Skip navigation

See It

3 Posts authored by: David Gutelius

Working with a wide range of Eloqua customers, one issue that has come up frequently is around email frequency management. Sending contacts too much email too frequently can kill an opportunity fast and lead to bad spam reputation. But it's hard to track how often contacts are getting emails in any given week when you've got 3 or 5 or 30 different campaigns going - even if you have a sophisticated stoplight program in place.


Surveying activity histories, we noticed that many of our Eloqua customers had some portion of their contacts getting emailed more than 10 times per week, and often a lot more. There's also the ever-changing battle against abusive bots spamming your online forms.


We've created new intelligent analytics as a part of Motiva AI's smart Frequency Management system that lets you explore which contacts are getting what emails from which campaigns so you can do something about it. And we're already uncovering opportunities to improve as well as surfacing plenty of abusive bot behavior.


Screen Shot 2019-04-15 at 11.23.49.png


So: are you over-emailing? If you're not sure and want to see your own results, ping us and we'd be glad to show you. And even if you are sure, we might find something surprising.

Audiences are key to everything marketers do, right? You can have the best content, timing, sequencing, and channel strategy in the world, but if you're not talking with right people none of it matters. This post outlines how it's possible to improve audience segmentation through message experimentation, and how you can deploy that learning into other marketing strategies beyond just email.


Audience understanding is often hard-won. A combination of intuition from personal experience, contact attribute requirements, and inputs from your team each play a factor in defining and refining audience segments. And often, once those segmentation definitions are established, they don't change very much.


One thing we've discovered running large scale campaigns for different customers is that with the right open attitude and some effective reporting, you can learn a lot from the results of testing different versions of messaging. It's something we frankly hadn't expected to see with Motiva AI, but very glad it is. Take a simple canvas campaign. The use case is the following:


  1. You have a segment for a given campaign. Think about a fairly generic campaign you might run over a larger audience, or one where you're just collecting new contacts automatically and flowing them into your pipeline.Screen Shot 2018-04-18 at 12.15.48.png
  2. You run a Motiva AI experiment step with a few variations of your message.
    Screen Shot 2018-04-18 at 12.15.06.png
  3. Motiva tracks response patterns from your populations on each message variation and gathers data as your campaign runs.
  4. You use the Who Responded report to see what Motiva is discovering about your audience. Here's where it gets interesting. More often than not, we're finding that within a more generic audience, there are actionable subpopulations who are responding to different versions of your messaging. Here's an example where we're looking at Company Size and Industry attribute data laid on top of response data across three versions of a message in our campaign:
    Screen Shot 2018-04-18 at 12.19.04.png

    Motiva lets you project whatever data you have on contacts against message response results. Here we're seeing that there are potentially at least two distinct subgroups - a Electronics industry group and a Charitable Organizations group who are responding to different messages at higher rates. Note that as opposed to what most marketers look to get out of messaging testing, we didn't get to a clear single message that was a "winner" for the larger population. That tells us that either our messages were equally good, too similar to make a difference, or that maybe we have subpopulations responding in different ways.


Now, maybe in the example here you realize you don't really care about the Charitable Organizations contacts because they aren't your ideal target customer. Yes, overall that Version A performed better than C, but that was among an audience that isn't your focus! You just learned that you could increase performance by tweaking your content to better fit the groups you DO care about. Maybe this is an opportunity to filter and re-segment. Maybe it's also an opportunity to use data to drive a discussion with the creative team about doing more of the stuff that works for the group(s) you're interested in. The point is that you're able to both accelerate your ability to target and adapt to your audiences, and improve the performance of your campaign.


You could imagine getting into a campaign creation pattern where you're constantly testing and refining those audience definitions not just on a campaign basis, but across your team. And as your team improves its audience intelligence, you can deploy that learning across your entire marketing strategy.


To make this super concrete, a large Eloqua customer ran a 12-way message test for an onboarding campaign triggered from an external download. Contacts flowed into their campaign, hit a Motiva step and Motiva executed the 12-way test as it received contacts. We were able to find a few different versions that the larger audience responded to (no single best message), but when the customer reviewed the campaign Who Responded report they realized they could use those results to target audiences much more precisely than they had been and corrected some assumptions along the way. They used these insight to improve and tighten up messaging, and then took several of the subpopulation definitions they discovered into improving PPC and SEO strategies. Team learning FTW!

AI is all over the news these days. It’s hard to know how to make it work for your marketing team. This post outlines some practical tips in using the Oracle Marketing Cloud and AppCloud offerings like Motiva AI to bring new adaptive intelligence capabilities into the way you design, execute, and improve your marketing programs.


When you get beyond the term “Artificial Intelligence”, you’re really just talking about software that learns to do things. There are lots of opportunities where you can get started with AI today that will have a measurable impact on your marketing. Here are five tips for how to go about it.


Tip #1: Start simple

Don’t try to take too much on with AI out of the gate. Start small, show some results, and then build on your success. Begin with simple proof of concept use cases that you can measure easily. A good candidate here is message testing in a single campaign – but going beyond simple A/B type testing. You can use a tool like our own Motiva AI to test and automatically find winning messages that lead directly to campaign response improvement. Profit!


Tip#2: Match the right task with the right tool

There are some tasks that machines tend to do better than people - and machine learning applications will get better at it over time. Here are some great candidates:

  • Audience segmentation and definition
  • Message testing and optimization
  • Personalization
  • Send time optimization
  • Data cleaning
  • Advanced analytics


Example: A large national healthcare company recently decided to focus on message testing and optimization, and used the Motiva AI Cloud for Eloqua on a patient-facing audience and saw a 200% difference in click-through rate by simply trying lots of message variations in the population. Motiva adapted to the audience preferences it observed, which allowed the campaign to adapt organically. A simple place to start, with big impact.


Tip #3: Look for “10x” opportunities

Ask yourself: where could we make the biggest impact in terms of customer response or labor savings? More often than not, machine learning can at least help the human marketers improve decision-making; in some cases, you can just outsource the entire workflow to intelligent helpers. Campaigns that most directly touch revenue or direct conversions are great places to improve pipeline dynamics.  Combine that with labor savings from automation. 


Tip #4: Measure and improve

It’s vital to think about what your definition of success is for a given use of machine learning and how you’ll measure progress towards your goals.


  • Will it be in terms of time saved for your marketers? Then track their time – develop a baseline for the workflow you’re interested in and the difference over time.
  • Will it be in terms of campaign performance? Again, make sure you’re collecting the data and reporting for the story you want to tell.
  • Will it be in terms of effects downstream in the sales process?  Ensure you can track your treatment effects all the way through your pipeline.


It’s not difficult, but you do have to do it. It’s just good modern marketing practice.


Tip #5: Remember it’s about your audience, not just the tech

Your number one concern should be how to develop that communication channel with your end audiences. Technology can be super useful here, but not all technology and not all the time. In terms of AI-driven tools, ask yourself:

  • Does this help me learn more about my customers and their needs?
  • Does it help me serve my customers better?
  • Does it strengthen the customer experience?


Make a connection between the tools you’re using and how they ultimately lead to positive customer impact.


The adaptive future

These tips are a place to begin to think about how to bring machine intelligence into your marketing. We are just seeing the start of a marketing revolution with machine intelligence combined with human intelligence.


For more information on Motiva AI, check out our free pilot in the Oracle AppCloud.

Filter Blog

By date: By tag: