19K spikes - no low numbers. Several accounts explained they weed those out over time and that their experience is that people in those ranges have a hi-propensity to follow back quickly, often RT, and generate the most Updates - possibly or probably for marketing or (RT) re-marketing needs.

I adopted this same theory and found it works well - and when there is a RT or a D/RT I can see/calc the rough audience reach.

Usage:

To use Tweepi rapidly (because its GUI loads slowly) I use dual monitors as follows:

1. Monitor 1: Load Tweepi and use normally. Note that as it loads a screen you can start your analysis by unchecking (not checking) those you do not want to follow. (much faster).

Note: While loading, press on any Twitter account's icon to get a feel for who they are and even proceed to L2 or L3 data.

2. Print out screen shots of the screen loads for metrics. (we have a plan for this)

3. Monitor 2: Using Outlook, I start to track metrics such as: Time-2-Fback and Who. Because, within 5 minutes of my first screen Follows (with Tweepi) we start to receive FBs

We compare who followed-back and when (not interested in Thank You DMs) on a sampling basis and commit this data to a spreadsheet.

Note: we receive 85-105 new Followers per day 'without' Following first. They find us.

Note: when a journalist RTd, we received 800 new Followers the next day.

Note: when we strike gold (heavy RTs and Mentions) we are near 530+/- per day

Steve

At this point I am Tweepi Power User and they asked me to help with product roadmap suggestions.