Marketing Ops – Great Campaign Structure Advice and My 2000 Millicents

Posted by:

When I led marketing operations and analytics at Siebel, drove the implementation at Oracle, and was a product marketing guy at Unica, one of the most critical elements of measuring marketing ROI was campaign structure. Since I’ve spent the last three years writing about big data and data science at Pivotal.io, I see even more reason why quality and granular attribution in this area becomes critical. I hope to make a bit of a series on this topic. For now, I’ll keep it high level – note – these are more stream of consciousness thoughts and deserve the time and attention to review, hone, and add more clarity.

Recently, I ran across some great advice on Google AdWords campaign structures (see the quote the following slides). It was a great article and really resonated with me – someone who had to build 50+ page slide decks on the quarterly results of marketing, sliced and diced by every variable possible as shown in the slides below (and also before Hubspot, Pardot, and Marketo existed). By the way, this is what sold Oracle on the fact that we did marketing operations better than they did.

or

This Campaign Structure Advice makes a Critical Point

If you are like most search advertisers, there probably isn’t much rhyme or reason to your campaign structure. A good portion of the PPC I manage involves taking over accounts that other people have been built out (and mismanaged). The one universal mistake I see over and over again is misaligned campaigns.

If you are like most search advertisers, there probably isn’t much rhyme or reason to your campaign structure. A good portion of the PPC I manage involves taking over accounts that other people have been built out (and mismanaged). The one universal mistake I see over and over again is misaligned campaigns.

What to consider with Campaign Structures in General

The chart above really paints a picture of this point. On top of all these attributes, each campaign has ad groups, keywords, ad text, landing pages, etc. There is a fairly complex campaign structure – then you add things like anonymous users and other audience segments, it gets even more complex. Now, add regions, different channels (like email, direct mail, YouTube videos, events), budget structures and bids. Now, we are talking about complexity. How do we deal with complexity? Do we ignore it and just do our best? We usually do. But from a critical thinking perspective, we can’t call it a precise, accurate, and sufficient measurement to depict reality or compare things. We can only say we are doing our best.

First and foremost, we have to start with the end result in mind – we want to have reports that are useful and give us the data we need to make better decisions for marketing portfolio optimization. This means we have to understand the inputs and outputs of the system. Here is a test, if you don’t know where every data point comes in to the resulting report, then you are lost and making bad decisions. You must have a clear picture of every input and output. In addition, you should know the accuracy, sufficiency, and bias in the data and understand any transformations or calculations, but that is not covered here.

Campaign Structures: High Level Expectations

After we have the end in mind, the link between budget and revenue becomes the most critical connection when it comes to analyzing marketing ROI (or any metrics) and forms the basis for campaign structure. Without linking these things, there is no way to really measure the result of the marketing portfolio. At the highest level, we can say that marketing spent $1M and sales closed $2M, so every marketing dollar spent produces two in revenue. But, we can see how the highest level provides an insufficient view for optimizing spend. Now, most people can measure at a more granular level for a one off campaign, like a direct mail drop with a measurable response code. But, to optimize spend across the marketing portfolio, across campaigns, or focus more dollars on certain segments, messages, regions, products, or sales teams, we really need to be able to compare things more effectively within the context of budget and result. We need to go a bit deeper.

A deeper measurement is what the CEO and CFO wants, but it isn’t nearly as simple as counting expenses by department or revenue growth per business unit. If every channel, campaign, and campaign child element captures attributes and metrics differently, we aren’t really able to compare things in an equivalent way. Still, some insightful measurement is possible and better than no measurement at all.

Minimally, we want to effectively measure an executive level of spend and result – but everyone should know that we cannot go deeper without further attribution, and, eventually, metrics become un-equivalent also making them un-comparable. For example, you cannot compare an email click to a website click and assume they are the same thing – clicks have types. The more depth desired, the more attribution required. The comparison we want, the more common attributes must be. As we graduate into more advanced attribution and analytics, we realize that marketing is not a funnel like sales – people are touched by marketing a myriad of ways and single-event attribution isn’t really a sufficient view of cause and effect. As we move to more analytical algorithms, like clustering, time series analysis, or machine learning, we can also be hamstrung by a lack of campaign structure.

While I am speaking in terms of a marketing department level portfolio, AdWords is broad enough to be it’s own portfolio within a portfolio.

Campaign Structures: Links, Attributes, Facts, and Dimensions

 

In the world of traditional analytics, all reports come down to facts and dimensions – the numbers and the attributes of those numbers. Facts are things like  “budget,” “number of clicks,” or “amount of revenue” and attributes are things like “product type,” “region,” “keyword,” or segment. Well, this is simple in theory – we just want to count how much of a certain thing – but we can have 100s of metrics and 100s of attributes – not to mention the fact that marketing has it’s own structure for campaigns and measurement is both highly temporal (longitudinal measurement of time) and multi-dimensional (a chart in 3D or more), but that is another discussion. Back to the point, we have to be able to measure effectively at a more detailed level, and, to do so, we should know how our metrics and attributes align – how they form a common level. We need to know that we can count on something to be truly comparable – like region in AdWords equals region in email equals region in direct marketing, etc. As we get to more and more granular levels, it gets harder and harder to do across channels, but we can at least do it for each channel individually.

What typically happens is that we can do it at a high level, then people want to drill into things like – is there a certain keyword that produces more revenue than others? Which emails are best? This is when we have to draw a line in the sand for what attributes and metrics can exist across the portfolio and which will be drilled into separately – again it is a what-is-shared-what-is-unshared agreement. We also have to look at things like, which campaigns are ongoing (like a website or PPC) and which are start-and-stop, like an event. This goes back to the first point – what do you want to measure across the portfolio and what is helpful to measure in each channel. To move money around and optimize spend, we have to compare across equivalent things much like a stock. On the drill down side, don’t compare email open rates to landing page bounces.

Let’s get into a bit more detail, after all, we are trying to lay the groundwork for building an effective campaign structure. In a campaign structure, we can have many child elements, each which has more child elements, etc. These elements are often facts and they are associated with the other elements’ attributes. We have to have a clear understanding of this – it is a core design factor – and if you ever want to measure marketing in a longitudinal way – you can’t go changing it all the time – or you will never get there.

The main design principle is that facts and attributes are limited to a certain level in the campaign structure, we cannot measure any more detail than the level it stops at. For example, if we have a budget for a campaign set at the campaign level, we cannot measure the ROI of individual campaign components only at the budget level. This is because the components roll up to a single budget and no budget is allocated to a single component.  To make the point even simpler – if you track the expenses for your rent or mortgage and all utilities in one bucket – and you never capture each detail separately, it is impossible to go back and look at the cost of electricity. You just can’t break it out because you don’t have the metric (cost) for the attribute (electricity) at a lower level. The same with emails – if we structure an email campaign at the U.S. level and send 5 emails (with no regional attribution), we cannot measure each email’s regional results, just the campaign as a whole at the U.S. level.

Looking to the Future of Marketing Operations

In data management and data science terms, we really have to capture as many of the attributes as possible at the most granular level of metrics possible. While we have to group things along the way, it should be a very cognizant and conscious design choice.

To really do it right, we should just make a default assumption – people will want to measure everything possible and ask any question possible.

Instead of going about this a chunk at a time, we can just assume that all attributes should be applied to all facts at the most granular data set possible. It sounds like a bigger hill to climb, but it is ultimately a shorter one and a more effective one. The problem is, we use so many different tools in marketing. Ok, here we go again. Have a common platform and independent platforms – know what is shared and what isn’t.

That will be for another blog post. #ShareThis @adambloom :)

Quote and Image Source: Are You Making These Costly Adwords Campaign Structure Mistakes and Missing Out On Your Most Valuable Customers?

 

 

0
  Related Posts