Vennli, Make Your Content More Effective: http://vennli.com/
Siftrock Uses Machine Learning to Save B2B Marketers Tons of Time on Email
Blog Feature

By: Paul Roetzer

Print this Page

August 16th, 2017

Siftrock Uses Machine Learning to Save B2B Marketers Tons of Time on Email

B2B marketers get a ton of email. In fact, the average B2B marketer gets 20 to 30 email replies for every 1,000 emails sent, which costs huge amounts of time when you are emailing thousands or millions of people says Adam Schoenfeld, CEO of Siftrock.

Siftrock is a machine learning tool that automatically manages and mines these email Siftrock marketing artificial intelligence institutereplies for B2B marketers. By integrating with marketing automation systems like HubSpot, Siftrock updates databases, improves email engagement, and even finds new leads. This all occurs without human oversight, freeing up marketing and sales teams to complete higher-value activities instead of managing their inboxes.

The result? More effective email engagement in less time. We interviewed Schoenfeld to learn how the company does it.

In a single sentence or statement, describe Siftrock.

Siftrock manages and mines email replies for B2B marketers; our product integrates with your marketing automation platform to clean your database, enable email engagement, and find new leads.

How does Siftrock use artificial intelligence (i.e. machine learning, natural language generation, natural language processing, deep learning, etc.)?

We use machine learning to categorize emails, mine and score contact data, and take action on different types of email replies and bounce backs. The typical B2B marketer gets 20 to 30 email replies for every 1,000 messages they send. When you look at companies with hundreds of thousands or millions of marketing contacts, this is a big problem and one that has traditionally been solved by manually monitoring an inbox (i.e. a big waste of time and human capital).

What do you see as the limitations of artificial intelligence as it exists today?

The devil is in the details. The last 5 to 10% accuracy is very challenging. This is particularly true when thinking about delivering accurate results for a specific use case. There is no generic approach that works across the board. Each type of data and each use case has tons of nuance and complexity.

This is certainly true in our world when trying to classify and understand messy email content. We quickly got things working, but it took years and millions of examples to build a training set and refine our algorithms to the 95% accuracy range that was required by our customers. In addition, there were lots of nuances to make the data work for our particular uses cases.

What do you see as the future potential of artificial intelligence in marketing and sales?

The technology appears to be taking off, and there is great excitement and buzz in the market. However, I think mainstream adoption will take a lot of time. Businesses will struggle for a while to define their philosophy on AI.

How much work do they want to let machines take over? How much risk are they willing to take while the technology is imperfect and immature? Will they let AI takeover interactions with customers?

Go to Vendor Website

What makes Siftrock different than competing or traditional solutions?

The main alternative to Siftrock is manual labor, unfortunately. We stand out in two main ways:

One, accuracy and speed without requiring human oversight. Two, the flexibility of our integration with marketing automation platforms (e.g. Marketo, Eloqua, HubSpot, Pardot) allows us to accommodate pretty much any customer use case.

Who are your prototype customers in terms of company size and industries?

We primarily serve B2B technology and services companies. This includes high growth, mid-market companies like Procore, Avalara, and Glassdoor and larger enterprises like Citrix, Frontier, and LogMeIn.

What are the primary use cases of Siftrock for marketers and sales professionals?

Email reply management helps with three big B2B marketing use cases:

  1. Improving database health and deliverability. We help invalidate out-of-date contacts, find job changes, and update contact info.
  2. Enabling two-way email engagement. We help route real human responses to the right rep, measure reply rates, and track all campaign responses in one place.
  3. Finding more sales contacts on target accounts. We surface new sales leads from out of office replies, find replacement contacts when people change jobs, and mine email signatures for new phone numbers.

 

Any other thoughts on AI in marketing, or advice for marketers who are just starting to explore the possibilities of AI?

Start with the business problems and use cases that you want to solve, then look at AI-driven solutions as one of many paths forward. AI isn't going to solve everything, but it has some great applications and should definitely be considered in the mix.

Also, look for vendors who are willing to listen to your feedback to improve their platforms. Most of this stuff won't be perfect out of the gate, but if the vendor shows a desire to listen and learn, you will see massive gains over time.

Get free access to the Ultimate Beginner's Guide to AI in Marketing: https://www.marketingaiinstitute.com/beginners-guide-access

About Paul Roetzer

Paul Roetzer (@paulroetzer) is founder and CEO of PR 20/20, author of The Marketing Performance Blueprint and The Marketing Agency Blueprint, and creator of The Marketing Artificial Intelligence Institute and Marketing Score. Full bio.

  • Connect with Paul Roetzer