What do you like best?
The strength of Infer is the validity of the model. Isn't that why we're using scoring as marketers to begin with? It's been my assertion since I first started down the predictive lead scoring path in 2013 that the company with the best modeling is going to win, and for us that has been Infer. We've tested other lead scoring vendors, some of which are more credible than others, but over time Infer's model has absolutely proven itself. Infer has kept its focus on predictive lead scoring (mostly based on profile) and hasn't drifted into broader service offerings (like demand gen applications - finding leads in the market who may not know you) but to their credit they've acknowledged where their core competencies lie and not tried to over-promise and under-deliver in other areas.
What do you dislike?
I wouldn't say there is anything I particularly dislike. I would love to have the success we've had with lead scoring applied to broader parts of our business like demand gen and data appending, but to Infer's credit those broader services are not their core competencies and I respect the fact that they want to stick with where they know they can be successful. A lot of people talk about UI for predictive models and being able to expose predictive variables to the organization, and wanting to see more of that, but a lot of that point of view reflects the fact that many marketers and sales professional quite frankly don't understand how these models work, so it's a bit of a red herring. I don't get wrapped around the axle with UI because it's not an individual variable that is predictive, it's a combination of different variables with different weighting and the way they interact that's predictive, which is hard to boil down and represent in a UI. I think Infer's UI is fine - their slider is the best part.
Recommendations to others considering the product
People looking at Infer and similar vendors need to do the back-test trial. Infer's staff is straight-up in what they build and honest in how they anticipate it will roll into production. We tested another vendor who essentially gamed the system during the back-test by using variables that would not work well in production. They did this by using variables that have information/data that is acquired over time, essentially by a lead or contact being worked by a sales or market development team and augmenting that variable with data that is highly useful in building a model. But that data is not available when a lead first comes in, so this vendor's model failed miserably in production. Basically, as buyer you need to be diligent in reference checking. Also, don't get sucked in by the UI - see my previous comments on that. The UI is useful, but the good models are highly complex - that's what makes them good. Infer's models are so much better than what comes out of the box in marketing automation - don't even go down that path if you're a serious marketer. But because they models are complex, trying to surface "magic predictor 1 and magic predictor 2" is just not realistic. If it were that easy, you could do it yourself. Test the model, work with Infer, and you'll be fine. Be a critical consumer of the other vendors. There are one or two others out there that are credible, and a bunch that I think are so lost they don't know just how bad they are.
What business problems are you solving with the product? What benefits have you realized?
My part of the business is SMB, and we have a strong content play with free content and a high volume of campaigns and inbound leads. We use Infer for lead prioritization as well as understanding what leads / contacts are in our database and how valuable they are, and using that to guide our targeting decisions. We've recently taken lead scoring and Infer's model further to integrate it into our SEM bidding platform.
The benefits have been clear - we're able to follow-up on the top leads and not get bogged down trying to follow-up on tens of thousands of leads and campaign responses that are low to zero probability buyers. It's important to note that Infer won't identify when someone is in a buying cycle, just that they're a good fit - so presumably the highly scored leads/accounts have a high probability of one day buying, but not necessarily at that moment in time. But I can share this: virtually the only companies that buy in any given period of time have a medium to high Infer score. Leads at the 50th percentile and lower almost never buy. So while we can't say that a lead scored in the 90th percentile, for instance, will buy today or tomorrow or next year, we can definitively say that a lead in the 10th or 20th percentile will never buy. And that insight has been invaluable. It's allowed us to help keep our market development costs down, helped us with lead routing and triage, and given us insight as to what campaigns, vendors and tactics are giving us the best yield.