Case Study: Marketing Budget Optimization, Multichannel Retail
This case study is a factual analysis that evaluates the validity of the following theory:
The Research Sample
The case study evaluates the actual results of 30 customer acquisition campaigns conducted by 10 different multichannel retailers. All 30 of the campaigns are catalog mailings. Three separate campaigns, each in a different season, were evaluated for each of the 10 brands. The 10 catalog brands are extremely diverse. They range from upscale home décor to value–priced women's apparel to men's clothes to mid-scale gifts. The 10 brands were chosen randomly and nothing was done to choose campaigns that had any particular outcome. Thus, the study is diverse and random.
Multichannel Retailers use a variety of calculations to monitor performance of customer acquisition campaigns. Some measure overall profit or loss in relation to revenue. Others measure overall profit or loss in relation to marketing expenditures. Others measure percent response. Others measure cost per new customer acquired. All of these are good measures, and each measure may be effective for determining performance for a particular brand. But some of them don't work for a study of multiple brands, where average order size, response rate, and other factors vary widely from campaign to campaign. The case study needs to be based on a measure that applies, regardless of the brand.
The measure we chose is "Marketing ROI". This measure arrives at a percentage gain or loss on a particular campaign, or segment of a campaign. Marketing ROI provides a clear picture of how a campaign actually performed, regardless of CPM, RPM, AOV, response rate, or other factors. If, for example, total Marketing Cost for a particular campaign was $1,000,000 and the campaign's contribution to overhead and profit was negative $100,000 then:
-$100,000 / $1,000,000 = -10%
In this example, spending $1,000,000 on marketing (Total Mail Quantity x Total Cost per Catalog Mailed) produced a $100,000 loss, or -10% Marketing ROI.
Some multichannel retailers seek a profit, or breakeven, from customer acquisition campaigns. Others are willing to invest. A vital question arises for these who are willing to invest to acquire new customers: how much investment is affordable? More specifically, how much can realistically be recovered from profit on subsequent orders, leaving the brand with acceptable or excellent long-term value?
Marketing ROI provides a common measure that is easy to understand. Most circulation professionals would agree that breakeven or profitable prospecting certainly produces good LTV. Prospecting at a 10% initial loss usually does too. What about a 20% initial loss? What about a 40% initial loss? Probably not. The right answer may vary from client to client, but Marketing ROI provides a common measure for comparison.
In order for us to accurately calculate Marketing ROI our clients had to provide two actual costs:
1. The cost of preparing and mailing the catalogs. Essentially, they had to provide accurate CPM (cost per thousand catalogs mailed, including postage, list, paper, printing, and other costs).
2. The cost of fulfilling orders, expressed as a percentage of revenue. This includes the cost of the merchandise sold and the cost of picking, packing, and shipping orders.
Wiland provided revenue as recorded for the campaign in question in our cooperative database. We verified with our clients that the Dollars per Book (Total Revenue / Mail Quantity) that we calculated is the same as the client's record of what the campaign produced.
Thus, we got the revenue right and the costs right, which means our calculation of Marketing ROI is accurate for every campaign studied, and for every model segment of each campaign.
Once we confirmed the accuracy of the facts, we were able to make the following calculation:
Revenue – Marketing Cost – Product & Fulfillment Cost = Contribution to Overhead and Profit
Then we calculated Marketing ROI, as follows:
Contribution to Overhead and Profit / Marketing Cost = Marketing ROI
Each of the 30 campaigns was ranked using a 20-segment Wiland custom response model. The model utilized transaction variables as they existed on the database as of the original merge/purge list cutoff date (Time Zero) for the campaign being analyzed. Thus, the model segment definitions and their associated results in the analysis accurately reflect the model's predictive capabilities. Segment 1 contains the best prospects—the ones our model suggests are most likely to buy. Segment 2 contains the next best prospects, etc. The worst prospects are typically in two places: the bottom segments, and the portion of the net file that is comprised of prospects whose overall transactional history suggests they are not good prospects for any offer. At Wiland, we filter out this latter group so that they are not eligible for promotion.
Results are presented for each of the thirty campaigns, and each of the 20 model segments were placed into 5 aggregate groups to help facilitate the discussion of results:
Results are shown below in the two tables below. Table 2 presents Entire Campaign results for each of the 30 campaigns, showing Total Mail Quantity, Total Revenue, Total Cost (Marketing + P&F), Profit (Loss), and Marketing ROI. Table 3 presents the Marketing ROI calculation only for the Entire Campaign and each of the Segment Groups.
The most significant high level observation is simple but profound: every brand wasted money by mailing to Segments or Segment Groups with weak predicted Marketing ROI in every campaign, without exception. The amount of waste varied significantly from brand to brand and from campaign to campaign, but wasted marketing budget was present in each of these 30 campaigns.
- Waste permeated the bottom Segments. We expected to see poor performance in the bottom Segments, and we did. All 30 campaigns in Segment Group 5 had negative Marketing ROI, as did all but one campaign in Segment Group 4. We also saw negative Marketing ROI in 23 of the 30 campaigns in Segment Group 3, and 19 campaigns in Segment Group 2. Even in Segment Group 1 there were 4 of the 30 campaigns with negative Marketing ROI, but the losses were all small—likely acceptable in new customer acquisition.
- Marketing ROI declined significantly at the bottom. In Segment Group 1, no campaign had Marketing ROI below -20%. In Segment Group 2, 10 of 30 campaigns had Marketing ROI below -20%; this number increased to 18 campaigns in Segment Group 3, 23 campaigns in Segment Group 4, and a whopping 28 campaigns in Segment Group 5.
- Low Marketing ROI at the bottom can ruin otherwise good results. We observed Marketing ROI below -40% as early as in Segment Group 2, and in Segment Group 5 nearly half of the 30 campaigns had Marketing ROI below -40%. When negative Marketing ROI appeared in middle segments, it created drag that compromised the success of the entire campaign. In fact, 5 of the 6 campaigns that had Marketing ROI below -40% in Segment Group 3 had negative Marketing ROI for the entire campaign, even though they had positive Marketing ROI in Segment Group 1.
- No merchandise category escaped the Marketing ROI decline. Whether selling clothing, gifts, jewelry, or home décor items, we observed the same trend. Strong performance in top segments ultimately was dragged down by poor—and sometimes abysmal—performance in bottom segments.
- Even the strongest marketer can benefit from Marketing Budget Optimization. Multichannel Retailer 4 had positive Marketing ROI for all campaigns in Segment Groups 1, 2, and 3, and Entire Campaign results were strong for all 3 campaigns. Even so, Segment Group 5 shows significantly negative Marketing ROI, and it is unlikely that there will be strong long-term value from new customers obtained from this Segment Group. Ultimately, the funds spent on mailing to those prospects would have been better spent elsewhere or simply added to the bottom line.
Evaluating Campaign Results
Reviewing the results of an entire customer acquisition campaign, we typically see either a positive Marketing ROI or a slightly negative Marketing ROI. In such cases our client typically determines that the campaign results are good; they are worth a small overall loss, because that loss can be made up with second orders from the new customers that the campaign produced. However, evaluation of results by Segment Group reveals that the results could have been much better—and often profitable. Once again, a study of Marketing ROI provides great insight. The following tables show the summarized results of all 30 campaigns; by combining the results of stronger, average, and weaker campaign results, we gain an unbiased view of Segment Group results.
Entire Campaign Results: Acceptable. On these 30 customer acquisition campaigns, the overall result is a 0.27% Marketing ROI. On average, the campaigns had a 7.60% positive Marketing ROI. The median Marketing ROI of -2.38% is actually a better measure because the average is highly influenced by a few very successful campaigns. Many would view these results as excellent; the goal is to gain customers, and to do so with a slight gain or minimal loss is a great achievement. Stopping here, though, robs us of the richness of knowing the full story.
Segment Group 1 Results: Fabulous. The overall results of Segments 1-9 tell an awesome story. Here we see that 41.49% of the mail sent generated 58.33% of the Total Revenue. And there is a bonus: these customers are very likely to have excellent long-term value. Total Revenue for these segments was $22 million, creating a Profit of just over $3 million—for customer acquisition campaigns! This works out to 40.84% positive Marketing ROI in total. Some marketers experienced an even higher Marketing ROI, because the average of 54.08% Marketing ROI is absolutely stellar.
Segment Group 2 Results: Mostly Acceptable. The overall results of Segments 10-12 show what we expect. New customers are obtained at a loss, but a moderate loss that most marketers would consider acceptable. The slightly negative Marketing ROI of -6.58% will be earned back with new customers' second order, and thus the loss will be worth it in the long run. Ten of the 30 campaigns actually were profitable in Segment Group 2. The median performance of -12.08% Marketing ROI is very likely to produce good long-term value for most marketers.
Segment Group 3 Results: From Acceptable to Concerning. The overall results of Segments 13-15 show a continued decline, and indicate that these Segments generally perform poorly. The Mail Quantity for these Segments is approximately the same as those from Segments 10-12, but this volume generated substantially less Total Revenue. For a few brands the result in Segment Group 3 is acceptable, but for most the results are more like tumbling toward the cliff edge. Overall, the Marketing ROI is -22.82%, which most marketers would consider unacceptable, even for customer acquisition. For some, the results are even worse, bringing the median Marketing ROI to -26.55%.
Segment Group 4 and Segment Group 5 Results: Absolute Loss. The overall results of Segment Group 4 (Segments 16-18) and Segment Group 5 (Segments 19-20 and the prospects Wiland considers to be "Not Eligible") show a staggering decline from Segment Group 3. Taken alone, these Segments perform so poorly that they drag down the results of the entire campaign. Only a very rare marketer could look at these numbers and consider a Marketing ROI of -38.04% or -40.10% to be acceptable for a customer acquisition campaign. Here the losses are so large that they should be considered intolerable, but they hide at the bottom of a net file and get covered by good results at the top unless exposed. And a key point must not be overlooked: half of the campaigns are worse than the -40.98% and -37.35% medians. For the brands in the bottom half of the performance, the need for Marketing Budget Optimization is extreme.
The purpose of the case study is to evaluate the validity of the following theory:
The clear conclusion is that the theory is valid. Our case study strongly suggests that the vast majority of multichannel retailers that mail a catalog to prospects are wasting a significant portion of their budget because a meaningful portion of the names that survive merge/purge actually are poor prospects that produce a terribly negative Marketing ROI.
Based on our conclusion that marketers are wasting significant portions of their marketing budgets, we have several recommendations. They apply universally to all multichannel retailers that are mailing a print catalog to prospects.
- Recommendation 1: Make the strategic decision to work closely with Wiland to establish excellent brand-specific optimization methods. Regularly optimizing merge/purge net files rather than mailing to every prospect that survives to a net file is a big decision. It is a strategic shift that has potential to add margin points to a brand's bottom line. It also has potential to alter arrangements with agencies, list brokers, exchange partners, printers, and others. It should be well understood and carefully tested and implemented over time. But, very importantly, it should not be delayed because it will take time to understand it thoroughly. It really does have potential to add margin points to a brand's bottom line, so it should be given prompt, significant attention.
- Recommendation 2: Provide Wiland with absolutely accurate cost information so that Wiland calculation of Marketing ROI is accurate. Some multichannel retailers are careful to provide accurate information to Wiland; others are not. If a client tells us their marketing CPM is $380 when it is actually $460, or tells us that product and fulfillment cost as a percentage of revenue is "about 50%" when it is actually 58.43% then our calculations are going to be wrong. We don't want to make optimization recommendations based on calculations that are wrong.
- Recommendation 3: Have Wiland perform an analysis that establishes the specific Marketing Budget Optimization opportunity for a specific brand. No brand should use the specific results from this case study to determine the brand's cutoff point for deciding which names from a merge/purge net file to mail and which not to mail. The optimum cutoff point varies from brand to brand. Have Wiland perform an analysis that reveals specific Marketing Budget Optimization opportunity for a specific brand.
- Recommendation 4: Test to prove value before fully implementing Marketing Budget Optimization. Wiland believes the case study to be absolutely valid, and our experience is that careful research is usually confirmed by "in the mail tests". However, this is not always the case. As soon as Recommendation 3 has been successfully accomplished, split upcoming net files into groups based on the Recommendation 3 analysis. Mail everything (or substantial test portions of bottom segments). Track results to determine whether bottom segments actually produce poor results.
- Recommendation 5: Confirm validity of the strategy on an ongoing basis. After Recommendation 4 has confirmed that bottom segments perform so poorly that they should be dropped, drop most of them and either save the budget (let it fall to the bottom line) or redeploy most of it to more productive marketing efforts. But, go ahead and market to a statistically reliable sample from the low segments, just to confirm the validity of the optimization strategy.
- Recommendation 6: Help change the industry. Mailing to the entirety of a merge/purge net file, or to all except a small portion dropped by the merge/purge service bureau, is the norm today. It is a bad norm that is costing multichannel retailers a lot of money. The losses it is causing extend beyond the initial customer acquisition campaign that produced the initial loss—reaching into the future, it is creating more losses, because new customers acquired from these weak segments tend to be poor customers that buy again at low rates, producing poor long-term value. Yet, the industry has accepted "mail to the entire net file" as the norm today. This needs to change. Internal circulation staff, agencies, list brokers, list managers, and list owners need to shift their thinking. Study the subject carefully, and help make the case for change. Wiland can implement this needed change immediately because we have invested heavily to streamline the process, but optimum results will come from arranging to pay only for names that pass the optimization process. Making such arrangements will take more work on the part of brokers and circulation planners. They need support and they need to be compensated for their extra effort. See the article Marketing Budget Optimization™ for Multichannel Retailers for a more thorough discussion of this topic.
- Recommendation 7: Make the right decisions concerning what to do with the budget saved by using Wiland Marketing Budget Optimization. Clearly, the money saved by Marketing Budget Optimization should either fall to the retailer's bottom line or be re-deployed in a more productive manner—or some combination of the two. But which is the optimum choice, or how much of each? The right answer will not be the same for every brand. There are a lot of possibilities. For some multichannel retailers the saved marketing budget can quickly and easily shift to different, superior direct mail prospects (read about Balance Modeling), making the direct mail channel even more productive. But often, although Wiland could easily make up the entire balance, the easy answer often is not the best answer. Without question, Wiland will have some high quality prospects available at the time of optimization that were not available prior to merge/purge, simply because hundreds of clients provide updates and Wiland posts new transactions constantly. Thus, Wiland Balance Names should be part of the plan. But we recommend more diverse testing to find the best use for the saved marketing budget. Here is a conservative approach that may be a good place to start:
Step 1: Don't spend all of the savings on other marketing programs. First, pay for the optimization out of the savings. Then, let 20% of the savings fall to the bottom line. Spend the remainder of the savings on other marketing programs, but not all on the same program. Diversifying the spend on a variety of marketing programs will increase the chances that something far superior to mailing to the omitted weak prospects will be identified. One approach to accomplishing such diversified testing is outlined in the subsequent steps below.
Step 2: Test Wiland Balance Names. Doing this is the best first test choice, for two reasons. First, Wiland won't oversell balance names. In fact, we will suggest that only a portion of the savings, perhaps 20% to 40%, be devoted to Wiland Balance Names (test first, increase volume if successful). Second, Wiland Balance Names are the best first test choice because Wiland's price policy makes this suppress and replace strategy very affordable. We charge progressively less for optimization as a brand takes more and more balance names. By operating this way, Wiland helps assure that the majority of the benefit of optimization stays with our client.
Step 3: Test something the brand has been wanting to test but didn't have budget. Whatever it is, if it can be tested for 20% or less of the saved budget, test it.
Step 4: Test Wiland's Digital Co-Targeting service. Most brands that test our digital co-targeting service discover that it not only pays for itself—it increases the performance of direct mail. Basically, we find prospects when they are online and serve them digital ads that reinforce the much more expensive catalog that they've just received or are about to receive. Even a small increase in direct mail responsiveness can make a significant difference in the bottom line. Test it with a portion of the savings.
There are many more test possibilities. Our purpose here is not to list every possibility, but to recommend a wise strategy: use the savings to pay for optimization, let some money fall to the bottom line, and test opportunities that have potential to substantially outperform the weak prospects dropped by Marketing Budget Optimization. Test, and find the ideal answer over time.
The analysis, conclusion, and recommendations in this case study are substantive. They call for a shift in industry norms. The shift they call for is good for multichannel retailers. The full positive impact can be dramatic: more margin points on the bottom line for almost every multichannel retail brand that mails catalogs. And there are bonus benefits: higher quality customers that deliver superior long-term value. To achieve these benefits, multichannel retailers and their agencies, list brokers and other service providers will need to make adjustments. Without the adjustments, multichannel retail profit margins on catalog mail likely will slip lower because costs keep rising. The potential benefit is huge. Even though it takes some adjustment, everyone needs to get on board and take advantage of Wiland's ability to identify the bottom portion of any net file—before the client's marketing budget is wasted.
Download the PDF of this Case Study here.