A 3 Yr. Pay Per Click (PPC) Case Study: Financial Services Customer Acquisition

The Pain and Glory of $5 a Click Campaign Management

Company Background

The outlined company represented here offers short term payday loans to qualified applicants. The website allows users to find loan information and fill out a form to apply for a loan. The loan application information is processed and verified by telephone representatives. This study covers campaigns started in 2005 with the campaign being abandoned in 08 due to cost of acquisition and a "saturated" PPC market for this company.

Client Overview

Before retaining IWB PPC services, the client was managing the campaign internally. The campaign had less than 100 keywords with poorly targeted ads and creative. The acquisition cost was at $120/Application and the loan conversion rate was a single digit. One of the problems specific to analytics for the clients' industry is what is deemed a "new customer" and a "completed conversion". Although a form may be completed by a qualified applicant and a loan extended to the applicant if the customer had a loan with the company in the past it was not deemed a converted loan.

IWB Opening Campaign, Adjustments and Results

Before doing any work on the PPC campaigns IWB first performed an Internet Integration Audit (IIA)  for the "service". The audit accesses the transaction process from customer click to Step 2 conversion (an originated loan). The audit is a review accessing the work/sale flow from web to transaction fulfillment. The "knockout" form (step 1 in conversion flow) was simple and straight forward consisting of yes/no tick boxes. The loan application form (step 2 in conversion flow) contained over 30 required fields with extremely strict input validation rules.

Knockouts are customers deemed to not qualify due to employment status, age etc. Many of these were later verified to be caused by over zealous application form validation (programmers... just because you can doesn't mean you should;-} or user input errors. For instance finding bank account numbers on a check isn't difficult, bank routing numbers are and they were knocking out incorrect formats.

Web Design Adjustments: The Value of Internet Integration Audits (IIA)

The original form looked daunting, took well over 5 minutes to complete and had an abandon rate that far exceeded the norm. The results of our IIA indicated that the original development team overlooked the users perception of the daunting form and complexity of the input validation. Our initial review took over 15 minutes to complete a validated form (the reviewer is a programmer). The first change we made was to improve the loan application form which was an obvious impairment to conversion.

The IIA Management interview identified that many of the fields were verified on other databases by the phone representatives during the loan application interview, therefore, the rep could ask for the information during the loan application interview. Using this criteria the 30 fields were reduced to 6 and none of the fields required validation.

The management interview also identified that 3 fields should be moved to the "knockout" form because they were a requirement to qualify a buyer. Bank, State, and zip codes were moved to the knockout form because they were actually qualifiers since some banks didn't allow direct deposit access to these types of companies and some States have laws that restrict this type of loan. I can only imagine how a user feels after spending time filling out the brute of a form only to find the service isn't supported by their bank! Improving the "knockout" form later became a very profitable move when IWB sourced other companies that could service the customer. This provided better service for the customer with the added benefit of revenue from the referral. In some cases there is high demand for some types of leads which provides an "aftermarket" for non converted PPC leads.

Management of the company were skeptical of the "short form" saying they had tried it in the past and it wasn't very successful. False positives were dramatically reduced by the IWB "short form" with no discernable change to the conversion rates. What the management failed to understand is that internet users don't give up info easily, especially, when they know you are phoning to verify it later. IWB's philosophy when accessing these user actions and perceptions is to ask ourselves "why do we need this info", often, that is the same question that is going through the users mind as they fill out a form on the web.

The "knockout" form allowed 3 chances to input all the correct answers. IWB added programming to save the incorrect "knockout" answers providing extra data for reps to identify potential misleading information given during the loan application interview.

Conclusions

Internet marketing is as much about integrating the two worlds, the virtual and real world, as it is about audience, usability and visibility. Often syncing the web service with the real world transaction process eliminates duplication of tasks by users or staff resulting in significant gains in customer satisfaction and sales. The departmental focus and protection of "turf" often results in duplications within the transaction process crossing the real and virtual worlds.

The IWB philosophy is to take as little input as required to complete the transaction. For instance our research shows turning down a form completion because the Credit card doesn't validate to the Credit Card company known format is an impairment to conversion. If you have an email or valid phone number accept it and verify the number later with the customer. Note you have to call the credit card company on purchases over $50 anyway so... why annoy the customer with over zealous form validation when you can engage them on the phone or by email?

Campaign Adjustments (The First Six months)

Like most in-house PPC campaigns the targeting was poor to non existent and the ads were weakly written so the CTR was low needlessly driving costs higher. The Google Network and Content campaigns were of course running in the same group because the client didn't know to split them, even though it is recommended by Google in the setup. Question for Google:
If you recommend these be split why does the default put them together? Could it be you want the advertiser to experience the wonders of unmonitored Content Network exposure?

The first thing IWB did was to split the Google and Content Networks into separate adgroups with much lower content bids. This resulted in a .75 to $1.05 decrease in overall CPC. These are big numbers when there are 1000's of clicks and that decrease represents a 25-30% decrease in CPC. The Content network can be a blessing or a nightmare the monitoring of it is usually the deciding factor. Later we saw a further decrease in CPC when the ability to set bids were added to the content network. In fact as Google improved the transparency and control the marketer had over the costs and locations ads displayed on the content network the CPC has always fallen. IMO, some day the transparency between the Google and Content networks will be equal.

Initial keyword research and analysis identified a total of 500 targeted "seed" keywords for the campaign. We created targeted “landing pages” for the campaigns to improve the conversion rate. Another very useful technique is spelling and input errors. We found that generally these fell into two groups very good converters or very bad converters. Nothing in between upon further review we realized that of ten the poor converters were coming from the content network and looked to be of "dubious" quality. I won't use the F word but... it did smell an awful lot like a mackerel left in the sun for a few days.

Targeted advertisements were created to improve the CTR (Click Through Rate) resulting in decreased ad costs and higher positions and conversion. After the first month we had identified many poor performing phrases, adjusted bids or deleted the keyword. Once we are at this phase IWB pays close attention to day parting opportunities. When clicks are $5 it is worth spending a day or two auditing the traffic of a large data set to discern when the best time to turn on campaigns with higher or lower bids. Also in some lead generating scenarios the business close at 4 and turn the campaigns off. We found this to be the case in this industry likely the result of the cost of the clicks and lower conversions rates later in the evening.

After the Google campaigns had been running well for about 3 months IWB began setting up accounts on Overture, Ask and a few secondary PPC platforms. The client purchased ads on one of the keyword systems that used applications on the users computer to overlay ads on content. This was a miserable failure and the provider was seemingly inflating clicks with bots or humans.

Ask was dropped after 2 months due to dreadful conversion and inability to use American Express. Yes a major company that couldn't accept Amex! IMO, that was just one of the many silly mistakes Ask made during its PPC venture. A toy platform didn't help either. Overture had to be watched like a hawk and eventually many of the budget setting techniques that worked so well on Google had to be abandoned because we found no matter how high you set the budget Overture spent it all usually taking conversion out behind the barn and putting it out of its misery.

Another major problem was unlike Google who do not carryover budgets from previous days, Overture does, once again with devastating hits to conversion. The fact we shut PPC off on the weekend was very problematic. In the end we had to set the budgets well below what we knew should be used to take advantage of traffic spikes.

The only performing secondary engine was SuperPages, the rest we won't even mention in case you take the same risk IWB did and lose. The Pay Per Call was way to expensive so we stayed out of that. The main reason being is their cost/click was higher then we were paying/conversion. No matter what you did that was going to be a drag on total cost/acquisition. The free listing and waiving of monthly fees did contribute to the success of the program. When we paused the campaigns for a few months they started billing us that fee, we shut it down at that point. When we started things back up Idearc refused to give us the same deal and at that point it meant that even if we maintained everything exactly costs would rise 30% basically pricing themselves out of the picture. It was no surprise to me when one of the US YellowPage companies went into bankruptcy recently. They have an unrealistic value of their programs. All their programs are grossly overpriced and overhyped with "unbelievable" stats to base their claims on.

At this point the budget had increased to $3,000/day from $300 on google and $500 on Overture.

Mid Term Results & Adjustments

IWB PPC analysts added 3,0000 keywords in different match types, more misspellings, input errors and keyword modifiers based on the performance of the original 500. We implemented budget and campaign management techniques which only companies doing real time bid monitoring can provide. We changed the budget settings so more clicks came from the Google content network. Over time the budget splits between the Google and Content Networks varied so we could control and maintain our acquisition costs. As the bids on the Google Network rose we simply lowered the budgets, day parted or turned weak adgroups off.

The content network is a lot of work, however, once you figure out the weak sites and add your negative keywords you can increase impressions and click volume significantly. The best move Google made in this regard was the ability to set separate default for the content network. At was at this point that we saw the Content Network take off for us.

Initial keyword research and analysis identified a total of 500 targeted "seed" keywords for the campaign. We created targeted “landing pages” for the campaigns to improve the conversion rate. We separated the Content and Google Search networks into separate campaigns for more control over budget and bids. Targeted advertisements were created to improve the CTR (Click Through Rate) resulting in decreased ad costs and higher positions and conversion. After the first month we had identified many poor performing phrases, adjusted bids or deleted the keyword. At this point the budget was increased to $4,000/day for Google and $1,000 for Yahoo!.

Yahoo! sponsored search is neither trusted, nor supported, by IWB at this time. IWB utilized streamlined Yahoo! campaigns removing almost 90% of the original keywords from the Google campaigns which were converting at 25-30% on the Google PPC Platform. We ran Panama exactly one day. Man that was scary! In the summer of 2007 the PPC program was discontinued and replaced with a much cheaper lead source.

Final Results

At it's peak the campaign budget was increased to $7000/day with a spend of 112,000 for Google AdWords and 12,000/mo. on Yahoo! which  was shut completely down within days of the release of Panama.  The overall campaign conversion rate never fell below 25% with the loan origination running around 15%. At its peak the campaign generated over 11,500 applications in December 2006 at less than $10/app! The campaign was turned on again in 2008 however, the costs had risen to the point that PPC was no longer a viable lead generator.

As mentioned one of the problems with customer acquisition in this industry is the way the conversions are calculated. About 18 months in we started noticing the churn rate (previous customers) was rising quickly. We were able to counter this by going exclusively with the Content Network. We were able to meet the acquisition targets for about another 12 months before the costs started to rise again and PPC was replaced with another lead source at a 1/4 the cost!