Announcement

Akkio + Salesforce: Powerful No-Code Machine Learning For Your Sales Team

by
Jon Reilly
January 28, 2021

Akkio is an easy to use, high performance, flexible, and affordable machine learning solution for Salesforce. 

  • Incredibly easy to use. Just connect your Salesforce account, select a key outcome to predict or forecast, and deploy the result right back into Salesforce.
  • High performance: Predictive power that meets or exceeds all major machine learning providers, but in minutes, not hours.
  • Fully flexible. Combine data from multiple, different sources with your Salesforce data to create predictive models. 
  • Very affordable. Pay for predicting key outcomes, not for the number of Salesforce users. 

Salesforce has Einstein, but it’s incredibly expensive - $150 per user per month. And a quick read through recent Einstein reviews also shows a range of limitations:

The ability to quickly and affordably integrate machine learning into Salesforce is a game-changer. ML can help you score leads, prioritize opportunities, estimate deal size, and drive substantial revenue growth as it increases your sales motion efficiency. 

This guide will show you how easy and powerful it is by walking you through an example -- using Akkio to stack rank new leads by how likely they are to convert into closed-won business. 

Step 1: Connect your Salesforce Account to Akkio

Log into your Akkio account. Create a new flow and under “Inputs,” and select Salesforce. You will be prompted to connect your Salesforce account (which will require you to log into Salesforce and approve the Akkio connection). 

Important note: your Salesforce account must have API access enabled to connect to external services like Akkio. 


Once your Salesforce account is connected, Akkio will begin loading in your data. After a short while, you will see something like this:


You can choose any input dataset to build a machine learning model. If you have key data in two places, use the Merge function to match records with their key outcomes. You can also merge data from external sources even if that data does not have matching unique ID’s using the Fuzzy Match flow step. 

For this example, we will run a conversion prediction on Leads, so we choose the dataset “Lead” as our input. 

Step 2: Predict key outcomes

Next, we add a prediction step by choosing “Add Step” on the left-hand side and selecting “Predict.” For this model, we will be predicting conversion - so we will choose “subscribed” as the prediction target from the list and click “Create Prediction Model.” Each custom ML model is trained on 80% of the data and tested on the held back 20% of the data. Training takes about 30 seconds, and when it’s complete, we are presented with the model report, which helps us understand how the model performs. Our results look like this: 


Zooming in on the performance, you can see that the model is going a great job - it correctly classifies the review’s tone as positive or negative over 90% of the time. 

The model yields a really valuable business result - especially when we combine it with the percentage likelihood of a lead converting closed-won. Running a quick backtest of the model, we can easily make the business decision to drop two-thirds of the current leads (those scored less than 5% likely to convert) while still capturing 98% of conversions. 

That is a massive improvement in sales team efficiency! Now that the model is ready to go, the next step is to push the predictions back into Salesforce and set it up to run automatically when new lead records appear. 

Step 3: Deploy to Salesforce

We add one more step on the left-hand Flow menu to deploy our predictions back into Salesforce - scroll to Deploy and select Salesforce as our target. 


Here we see a preview of the output fields and have a few options. First, we will deploy our predictions back into the “Lead”  object in Salesforce, so select “Lead.” Next, we can choose our run frequency (every half hour in this case). We can decide if we want to apply the model to every existing record right away - we do, so we select “True” for “Run on Deploy.” 


Finally, click Deploy at the top right, and the predictive fields will be added into Salesforce and populated. The first time the process runs, it can take a bit of to complete, depending on your SF database’s size. 

Step 4: Add Akkio Predictions to your Salesforce Views

Now go back into your Salesforce account and navigate to the Leads view. Under the Settings wheel choose “Select Fields to Display” and add the Akkio fields into your view (in this case, “Akkio subscribed,” and “Probability subscribed is yes”). 


You should now see the model predictions, and you can quickly sort them from highest to lowest probability of conversion!


After the first time running, the model will only predict on records that are changed since the last time it was run. New leads that populate into Salesforce will be scored every half hour. 

Wrapping Up

Akkio is an easy to use, powerful, flexible, and affordable way to bring the power of machine learning to your Salesforce workflows. Whether you’re ready to do better than Einstein or just looking to use the power of ML more affordably, Akkio makes it easy to crush your sales and marketing KPIs while staying under budget. Give Akkio’s free trial or free usage tier a try and see for yourself.

SIGN up
Grow Faster with No-Code ML
Now everyone can leverage the power of AI to grow their business.