Thoughts On Multi-Channel Strategy
This article in slightly different form first appeared in Catalog Success, May 2007
Thoughts On Multi-Channel Strategy
Scratching your head over the interaction between your online and offline marketing efforts? Not sure how much to advertise online? Uncertain about the true impact of your catalog mailings? You’re not alone. This essay won’t completely solve these puzzles, but will try to offer some relevant ideas.
How Much To Advertise? Begin With Your Goals
First, let’s assume you’ve already established your high-level financial goals, either for your online program or for the business as a whole. Such goals should be specific, numeric, and time-based. Be sure the whole team understands and buys into these goals, and share your progress towards meeting them each week. Typical goals are P&L-based and include a revenue and an earnings component. Rapidly growing companies may also have a cash flow goal.
Build a planning P&L statement based on these goals. Usually, a marketing P&L stops at marketing contribution, as marketing isn’t typically held accountable for fixed costs and overhead. If your margin doesn’t vary greatly across your marketing efforts, this planning P&L effectively reduces to understanding the tradeoff between marketing expense and resulting sales. I find it helpful to discuss sales in terms of absolute dollars (volume) and marketing expense in terms of a percentage of net sales (efficiency). Expressed as percentage of sales, the marketing spend becomes the advertising-to-sales ratio (“A/S”). Don’t allow different departments to use different assumptions for margin or variable cost (eg. do vendor coop funds reduce marketing expense, or rather reduce cost of goods?). Task your finance team with ensuring marketing and merchandising use consistent numbers.
Your programs in aggregate should meet your efficiency goals, and you should allow your marketing team leeway to cross-subsidize programs within the portfolio at their discretion. If you require every program to meet your target efficiency, the portfolio as whole would be too efficient, leaving potential revenue on the table (Figure 1, at bottom of post). Do dig into the details, making sure your team can explain which laggard programs they’re propping up and why. Good explanations are “testing new opportunities”, “trading some efficiency for increased volume”, or “strategic considerations.” Bad explanations are “we don’t know” or “we’ve always done it that way.” This perspective applies to list rentals and paid search keywords: require the portfolio “works” in aggregate, allow your managers discretion within the portfolio, and occasionally inspect disaggregated performance by list or keyword.
The Advertising vs. Sales Tradeoff Curve
To determine the tradeoff curve between advertising and resulting sales, turn to historic data, or construct controlled tests. Earmark 5% of your marketing budget for testing brand new opportunities. Budget another 5% to test “over-marketing” of existing channels: small selects from marginal lists, small ad spends on marginally converting keywords, mailing a bit below your cutoff to your buyer file, and so on. Modest ongoing “over-marketing” lets you collect critical data on the shape of your advertising versus sales trade-off curve.
The advertising-to-sales tradeoff curve isn’t usually smooth — you either take a list or keyword, or you don’t — and often has a very steep cliff — once you’ve bought the good stuff, the quality plummets, and additional sales come at a huge ad cost.
For modeling purposes, it can be helpful to assume this tradeoff curve is smooth for small changes. The “square root rule” takes this approach, stating that sales increase linearly with the square root of advertising. This assumption isn’t correct for large changes, but it is a useful approximation for small changes. The square root function show decreasing returns to scale, is internally consistent, and offers nice theoretical properties. On our website, we provide a free Excel spreadsheet which uses the square root rule to model the potential effect of slight increases or slight decreases in your advertising spend, as well as suggest the advertising level which optimizes contribution (Figure 2, at bottom of post).
One rule-of-thumb that emerges from the square root assumption is that direct marketers maximize their marketing contribution dollars by spending one half of their effective margin on advertising. For example, a marketing selling products with 45% COGS and 12% other variable expenses should aim for an advertising-to-sales ratio of (1/2)*(1.0 – 0.45 – 0.12), or 21.5%.
Acquisition vs. Retention, Meet Brand vs. Non-brand
Catalogers have traditionally partitioned their marketing efforts into acquisition and retention programs. Many firms ran their acquisition efforts to break-even or below, satisfied to bring new buyers onto the file sans profit for the lifetime value of these new customers and for the top line benefit. When marketing to their active buyer file, catalog firms would only mail catalogs to fresh highly profitable segments, as the resulting orders had to cover not only the cost of marketing to active buyers, but also of marketing to prospects, covering overhead, and generating overall profit.
The tremendous economic importance of their active buyer files caused many catalogers to overestimate the loyalty of these customers. The rise of e-commerce and paid search has totally changed the concept of customer loyalty. In years past, a retailer’s catalog might be the only exposure a household received of merchandise in a particular category. No longer – the web puts every SKU in front of every consumer. In the age of the web and of WalMart, customer loyalty is hard-won, rare, and easily lost.
In paid search, the traditional catalog paradigm of “acquisition” vs. “retention” is replaced by “non-brand” vs. “brand” search. When searchers find you online using your brand name (“Lands End Shirt”, “Harry And David”, “Chase Credit Card”), they are using the search engine like a White Pages. For this search, they have exhibited enough loyalty to find your site by name – regardless of whether they are on your active buyer file or not. In contrast, searchers who find you online though non-brand search (“men’s oxford shirt”, “fruit basket gift”, “rewards credit card”) are using the search engine like a Yellow Pages. They are comparing you to all your competitors. This order is “in play” – regardless of whether that customer is or is not on your active buyer file.
Just as catalogers mail their most recent buyers, online retailers should usually buy search clicks on your brand terms. They are low-cost and high-converting. However, realize the resulting sales are largely non-incremental. (To increase the number of these brand clicks, mail more catalogs.) While you receive lower efficiency on non-brand paid search ads – that is, you need to pay a greater share of revenue in advertising – more of these orders are incremental, capturing market share from your competition.
Order Allocation Rules
In the past, when households received your marketing messages infrequently, allocating orders to customers was simple: give the last marketing communication credit for the order. This rule doesn’t work today, as our messages barrage potential buyers in overlapping torrents. Perhaps the last touch before the order was an email, but the prospect received a catalog the week before. Or the last touch was an email, but the email signup was driven by a paid click.
The irony is that that our increased ability to track orders and marketing – due to the web and better technologies – has decreased our true understanding of response. The multichannel order allocation problem is far from solved. While we as industry struggle to develop the accurate methods for multichannel planning over the next few years, in the interim expose and discuss your current order allocation rules. Do not bury these assumptions deep in your business logic. Bring them to the forefront to determine if your strategic models are highly sensitive to these assumptions. They will be, and that is terrifying. The “right” way to determine the true marketing drivers of customer behavior are ongoing cross-channel hold-out tests, which are difficult to design correctly, hard to implement, slow, and costly.
Multichannel has changed the physics of direct marketing. Search has reduced customer loyalty-via-ignorance. Simple rules of house-file vs. acquisition economics are no longer so simple. Order allocation rules matter, more so than you may think. Track sales and marketing costs at the customer level across channels. Differentiate sales from brand vs. non-brand search phrases. Monitor efficiency just above and just below your advertising cut-off, and start hold-out testing to determine your true marketing drivers of behavior.
Figure 1: Marginal vs. Average Efficiency For Hypothetical Cataloger. If this direct marketer needed 16% A/S to make her P&L goals, she would continue advertising in channels A, B, and C, as the aggregate A-B-C portfolio meets her A/S target, even though channels B and C in isolation do not.
Figure 2: Using the square root rule to model small changes in advertising levels. Spreadsheet planning model available at www.rimmkaufman.com/squarerootrule.