Q:              What did you learn from the experience of 2000, that you carried into 2004?

SHAW:       I think we learned — this is going to sound obvious — but to trust the data, that dynamics do change from election to election.  That states for idiosyncratic reasons — sometimes it’s population shifts, sometimes it’s change in the political climate at the top of the state.  But they move around a little bit.  So while there is a rough structure, and a consistency to the rank order, things vary from year to year.  In our election, West Virginia went from being a — you know, a pie in the sky state, to being a cornerstone to the Republican coalition.

Later on in 2008, I think the Obama people probably had the same realization with Indiana, which was — there was no reason for them to have targeted Indiana, except they kept having data showing that it was a pretty good target, and that was a reflection of the demography of Indiana, which has changed a little bit, the presence of an African American candidate [00:24:00] at the top of the ticket, and things of that nature.  We had a couple of those realizations in 2000 that carried forward to 2004.

Mechanically, the other thing I would suggest, and returns to the question, the fundamental question I asked earlier about how do you target, and that is that what we ended up doing was, once we had identified a set of battleground states, the targeting actually shifted from state-focused to market-focused.  So in other words, you were either a battleground state or you weren’t.  Once you were in the battleground state range, we listed all of the media markets within those subset of battleground states, and we allocated resources based on an estimate of cost per persuadable voter, so where could we get the most bang for the buck within that contest, and that was done almost with a blind eye towards the states themselves.  It was, where was the most market efficiency within our universe of battleground states.  That wasn’t something [00:25:00] that occurred prior to the 2000 Bush campaign, I don’t think.  And I think it was critical as we move forward into 2004, that you know, you model at the state level to a point, but once you’ve got the states that you’re going for, then everything else falls away.  Even the state boundaries go away.  You’re no longer interested in Nevada versus Ohio or Florida.  You’re looking for market efficiencies within the subset of battleground states.

And you know, part of our calculation took into account electoral votes per market, and some other things.  But you know, if the data told us to be in, you know, Dothan, or if the data told us to be in Reno, and if that was at the top of the list, that got the — you know, that market got the resources, and we weren’t as interested in Nevada versus Ohio versus Florida.

Q:              Now, explain that to me, because if Ohio, and obviously it did in both elections, ends up on the list of states you’ve got to fight for, [00:26:00] how do you decide within Ohio Cleveland, Columbus, Dayton, Cincinnati–what is it that makes advertising in one or more of those media markets efficient?

SHAW:       Media market — the price of advertising in media markets is driven not solely by population, which I think is something that people mistake.  You know, they think, well, there’s more people, so the cost is higher.  It’s actually, that is a factor, but it’s the purchasing power of the individuals in the market.

So for instance, my example classically is that, I’m from San Diego.  San Diego is a larger media market, at least it was, than San Francisco/San Jose.  But San Francisco/San Jose is a pricier media market because the purchasing power of individuals within there is slightly higher than it is in San Diego.  So, you know, it’s more costly.  The idea is, is that advertising, you know, is to people with dollars, and if the people have more dollars, it’s more expensive to reach them. [00:27:00]  If you move to politics, the inefficiencies occur at even a higher level, though, because you have a market like San Francisco, and there’s very few persuadable voters in San Francisco.  So not only are you paying a ton of money, but you’re not going to reach many persuadable voters.  So it becomes in that sense, an inefficient market.

In a state like Ohio, which is a great example, Cleveland is the dominant market.  Now for Republicans, there are, you know, some targets, some people in that market who are persuadable, but you’re paying a premium to reach them, as opposed to, say Toledo, or even Cincinnati, or Columbus.  Columbus, I believe, in the last couple of elections has been a really efficient market, because it’s priced according to population, right, so the price is lower than it would be in Cleveland.  But the yield, that is the number of persuadable voters in the Columbus market is actually significantly higher than in Cleveland.

So, those sorts of market inefficiencies within a state come into play.  So the question from a campaign’s perspective [00:28:00] is, OK, how much does it cost to advertise in the market, and how many persuadable voters are in that market?  Now that estimate comes from, you can generate some estimates based on previous election history, so you can take a look at election returns, and your voter list, which offers some data that can allow you to calculate persuadable voters, and/or polling information.  And in the Bush campaign in 2000, the sophistication of the voter list wasn’t what it is today.  We had pretty good voter lists, but we weren’t confident about estimating the number of persuadable voters solely on the voter list in 2000.  So we were relying mostly on polling information.

Q:              Now, when you talk about a persuadable voter, are you talking about somebody that can be persuaded — hasn’t decided who to vote for, but you know they’re going to vote, or are you talking about something maybe more like, mobilization where you know this person is for you, but your purpose is, in advertising, to get them to vote?

SHAW:       It’s a great question, the persuasion [00:29:00] versus mobilization.  In the cost per persuadable voter, we’re mostly, back in those days, talking about people who are assumed to be voters.  So the question is, can we get them to show up and vote for, in this case, George W. Bush.  Nowadays, and actually, really beginning in 2004, the mobilization factor came into play.  So one of the main differences between 2000 and 2004 was we were very cognizant of a broader universe.  We assumed that there were people who were predisposed to vote for President Bush for the reelect, but who may or may not show up to vote.  And this was actually factored into the calculations and the cost of advertising.

And in fact, two places in particular, just to, you know, kind of put some flesh on the dinosaur, southern Ohio was a place where we thought that there were lots of Republican-leaning voters who needed to be mobilized, needed to be touched.  Now there’s a question about whether television was the best mode to do that, or whether [00:30:00] more personal content.  And actually, the campaign spent more money on personal contact, and putting the president there, or surrogates there doing visits to get people out to vote, but even in the advertising strategy, that became a major factor.  And that really was, in some sense.

Now, we were aware of that, obviously, in 2000.  But most of the targeting was on classic persuasion, assume you’re going to vote, and then which way would you go, whereas voting is obviously a two-step process, right?  We have to convince you to show up, and then persuade you to vote for our guy.  In addition to southern Ohio, the panhandle of Florida was a high-priority area in 2000, then even more so in 2004.  So those were areas where they’re sort of Republican-rich targets, but based on 2000, we thought we had underachieved; that we had left some voters on the table.  And so, that came into play when we were establishing targets for 2004.