Many of you know about the Glyphius software I wrote to analyze ads for profitability. It analyzed the text of millions of profitable and unprofitable ads to find out the words, phrases, capitalization, punctuation and numbers that were more often associated with profitable ads vs. unprofitable ads.
That database was then flipped around and allowed the Glyphius user to score several ads against each other to find out in advance how profitable each ad was likely to be when compared with competing versions of the same ad.
The score was arbitrary. One could start with an ad of “Lemonade 10 cents” and press the score button. The score might be 311 which is meaningless. But then if you enter “Ten Cent Lemonade” and press the score button, that version might only score 240. The software was using predictive analysis to determine that the first version of the ad was much more like profitable ads as compared to the second version of the ad.
Glyphius was validated with a large dataset of ads that were then actually run in A/B tests. The overall percentage of time that Glyphius predicted the winning ad in advance was over 85% of the time. That saves a lot of money.
Can the same thing be done for telephone calls? With text, we really just have the length of the text, the words, phrases, punctuation, capitalization and digits to analyze. In some ways, voice is easier to analyze. It doesn’t have capitalization, punctuation or digits to analyze. To analyze the words and phrases (groups of words), you simply have to run the recorded phone call through a speech to text process. Easy; right? (Disclaimer: Not all states allow a phone call to be recorded with single party consent… check with your attorney before recording business telephone calls). Analyzing the length is just as easy as with text.
In some ways, analyzing audio is a little more difficult. Audio has the additional components of tone, tempo and amplitude. In other words, does a deep voice or a shrill voice (or something in between) yield higher profitability in sales calls? Or does it matter at all? A study could find out.
The same goes for tempo and amplitude. Should one talk fast or slow or is a medium tempo better for closing sales? How about amplitude? Should you talk louder or more quietly than the person on the other end of the line?
I am imagining a tool that does all of that analysis. What if you could analyze the words and phrases post-call? Then you could drill into your sales team what phrases should be avoided and what words and phrases are very consistently used in highly converting sales calls?
One of my colleagues mentions the other day that the word “demo” was being banned with his inside sales team. Is that a good idea? I don’t know. But a Glyphius like study of the words used in telephone calls that ultimately lead to a sale can be compared (percentage wise) against telephone calls that ultimately lead to a failure to close the deal. The answer can be had using pure science. The Glyphius engine can be used to answer that question and thousands of other questions about other words and phrases.
The other part of the tool that would be specific to telephone calls would be based on monitoring an active call. After a Glyphius like database was compiled to answer the questions about tempo, tone and amplitude… an automated coach could be devised. The software could monitor your call actively and give you coaching about when you speed up your tempo, speak in a higher or lower tone of voice or to change how loud you are speaking. How cool would that be? How would you like a computer coaching you with real data during a telephone call to maximize your profits?
Those tools don’t yet exist, but I have created similar tools in the past. What do you think? Would you use this kind of tool to optimize your business if it existed?
-James D. Brausch