MBE Capital Partners is an organization committed to helping minority and women owned businesses get access to the capital they need to grow. When Congress extended the Paycheck Protection Program (PPP) in July 2020, MBE Capital Partners turned to Fwd People to help them find eligible small businesses owners and encourage them to apply for the PPP funding they were entitled to.
There were two key challenges with this engagement:
- With one round of PPP loans already complete, we needed to find business owners who did not know they were eligible for PPP and therefore had not already applied.
- We needed to be highly efficient with our media spend. The client had a limited budget and given the importance of the mission we had to ensure that we reached as many small businesses as we could with the funds available to us.
We began by casting a wide net to quickly learn as much as possible about our target audience. To be efficient with our efforts we focused on three main digital channels: Facebook, LinkedIn, and Google. Within each of these channels we set up a series of test campaigns based on our hypotheses about which potential target audiences would be most responsive to a wave two PPP campaign.
A test and learn approach
We quickly got our campaign live and began a five-day test phase where we intentionally spent only a small percentage of our overall media budget. The goal of this phase was to refine our channel and audience mix so that we could scale the campaign significantly. Fortunately, we learned a lot in these five days.
We then optimized away from Google and LinkedIn and pushed all of our spend to Facebook where our conversion rates were 10-15x what we were seeing on those other channels. We also began to hone in on the right target audiences: independent contractors, restaurant owners, and Facebook users whose behavior indicated that they were entrepreneurs.
Further refining our targeting
While we saw dramatic improvements in our cost per application (CPA) during our test and learn phase, there was still room for improvement. By day five of the campaign our CPA was still coming in at $37 per application — still 85% above our target of $20. At this point we made two key optimizations to the campaign:
- We chose to focus almost exclusively on the independent contractor audience with creative that spoke specifically to people who received 1099s.
- We introduced an additional set of ads targeted at Uber and Lyft drivers — a subsegment of the independent contractor audience where we were seeing extremely high conversion rates.
These changes had their intended effect and by day eight our CPA had dropped an additional 51% to $18 per application. This increase in efficiency enabled us to steadily increase our daily spend which in turn caused the total number of daily applications to skyrocket — all without without negatively impacting our CPA.
Maximizing our spend
At this point we implemented our final optimization. Using the Facebook pixel we placed on our landing page, we introduced a lookalike audience that targeted additional Facebook users who most resembled the individuals clicking ‘apply now’ on our landing page. This change significantly altered the trajectory of our campaign, sending our CPA all the way down to $3.67 on day nine.
Because of this very positive result we could then comfortably maximize our media budget and our focus shifted to spending as much as possible before the PPP extension expired. While our CPA fluctuated slightly over the remaining two weeks, it never went above $7 per application — well below our overall target. Meanwhile, by spending the remainder of our media budget we were able to consistently drive. between 300 and 400 new applications each day we were in-market.
- Drove 5,675 new loan applications totaling $171,381,980 in capital dispersed to small businesses
- Hit a CPA of $7.43 over the full duration of the campaign — 163% below our target CPA of $20
- Saw a CPA of just $5.81 over the final two weeks of the campaign, a period when we spent 74% of the total budget
Return on ad spend