Data Driven Marketing in an Election and Business

Continuing on my trail to promote data as a means of marketing, I turn to the Obama campaign. I know you’ve heard a lot from the Obama campaign, but I guess not as much about the data behind winning the campaign. Yes, the knocking on doors had its effect on the outcome, but did you know that over $500 million of the campaign budget came from online donations. That’s more than twice as much (152 million dollars) as came from offline donations. The video in this blog post explains some of how data was used in order to create a winning team for the Obama campaign.

I will extract some of the more interesting remarks made by Dan Siroker below in this post, and highlight how this relates to your business.

1. Define success in terms of measurable metrics

First they break the ultimate goal – ie. elected dude – into stuff needed to be in place for that to happen. In terms of the US election, those sub goals were defined as:

  • Recruit volunteers
  • Register voters
  • Canvas door-to-door
  • Make phone calls
  • Raise money

All of which is needed if you are to win an election on US soil. In terms of business, the ultimate goal might be a profit increase or an end year sales goal. In order for that to occur you need to:

  • Recruit great staff
  • Get leads to sign up for your services
  • Get retailers onboard
  • Make phone calls
  • Sell products

I think you get where I am going with that one. In the video, Dan focuses on the last one and tries to make some sense out of how they used data in order to win an election.

He starts out by showing a “user funnel”. A user funnel is basically a behavior of your visitor to the website. This means, the way your user behaves, clicks, waits, returns to your website. Usually you put up a goal which you want him or her to complete before they leave. In the video the ultimate goal is for the visitor to donate money. But they realize that they might be better off getting people to sign up for the e-mail first and then get the donations secondly.

So the initial funnel is: Website visitor -> E-mail signup -> Raise money

What they realized early was that they had to come up with quantifiable success metrics that would then help them optimize the website for the purpose of getting as much money per visitor as possible. They used some key metrics including cost per click leading to the website and e-mail sign up rate that led to two final KPIs in order to measure the success:

$ per page view

The dollar per pageview was determined by looking at how much money a user was likely to donate given they had visited a specific page. Once they know this number, they are able to calculate how much they can spend in order to get a visitor to that specific page. This is much like you should be thinking of your different pages on your website. How much money does this page make me? Not – how much money does my website make me.

You should think in terms of basket value. If a visitor has visited a specific page and you can determine that every visitor who visits that page either buys more or less from you, you can also determine what you are willing to pay – with positive ROI – to get a user to visit that page. Simple enough. Right?

$ per e-mail

This was virtually the same thing. But I will go through this second step a bit in more detail as he sort of runs it through a bit too fast in the presentation.

  • The number of recipients per e-mail
  • The opening rate of the e-mail times the click through rate
  • The conversion rate, ie. the people who donated from that e-mail click
  • Multiply that all together times the money they donated

2. Create a function for analysis purposes

Which gives you a rough amount of dollars per recipient. So, what this really means is the following example. Let’s say you have a send list of 100 000 recipients of which 20% open the e-mail. Of those, 50% click the link and of those 5% donate a total of 100 000 USD, we get the following formula.


X = Dollars per e-mail

R = number of recipients

O = Opening rate

C = Click through rate

T = Transactions or number of conversions to donators

N = Total profits

This example: 100 000x * .2 * .5 * .05 = 100 000 USD

x is in this case the dollars per e-mail.

The importance of stating it as a function is that you are able to see how changes in one effect the rest of them. Here, you should really use regression analysis which you can really skip learning and simply use one available tool out there such as SPSS.

2 a. Analyzing the effects of changes in subscribers

Cause, think of it. Let’s say you only look at the send list subscribers and the outcome in USD. Let’s say you increase the number of subscribers only to realize your total donations remain the same. This means the dollar per user has really decreased. What you have to ask yourself is not only if you have the right subscribers, but if you are feeding them with the right types of messages.

So, although your send list grew, your messages do not make the income grow. Unless you look at it through all the steps you will be unable to make changes along the way necessary to increase your end game profits. It doesn’t matter how much you spend on getting more people to sign up unless you are able to figure out how to make more of them convert.

2 b. The case of the opening rate

Continuing with the example. Let’s say your opening rate has drastically decreased. Let’s say it has halved. If that is the problem, then perhaps your increase in subscribers might have effected your white list or delivery rate, or perhaps there is something wrong with the subject? All of which you need to analyze if you realize that it is the opening rate that has gone down.

Or perhaps it is the click through rate that has gone up or down. Or perhaps it is the conversion rate into donators that has gone sour. Either way you will know if you split each step of the user funnel into parts rather than looking at the nominal before and after.

2 c. The case of the bad subscriber

One interesting scenario occurs when you have improved your opening rate and your CTR, but you realize the visitors still aren’t converting into donating or paying customers. Or perhaps your donations even decrease. In this case you have to look at where the subscribers came from, what KIND of call to action you are using etc.

It is just like sourcing for the wrong kind of traffic from the search engines. You can get all the traffic in the world, but just because you rank number one for Britney Spears or Lady Gaga, doesn’t mean they will buy more of your kitchen hardware. Same goes for your e-mail campaigns. If they signed up thinking they would get images of pink zeebras – something I would sign up for – they are not very likely to buy your kitchen supplies either. Same thing goes for the “free stuff” e-mail sign ups out there. Sign up and get free… whatever.

Although a lot more people might be inclined to sign up, they might be less valuable to you in the long run as they do not buy. So although your cost per conversion into subscriber has gone down, your net effect on profits has been severely hurt.

What I am really trying to say is that a lot of subscribers or traffic might not be what you want. You might actually want less subscribers, or traffic, that you pay considerably more for, because they make you more money. A lot of traffic or subscribers might actually cause you a negative ROI. I don’t know if I am making my point here, but I hope I am.

3. Focus on your weakest links

Once you have your analysis methodology setup Dan argues says you need to start focusing on your weakest links. When he started working with the Obama campaign, they had a very strong number of both visitors who donated as well as subscribers who donated. What they didn’t have was enough subscribers. Or moreover, the subscription rate was too low, meaning, out of every 100 visitors who came to the website, the conclusion was, more of them should sign up to the e-mail send list.

3 a. Don’t get fat, stay smart

In fat markets, usually in early stages, where competition is not fierce, vendors or retailers generally do not care about fixing the weakest links. They work with the stuff that makes them the most profits. Ie. the strongest end of the rope. They usually believe that it is where they are strong that they can make the most money. And perhaps they are right at the time being, but I have seen numerous occations where they end up crying in a corner as a result of it. Cause although the sweet taste of sugar might blind you from looking over your shoulder, you SHOULD know that were ever there are profits, competition will follow.

If you do not have your stuff in order, your weakest links will come to haunt you. (I really hope some of you readers out there know exactly what cases I am talking about… yepp… I hate to say “I told you so…”…) One thing with digital is that it is never to late to change. If your competitors managed to move passed you, there is still room for you to retake your position.

So. In the video, Dan takes you through a series of multivariable tests they did on the website. They optimized for an increased subscription rate going from an 8.26% subscription rate on a page looking like this:

They started out by optimizing the text on the button as well as changing the image around. They had some videos in the campaign that you really have to watch the video inserted above in order to have a look at again. However, if you forward to about 18 minutes into the video this example should be there.

3 b. What to take away from the tests they did

The button saying “Learn more” was victorious, for the media image, the family message did the best. One interesting thing is that every single video did worse than any of the images. Perhaps not a general conclusion to draw, but you should really think about where and when you insert your call to action in your video. Always think about the bottom third – the space where you can insert a banner with a call to action in any video… just like YouTube does forcefully with a pop-in, so can you in your actual video – and you need to remember to put a call to action there. Make it evident so that even the user who does not want to look to the end of the video, knows what to do next.

In the end, they increased their conversion rate by 40%. From 8.2% to 11.6%. Over time, think about what such improvements could do to your business.

3 c. Why epic content sometimes isn’t enough

So, as you can see from the example, the probably most compelling video gets NO traction what so ever with regards to conversion rate. You MUST realize that even the best of content put in the wrong context, makes absolutely no sense. When I discuss gut in one of my previous posts on data driven marketing I also discuss the fact that although your gut feels something you HAVE to be ready to capitulate to the mere facts of data.

Thus, you can make a really compelling video, but if it actually hurts your conversion rate as it pulls eyeballs from the goal you are trying to get your visitors to perform, then TAKE THAT SHIT DOWN, cause it is really shit if it doesn’t help you achieve your goals.

And YES that video might have been what made a person come to your website in the first place. But it might not be what makes your visitor convert. (Although you probably must work with maintaining scent so that they know that they’ve come to the right place.)

4. Never over-generalize

In the video, about 27 minutes in, there is a really great lesson to be learned which is Don’t over generalize. In modern digital marketing and analytics you can actually serve different content to different users depending upon their behavior. If you spend time doing this, which would cost you probably one tenth of your standard banner budget – which is completely useless if you don’t do any of this – you should be able to increase your profits significantly.

Here are some of the results from the video showing different users acting on different calls to action:

Thus, depending upon where and who you are as a visitor, you react differently to the messages being fed to you. Remember that this is a very simple example and that the sophistication in the way you decide to do it can be even greater. Let’s say you would run different calls to action on returning vs new visitors. What if you would do it for visitors who were returning and who had signed up for something previously. What if you could increase profits of a second purchased based on changed copy in a button fed to people making small initial purchases.

All of which is possible, but you have got to optimize for it.

5. Take advantage of circumstances

What he argues is simply that you should take advantage of all those small things that happen out there even though you haven’t planned for them. Moving away from big idea marketing you have to be able to make changes and move on circumstances that occur in your world. If an opportunity appears, just act upon it with a tactical movement.

Sometimes something that happens in your surroundings might effect you as well. If your competitor starts to run TV-ads, have a look and see if they bought the PPC matching the increased search volumes that come as a result. Perhaps they optimize their budgets for the day whilst their TV-spot run at night. You might be able to counter their offer with a really nice counter offer so that when someone search for the competitors product you put yours on top of the search engine with a cheaper price, a better delivery circumstance or simply… you’re there… whilst your competitor burnt all their money on creating the traffic and not taking care of it.

6. Question assumptions

Look at the example about 31:30 into the video above. This is basically asking you to think about your value proposition. One of the key drivers into making someone convert on your website is to offer them something that they want. I know this sounds extremely idiotic to say, but many of you still have trouble offering a customer anything at all, yet alone something they actually feel tempted to buy.

Change it around. Work towards an increased ROI rather than a number of sold items. Look at your costs and what brings you the best profits. Then question it all again and test it until you are blue and red from starting to doubt yourself to much.


I think this video is great! It is not good, but it is great. The only thing that concerns me is that there are a million people talking about the Obama campaign as something successful with regard to social media, but this video only has 7000 something views. That is shameful!! Now go ahead and share it as it should be seen.


1 thought on “Data Driven Marketing in an Election and Business”

Leave a Comment