A Decision Making Model for Spend in Marketing Channels

A Model for Optimizing Marketing Spend

Recently I was teaching my marketing course at the BMCC and re-examined with my students the concepts of BCG’s famous growth-share matrix as well as the experience curve.  While only moderately related to the marketing discipline, these concepts are in the mandated text book I teach from, and they already had a full set of power point slides ready, so in the spirit of laziness, it was too easy not to teach.

However, this did get me thinking that there are not enough models in the marketing decision making process such as those that have been in practice in management consulting for decades. Marketing spend is too often determined via a HPPO or an arbitrary amount of traffic that the company wants to achieve.  Yes, many companies have spreadsheet after spreadsheet, or data visualization tools, but many do not understand how that data helps them to make a decision.

 So I sat down, embraced the remix and re-worked some time honored management consulting models for marketing purposes. Specifically, I created a model for how to optimize spend across marketing channels.  The basic principle is simple and is based on the law of diminishing returns – for every marketing channel there will be a point where for every dollar of extra marketing spend, there will be a diminished return on conversions (Figure 1).  Conversions being defined as actions that you want a visitor to take such as leads, downloads, or of course, direct sales. This is the case because, inevitably, an individual channel or keyword will become saturated with your ads and reaches more and more people that are not interested in purchasing your product, or the same people over and over again that have no interest.

The individual points along this line that is being plotted are known as “effective cost per acquisition” (eCPA), or the average cost needed to obtain a conversion through that channel. eCPA can be a little tricky in some cases to determine because of the difficulty in attributing which channel led to the conversion. A lot has been written on the subject including a post from myself and an excellent post from Avinash Kaushik, so we won’t explore calculating eCPA here, but will reference it throughout the post.

eCPA is a useful metric for determining at what point an advertising medium is profitable. For example, if I sell a product at $12 and have a Cost of Goods sold (COGS) on that product of $6, then I better be able to achieve an eCPA of less than $6. Anything over $6 is not profitable – it is costing me too much in advertising to sell it profitably. Let’s refer to a channel that, no matter what you seem to do, is not profitable as a Dog.

Campaign Creative Optimization

Spend is not the only lever that can be manipulated to alter a campaigns performance and ultimately result in a better eCPA. Testing ad copy/creative, landing pages, and website conversion can result in a better performing campaign (see figure 2). Again, some great work has been written describing how to do this process, like this one and this one. Let’s refer to a channel that has only been thoroughly optimized for campaign criteria or a channel that hasn’t been touched at all as a Question Mark.

Campaign Spend Optimization

In some cases the optimized eCPA and acceptable eCPA are not the same (see figure 3). I call the gap between the two (optimized and acceptable) the spend-profitability gap. The question is, should we close the spend-profitability gap by spending up to the point of acceptable eCPA? The answer, for an individual channel, is no. The money that would be spent to reach the Acceptable eCPA should be diverted to a new channel that has not yet reached the point of Optimized eCPA. This approach maximizes conversions and profit. Let’s refer to a channel that has been only optimized for spend, and is returning a good resulting eCPA as a Star.

However, after optimizing every channel (which is difficult, time consuming, and quite frankly unlikely) it may make sense for overall spend to be increased across all channels to reach acceptable eCPA, depending on the organization’s goals. If an organization is trying to maximize market share, for example, they may be less concerned about profit and want to simply increase sales to their maximum level (think Amazon).

While optimizing EVERY channel might be nearing impossibility, you can certainly get close to “as good as it’s gonna get” with individual channels in both spend optimization and campaign optimization. That channel is producing consistent and profitable sales. It doesn’t need more money not does it need time and effort on optimizing campaign details. This specific channel, we’ll refer to as a Cash Cow. It is important to note that today’s cash cow may not always stay that way. Channels require monitoring to ensure that they remain performing well.

Applying this concept to decision making

The BCG growth share matrix utilizes the concept of categorizing business units/divisions in the same manner as we have here, Dogs, Question Marks, Stars, and Cash Cows. Using this same approach, we can categorize large numbers of channels and make decisions with what to do with them.

  • D    Dog – A channel, after attempted optimization, that has an eCPA higher than the acceptable (profitable) eCPA. Do not spend on this channel. Reallocate those funds to others.
  • ·         Question Mark – A Channel that has not been attempted to be optimized, or has only been optimized for campaign criteria, such as conversion percentage and Click through rate (not spend or eCPA). Run test spends here to understand how incremental increases affect sales.
  • ·         Star – A channel that has been optimized for spend and eCPA. This channel is worthy of time and effort towards campaign criteria optimization, but is well optimized for spend. Maintain current advertising budget, and work to increase conversion through campaign criteria.  
  • ·         Cash Cow  - A channel that has been pretty well optimized for both spend and campaign criteria. Not much left to do here, but check in, maintain spend and ensure continued performance.


There are several ways to apply this concept. One approach is to segment keywords (or keyword groups) from a search campaign, such as Adwords, into the categories. I’ll demonstrate using my company hotsauceadventureclub.com (HSAC). Hot Sauce Adventure Club is a hot sauce of the month club, providing flavorful Hot Sauces sourced from around the world. Some keywords that HSAC targets are:

·         “Hot Sauce” – Great eCPA, Optimized landing pages and campaign creative. Not optimized for spend.

·         “Hot Sauce Gifts” - Good eCPA, Optimized for spend, but have not tested landing pages or creative.

·         “Foodie Gifts” – Fluctuating eCPA due to budget constraints, uncertain what increased budget would do. No attempt at campaign creative optimization.   

·         “Food Gifts” – High cost per click and competition has made this a toxic key word to bid on. It has a very high eCPA despite ok conversion.

·         “Hot Sauce Club” - Great eCPA, Optimized landing pages and campaign creative. Optimized for spend.


STAR – Invest in campaign optimization


Hot Sauce Gifts

QUESTION MARK – Run test budgets


Hot Sauce

Foodie Gifts


CASH COW – Sit tight and monitor


Hot sauce Club



DOG – Eliminate this keyword/channel


Food Gifts


Another way to apply this concept is to specific websites and channels altogether. Programmatic can help a great deal with these decisions at scale, but there is still great value in seeing how it’s broken out. Here is what a possible matrix may look like for a generic advertising campaign:


STAR – Invest in campaign optimization


NewYorkTimes.com Banner Ads

Facebook.com Banner Ads




QUESTION MARK – Run test budgets


NewYorkTimes.com Advertorial

Retargeting Campaign

Twitter Sponsored Ads

Timeout.com/San-Francisco Banner Ads


CASH COW – Sit tight and monitor


Affiliate Marketing Channel

Digg.com Sponsored Ad

Timeout.com/New-York Banner Ads



DOG – Eliminate this keyword/channel


Foxnews.com Banner Ads

YP.com Listing