Marketing Budget Optimization for Store Retailers
During the course of developing predictive response models for thousands of clients, Wiland has observed that in nearly every prospect marketing campaign a portion of the audience performs abysmally. These low performing segments generate few responses. The handful of new customers generated would have to produce stellar long-term results that are "beyond reality" to ever recoup the money that was spent to acquire them. And to compound the misery, it turns out that most of the customers generated by these inferior segments are inferior customers with poor long-term value. The cost of acquiring them and then continuing to promote them is so great and their performance is so bad that they will never be profitable. The more promotional money a retailer throws at acquiring such customers the deeper it goes in the hole. We help retailers save this money so that they can let it fall to their bottom line or redeploy it in more productive marketing efforts.
Our predictive models, built primarily to find good prospects, find the bottom too. We use them to optimize merge/purge net files via a solution that we call Wiland Marketing Budget Optimization. The name derives from the fact that, often, marketers obtain prospects from many sources, combine them and eliminate duplicates with a merge/purge process, and produce a net file that is either sent to a printer for direct mail print production or to an ad tech company for digital advertising deployment. Some use only one source for prospects and skip the merge/purge step. Regardless of how the file of net prospects comes into existence, we can optimize it to improve campaign performance. Virtually every campaign can produce better results than it produces today! This is how it works.
At least two weeks before the merge date for an upcoming campaign, we do two things to prepare for successful implementation of Wiland Marketing Budget Optimization.
1. We work with our client to identify three past marketing campaigns that are "similar" to the upcoming campaign. The similarity that matters most is similarity of offer (pricing/discounts, season orientation, merchandise, creative treatment, etc.). For example, we would not want to study performance of a postcard mailing featuring an extra 10% discount on summer clearance merchandise to predict who will respond to an in-season holiday offer. Rather, we request prospect promotion files from campaigns in which a similar mix of merchandise, at similar price points presented with similar styling, was offered. The campaigns should be recent so that the brand is unlikely to have changed significantly since the campaigns occurred, but long enough ago that results are largely complete so that the majority of responders can be incorporated into our analysis. Once the three campaigns are identified we use them to create a new predictive model that will be used to perform the optimization process.
2. Next, we need to understand campaign costs so that we can make accurate Return on Investment (ROI) calculations in our analysis. How much revenue did the campaign produce, expressed in revenue per thousand or revenue per person promoted? What was the campaign cost per thousand (CPM)? And finally, what are campaign level product & fulfillment costs, expressed as a percentage of revenue. This information enables us to calculate actual campaign performance at the variable contribution level, a vital aspect to our model development process, as described below.
These advance preparation steps are critical aspects of the process. When the first advance preparation step is done well, we will correctly predict prospect behavior. When the second step is done well, we help make informed decisions concerning which prospects are worth the marketing investment to promote, and which are not worth promoting.
Wiland Marketing Budget Optimization Production Steps: Net File Processing and Evaluation
As soon as the net file exists, it is sent to us via our secure FTP site, and the sender notifies our team that it has been sent.
1. We optimize the net file quickly, in no more than two business days—usually within 24 hours.
2. Each prospect record is assigned a score that indicates their likelihood of buying and their desirability as a customer if they do respond. Highly desirable, likely-to-buy prospects get a high positive score. Prospects not worth the marketing investment required to promote receive a low negative score.
3. We divide the prospects into 20 segments based on score. We know the actual ROI of prospects scoring in the same segment from the past marketing campaigns analyzed. We often find that Segment 20 produces triple digit negative ROI—losses so large that it is virtually impossible to recover the investment cost with subsequent purchases. In most cases, the bottom 2-5 segments of the past campaigns produced such poor ROI that they should not have been promoted. The theory, of course, is that if a segment was that bad in three past campaigns it should not be used in the upcoming campaign.
4. We provide the client with a Marketing Budget Optimization Long-Term Value report based on model performance and consult with them to reach agreement on which segments to omit. Maybe -40% ROI is acceptable in some cases, but it almost never makes sense to promote to segments that have worse than -40% ROI. We work with our clients to choose the appropriate cutoff point, and we omit accordingly.
5. After omission of the agreed upon segments, the net file is smaller. Our client then decides whether to let the savings fall to their bottom line, or to supplement the net file with higher scoring modeled names using Wiland Balance Modeling, or to redeploy the saved budget to other marketing campaigns at Wiland or elsewhere. When the net file is finalized, we provide it to Wiland Digital Solutions™ if it is for a digital marketing campaign, or to a service bureau or printer to be presorted and moved into production if it is for a direct mail campaign.
6. We consult with the client and discuss options for how best to utilize the saved marketing budget.
Wiland Marketing Budget Optimization cuts the waste out of prospecting. Wiland clients have seen significant improvements in overall customer acquisition results, in terms of both immediate performance and longer term customer value, as a result of utilizing our optimization service—even when the rental cost of dropped records from outside audiences is considered. Ideally, retailers will be able to make advance arrangements with audience providers to pay only for the portion of the audience that survives optimization. Wiland leads the way in this regard. If any portion of an audience we provide as input to merge/purge is later classified by our Marketing Budget Optimization service as too weak to justify promotion, Wiland will credit 100% of those not utilized. Other providers should do the same. This is about producing good results for the marketer. Optimization of net files just prior to utilization is the future, and Wiland wants to deliver a better future, one of increased client revenue and margin. Avoiding wasteful spending is a key part of accomplishing this.
We are confident that the lowest-rated segments will generate sales that are only a fraction of the cost of marketing them.
Some may wonder how it is possible that during Marketing Budget Optimization we would classify prospects we provided just a short time ago as too weak to justify the marketing cost. This usually happens only if there is significant time lapse between the two events. If elapsed time between when we supply prospects and when we perform net file optimization is short, few or none of the prospects we provide will be classified as weak. But if significant time passes, we will have more current information available—such as new transactions, which are posted constantly, and new clients, which are added daily— for ranking consumers during optimization. Even though it can be challenging for marketers to build the necessary time for net file optimization into their schedules, this should become a regular part of the mail file production process. Failure to leverage the predictive power of Marketing Budget Optimization results in big marketing losses that are absolutely avoidable.
The first time or two that a marketer uses Wiland Marketing Budget Optimization we encourage them to mail all segments in order to prove that the segments we classify as weak are actually so inferior that they are not worth the marketing cost. We are confident that the lowest rated segments will generate sales that are only a fraction of the cost of marketing them. For one recent retail client, the segments that we predicted would perform poorly cost $40,000 to mail plus $3,000 in product and fulfillment costs for a total cost of $43,000. These prospects generated only $7,000 in sales. No matter how well this handful of new customers performs in the future, our client will never recoup its original acquisition cost, or at best, it would take many years! Testing should be done to prove it, but in almost every case the theory will be proven valid.
Sometimes only the last segment, about 5% of the prospect universe, is so weak that it is not worth utilizing. More often, the bottom 20% to 25% of the prospect universe is weak. Sometimes the percentage of weak prospects is even higher. There are several reasons that so many poor performers end up surviving to a net file:
1. Use of Marginal Sources. In an effort to reach a desired promotion volume, many marketers use audiences that are only marginally acceptable. If an audience is marginal in its entirety, it stands to reason that not every person in the audience is of equal value. Some are better than average, some worse. Many of the poor prospects identified by optimization come from these marginal audiences, though even the best audiences include segments that perform at a significant loss.
2. Time Lapse. Another reason for the presence of poor prospects is the significant amount of time that often lapses between selection of a prospect audience and merge/purge processing. There was a time in the last century when marketers ordered lists just 1-3 weeks prior to the mail date. Now, it is not unusual to see audience orders placed 30-90 days prior to the launch of a marketing campaign. In such cases, time lapse causes degradation in the recency factor of prospects provided. At Wiland, we post updates to our database constantly, with hundreds of thousands, usually millions, of transactions uploaded daily. This is important for two reasons. First, it enables us to identify net file prospects that may have had great recency a few weeks ago but haven't bought anything since, making them less qualified than they once appeared to be. And second, we can identify recent buyers that look very good now despite not having appeared to be a good prospect a short time ago. Through optimization, we utilize this new information to remove prospects that are no longer qualified and will, upon request, replace them with good ones that have very fresh purchase activity and, consequently, are more likely to respond.
3. Poor Audience Decisions. In established companies with seasoned staff and talented brokers or agencies, it is fairly uncommon for audience decisions to be made unprofessionally. But it happens, especially in companies with little experience, too little staff, or other issues. Poor audience decisions can also be the result of order fulfillment errors at a service bureau or in-house fulfillment department. And, uncommonly, they can be the result of an audience owner padding a good universe—intentionally sprinkling in some bad stuff to increase order volume. One client who was a victim of this called it "watering down the scotch." Whatever the reason, poor audience decisions do occur. They ought to be caught and eliminated through optimization rather than wasting money on poor prospects.
The aforementioned factors virtually guarantee that every net file will contain a high percentage of poor prospects, with 20% to 25% being typical. Thus, every net file should be optimized and, ideally, supplemented with high-ranking Wiland balance names.
Overcoming Obstacles to Marketing Budget Optimization
The benefits of Marketing Budget Optimization are enormous, and the opportunity for us to help our clients is real. However, some have expressed reservations about Marketing Budget Optimization, citing a variety of reasons for not taking advantage of this valuable service. Here we will address real and perceived obstacles, suggest solutions, and encourage industry thought regarding these important topics.
Obstacle 1: There's not enough time in our production schedule for Marketing Budget Optimization.
While Wiland Marketing Budget Optimization does add a step to the production cycle, it can usually be accomplished in a day, sometimes two. For marketers working under extremely tight schedules, the addition of an extra 24-48 hours is not a trivial matter. However, we strongly believe that the financial benefits of optimization far outweigh the costs and inconvenience of modestly lengthening the production cycle.
What a difference a day makes, twenty-four little hours.
Many years ago when the USPS began offering presort postage discounts, Wiland Services was the first company to have software and a carrier route directory to enable presort services. We offered the only path to presort savings. Many of our clients and other marketers insisted that there was no way that they could add time to their processing cycle to allow us to perform presort. However, within a few months they realized just how much money they would save on postage and how it would speed delivery of their mail. Within two years, virtually every marketer of any size had either added time to their production cycle for presort processing or figured out how to compress time allowed for other functions.
Can you imagine a large direct mailer skipping postal presort because it doesn't have time for it? Of course not. It doesn't take long, it saves money, and it speeds delivery of the mail. Just as we introduced presort decades ago, we believe our emphasis on Marketing Budget Optimization today will have a similar result: everyone will eventually do it because to not do so is to accept budget waste that could be avoided.
Dinah Washington, a popular jazz vocalist in the 1950s, sang, "What a difference a day makes, twenty-four little hours." A great song. And in this case, Wiland Marketing Budget Optimization should be music to the marketer's ear. Careful, well-tested integration of Marketing Budget Optimization into the production cycle is likely to have a strong and positive effect on the marketer's bottom line, much like the magnitude of presort when it was first introduced.
Obstacle 2: I have to pay for all net prospects whether I market to them or not. Isn't it a waste of money to drop names for which I've already paid?
Our experience has shown that clients utilizing Wiland Marketing Budget Optimization see a significant boost to their bottom lines by avoiding wasteful spending on poor prospects—even when the cost of dropped rental names is factored in. We encourage clients to test this for themselves by going ahead and mailing low-ranking prospect names identified by Wiland under a separate keycode. However, we also advocate for changes in industry practices so that marketers are not forced to pay for dropped audience segments destined to waste marketing budget.
The marketer should not have to pay for marketing to demonstrably bad prospects nor should list suppliers be deprived of rightfully earned compensation for their services.
Many in the industry—including list owners, brokers, managers, agencies, cooperative databases, and other prospecting sources—oppose providing credit for names dropped in the optimization process. This is understandable, as the majority of audience providers are compensated on a per thousand basis. A list broker, for example, may be accustomed to being compensated $X per thousand for arranging audiences on behalf of marketers. They will see their income plummet if we remove PP% of the audience and the marketer does not pay for the removed portion. Their objection is valid. No one wants their income to go down. And in this specific case, the list broker likely would do more work for less compensation because it would need to negotiate net payment to the audience owner. The same thing applies to agencies that are paid a percentage of production costs. There are several possible and necessary solutions to these issues.
1. Continue paying prospect sources for dropped names where required, but determine the optimum use of marketing budget saved by not mailing those segments that are predicted to be poor during optimization and redeploy the savings to other, more productive marketing efforts. There are many ways that preserved marketing dollars can be redeployed. Several of the more popular options include:
a. Order more marginal audiences arranged on a net basis. Use only the consumers in these marginal universes that score above average in the optimization process. This may drop 70% of an ordered universe, but what remains should be highly productive. This allows the marketer to tap into previously disregarded list universes which expands prospecting affordably.
b. Use Wiland Balance Modeling. This will find productive replacement audiences based on transactional information added to the Wiland database very recently.
c. Depth Test and Refine. Depth testing is the process of testing deeply into a large audience identified by a predictive model by selecting nth cross-sections of the top scoring segments. Many multichannel retailers do not conduct adequate depth tests. Depth testing, while it sometimes produces initial losses, is far superior to marketing to demonstrably weak prospects eliminated by optimization for two reasons. First, the first depth test almost always performs better than the weak segments eliminated by optimization, even if it does not meet desired performance level. Second, and most importantly, depth testing often leads to something grand: a refined model that will usually produce a larger volume of successful audience than would have been possible without depth testing. In other words, it can increase the brand's revenue growth rate and do so cost effectively, producing profitable new customers.
d. Thoroughly Test Alternate Channels. Some of the promotional dollars saved by not marketing to the inferior prospects identified by Wiland Marketing Budget Optimization should be shifted to thorough testing (or rollout) in alternate marketing channels. Retailers whose digital audience we have refined can shift some budget to direct mail. In cases where the net file we optimize is for a direct mail campaign, some of the saved budget can be redeployed in digital channels. Whether you call it multichannel, omnichannel, or some other term, we don't live in a single channel world. Marketers should test alternate channels, keeping budgets small until results are proven. Learn more at Wiland Digital Solutions™.
2. Change broker and agency fee structures in a manner that rewards optimization. For example, they could be compensated as they are today for non-optimized audiences but at 1.2 times the current fee for optimized audiences surviving the optimization screen. This should provide incentives for everyone to optimize for the client's benefit. The principle concerning compensation is this: the marketer should not have to pay for marketing to demonstrably bad prospects nor should list suppliers be deprived of rightfully earned compensation for their services. Multichannel retailers should thoroughly test Wiland Marketing Budget Optimization to determine whether, in their case, the benefits of optimization are worth re-working list provider compensation to make it fair and ensure that providers are properly compensated for their work. Most marketers will find that it is indeed worth the effort.
Obstacle 3: Won't Marketing Budget Optimization negatively impact the workflow at my service bureau?
While Marketing Budget Optimization will indeed impact service bureau workflow, we've found that the process is relatively seamless for the vast majority of our clients. Let's say a service bureau is doing merge/purge and presort now and sending the net file to a printer. When optimization is introduced, the service bureau must now send the file to us for modification after the merge and prior to presort. Once optimization is complete, usually within 24-48 hours, we send a new file back to the service bureau, where it is converted, presorted, and shipped to the printer.
While marketers must work out these additional details with their services bureau, we've found that most are well equipped to handle the extra work, and clients using Wiland Marketing Budget Optimization find that it's worth the effort in virtually all cases. The financial benefits associated with cutting out poor prospects from a promotion are just too great to ignore, even in cases where the service bureau imposes a small fee for the additional processing. The key is that service bureaus should work smoothly with Wiland's optimization process for the client's benefit.
Obstacle 4: I've planned my campaign around a set volume of acquisition names. Won't Marketing Budget Optimization cause me to fall short of my desired promotion quantity?
Direct mailers usually must estimate final print volume well in advance so that the correct amount of paper and other materials can be procured and Operations departments can be properly staffed, along with numerous other considerations. Marketing Budget Optimization makes this prediction more difficult. How many names will we drop in the process? Of the portion dropped, how much can be replaced with productive audiences?
We are eliminating poor prospects that cost the merchant greatly.
These are not easy questions for a merchant to answers. But each campaign will build knowledge for the next, and most clients take the first opportunity as a verification test and will actually promote the quantity that should be eliminated, thus more deeply understanding their metrics and setting in motion the forecasting plan for the next promotion. While we could simply supply as many balance names as needed to meet the target quantity (which is actually the optimal choice in many cases) our goal is to optimize to our client's benefit. This may require reductions in overall promotion volume, or it may mean adding new prospect sources as input, sources that agree to be compensated only for names surviving the optimization screen. The fact is, from an order volume standpoint, we have found that the elimination of the bottom segments in an optimization typically has minimal impact due to the abysmal response from these names.
The key is this: We are eliminating poor prospects that cost the merchant greatly. Even if final marketing volume is smaller than would otherwise have been the case there should not be major impact on sales volume. We're eliminating low likelihood responders. Thus, marketers should work within their organizations to encourage optimization that benefits the brand.
Many retailers need to improve their bottom line and increase the long-term value of the new customers they invest to acquire. Wiland Marketing Budget Optimization can deliver on these objectives. The high price paid to promote unresponsive prospects takes too much of a bite out of marketing budgets. Retailers and their agencies, brokers and other service providers have an opportunity to work with us to make marketing budgets more productive. Doing so is good for everyone, and can be achieved starting today.