Our team loves to tackle challenging problems so we were thrilled when we had the opportunity to flex our muscles on a project for a pay-per-call affiliate.
Our challenge was to create an online marketplace where publishers could list a public phone number for advertisers to bid on for incoming calls.
This is a worthwhile business model because it results in new sales leads for the auto insurance carrier.
Typically that process looks like this:
- A publisher creates a phone number (888) 888-8888 which is shared within the auto insurance industry.Auto insurance carriers and advertisers are notified of the phone number.Auto insurance carriers place a bid to have incoming calls to the number redirected to their call center.
- With such a complex model we understood that we would have to create something special so we recruited help from one of our favorite tools, Twilio.
- Twilio’s expertise in communication based apps, developer friendly API and documentation made it a perfect fit for this project.
To solve the problem we defined and built an algorithm using Twilio that would route phone calls based on a live bidding system.
Here’s how we did it.
When it comes to building algorithms we, first, identify parameters and inputs that will influence the outcome.
For this particular project the ideal outcome would be for the incoming call to result in a sales conversion, so understanding this we identified three inputs:
- Geography of incoming call
- Time of day
- Call quality/duration
Next, we used Twilio to describe each of these inputs via their API.
For each input we used the following:
- Geography: Capture phone number approximate location using the area code.
- Time of day: Capture call start time.
- Call quality/duration: Capture call transcript (text-to-speech) and length of time on call.
Using these inputs and an estimated bid for each incoming call we can determine a score for each participating auto insurance carrier.
The insurance carrier with the highest score receives the incoming call.
When it comes challenging projects like this it’s easiest if we break it down into smaller pieces.
In this case, Twilio answered (pardon the pun!) a huge piece of the problem by creating the “call center” functionality.
Because Twilio simplified this piece of the puzzle our team was able to focus on the key value add: creating the intelligence to redirect phone calls to the right place at the right time.
If you’re interested in learning more or want to get into the nitty gritty of the algorithm, feel free to contact me.