Does progressive profiling work?
August 18, 2015
In the marketing ops world, you hear a lot about “progressive profiling” as a way to increase lead generation and quality. The basic idea is, if you want to collect 10 pieces of data about a lead, you do it in subsequent visits:
- [First Name]
- [Last Name]
- [Company Name]
- [Job Title]
- [Company Size]
- [Phone Number]
- [Interest in buying]
And so on. The upsides seem pretty obvious, right? Why wouldn’t you want to reduce your form lengths, while gathering more data about your leads? It seems like a way to boost conversions (by reducing form lengths), without the tradeoff of having less information.
The downside to progressive profiling
However, progressive profiling has one really obvious downside, which is that it requires repeated visits in order to get that data. As a result, you:
- Can’t send repeat visitors directly through to the content; they must fill in a form every time.
This is a bad experience for them, and it reduces the chance that your prospect will actually read the material you’re producing. On the other hand, maybe you want your visitor to jump through hoops… to show commitment?
- May have difficulty reactivating dormant visitors.
Let’s say you wait to ask for Company Size until the second visit. But they don’t reconvert for a while after their first visit. Do you send them the Small Business or the Enterprise reactivation campaign? What if they keep coming back to your site (showing interest), but they never convert again? Why not just ask upfront?
- Get locked in to your marketing automation vendor.
If you have a bunch of progressive profiling data in your Marketo instance, you’ll end up resetting all visitors to Form #1 if you switch to Pardot or Eloqua. You could use your own JS instead, but that’s an extra layer of technical complexity and only works if you are hosting your forms on your own site.
- Get bad funnel data.
A use case that’s often proposed for progressive profiling is that it tells you where your visitor is in the funnel. If they’re coming back to fill in another form, they must be further in their buying journey, right?
I’m not sure this is true. It seems like the most reliable way to know this is to infer it based on the content. If you produce a whitepaper called “what is Product X?” and another called “Evaluating Product X vs. Product Y”, data on which is being requested is probably more reliable in determining buyer journey stage, than simply seeing how many times a visitor has come into the site.
Another pernicious effect: marketers will use a single form, and then rely on progressive profiling to hide certain fields based on the assumption that number of visits equals readiness to buy.
Seems like it would be much better to have separate forms tailored to separate funnel stages based on content, and only then, if one of those lacks necessary data, implement progressive profiling to make sure it’s captured.
Or just cut down your forms so that getting all the data from a late-funnel lead is not arduous.
I’m not sure what the right answer is
I totally get the appeal of progressive profiling.
And yet I’ve never seen any data on whether it actually achieves its stated goals. In fact, I’ve never seen anyone propose a test for it.
I guess a simple one would be to divide your leads into 2 pools at random, one using progressive profiling and one not, then see which one generates more revenue over time. (But isn’t that every test?)
A more practical and faster one might be to:
- Decide what constitutes the information that you truly need for effective marketing and qualification.
- Present this all in one form to a set of visitors, to establish a conversion baseline.
- Present this in separate progressive forms to another set. See how many visitors get through all the forms.
It really boils down to: Which do you think is more likely? That a visitor will complete a form with eight fields?
Or, that they’ll complete a form with 4 fields, visit again, complete another form with 2 fields, visit again, and then complete another form with 2 fields?