Finding where leads drop off in a multi-stage pipeline
An organization I worked at had a lot of applicants entering the pipeline, but very few made it all the way through. I wanted to find out where the drop-off was happening and what we could do about it.
The short version
The biggest drop happened at the very first stage. When I broke the data down by where applicants came from, I found that applicants from one specific source converted at three times the rate of any other. We added a small fee to filter out low-intent applicants and prioritized follow-up with the high-conversion source. Conversions went up.
The problem
The organization had no shortage of applicants at the top of the pipeline. People were entering the funnel in healthy numbers. But by the end of the process, only a small share had actually completed the steps to convert.
The team was spending a lot of time following up with applicants who never moved forward. That's expensive. Every hour spent on someone who was never going to convert was an hour not spent on someone who would.
I wanted to answer two simple questions. Where in the pipeline are we losing people? And is there anything in the data that tells us which applicants are worth our time?
What I did
I pulled the applicant data and built a four-stage funnel showing how many applicants reached each stage. Then I counted the drop-off at each step to see where the biggest leak was.
Next, I split the data by where applicants came from. The intake form asked everyone how they had heard about us, so I could group applicants by source and check whether some sources produced applicants who actually completed the process and others didn't.
To make sure the differences I was seeing were real and not random, I ran a chi-square test on the data, and followed up with pairwise tests between sources.
What I found
Finding 01
The biggest drop was at the very first stage.
A lot of applicants entered the pipeline and then went silent before the first real interaction with the team. That told me the problem wasn't with how we ran the rest of the process. The problem was with who was entering in the first place. A lot of them were never serious about going through with it.
Finding 02
One source converted three times better than the rest.
When I broke the funnel down by source, the differences were huge. Applicants who came in through one specific source converted at 58 percent. Applicants who came in through a generic online search converted at 19 percent. Same team, same process, very different outcomes depending on where the applicant came from. The chi-square test confirmed the gap was real.
What we did about it
Two changes, each one matched to one of the findings.
First, we added a small fee at the start of the pipeline. The idea was that a small fee would filter out people who were not serious without scaring off the ones who actually wanted to go through with it. The team stopped wasting time on applicants who had no real intent.
Second, we changed how the team prioritized follow-up. Applicants from the high-conversion source got contacted first and more often. The data made it clear that an hour spent following up with this group was worth a lot more than the same hour spent on a low-conversion source.
After the changes, total conversions went up.
Materials
- Full analysis (Excel) Workbook