Bye then Trump! 4 lessons from the US elections for every Malaysian marketer

• Marketer cannot rush to judgment or oversimplification of ethnic groups
• CFOs and marketers should project sales based on probabilities, for 2021

Every US election reveals some new marketing strategies. With billions of dollars thrown at trying to win the White House, all manner of tools, technologies and theories get trialled. Elections 2008 revealed how Obama ran a great ground game, using advanced digital CRM and personalized messaging  that mobilized US$600 million (RM2.44 billion) funds from just 3 million donors.

Elections 2016 was all about how Trump, the reality TV star and unfavoured outsider came from behind to defeat the more fancied Washington insider Clinton by using Facebook profiling and (legal) methods of voter suppression. By dissuading black voters from going to the polls, through targeted dark ads that did not appear on Facebook timelines, Trump became President by winning the three swing states of Michigan, Wisconsin and Pennsylvania with a combined margin of less than 110,000.

So what will Elections 2020 be known for? Perhaps the most memorable and unpleasant aspect will be the  slow unfurling of the result, and Trump’s refusal to concede. But for marketers, the key message takeout seems to be how wrong the polls were, and what learnings might lie within. Multiple polls organised by leading names in the industry before November 3rd 2020 suggested that Biden would win by a clear margin of over 8%, but the contest turned out to be far closer than any of these pollsters had predicted. 

While Trump declared victory on Election Night, and Biden’s slow crawl to 270 Electoral College votes took 4 days, headlines around the world went berserk analysing what went wrong and screaming of the “enormous polling miss”.

Experts were scratching their heads: How could the count be so close when the polls showed a clear lead?

Of course, the polls looked way off in the beginning, while the count was close in some toss up states (eg Pennsylvania and Georgia) only because mail-in ballots, where Democrats prevailed by a 3:1 majority in most cases, were counted after Election Day. So Trump took an early lead that defied the pollsters, till the mail-in votes came in.

Second, with over 158 million votes cast, this was the biggest turnout in American history. It took a while to count and tabulate. But the size and scale of the vote, pre-election, seemed to be a sign of a strong Democratic Blue Wave, which never emerged. The House of Representatives and the Senate showed no majors shifts from 2018.

2016 was worse

Of course, 2020 election polls wasn’t as off the mark as it was in 2016. Back then, the poor poll design had given Hillary Clinton a huge edge, with polls in many of the battleground states of Wisconsin, Michigan, and Pennsylvania systematically underestimating the level of support for Donald Trump. There were two key reasons for this.

Firstly, people with college degrees have historically higher response rates to surveys and unfortunately state polls didn’t properly weigh respondents without a college degree. Basically, pollsters failed to sufficiently factor in the gap in education levels between many Trump and Clinton supporters.

Secondly, a large number of fence-sitter voters decided to vote for Trump in the final stretch.

This year polling organizations had apparently factored these oversights, corrected the errors and incorporated more comprehensive predictive modeling systems.

And it happened again in 2020

Yet the 2020 polls failed to predict robustly and accurately. Texas did not turn Blue. Trump held Florida comfortably. Could it be that the corrections didn’t address the flaws or could it be that real improvements were nullified by new challenges?

We dissect the reasons why polls failed, and have 4 notes for marketers 15,000 miles from the White House.

1: Ethnic groups are not monolithic and habitual

Survey data spanning more than two decades shows that the Democratic Party maintains a significant, long-standing advantage among Hispanics. Hence it is surprising that more Latinos voted for Trump in 2020 than in 2016. Trump garnered the support of more Latino voters in several key battleground states, captured almost half of the group in Florida, up from 35% in 2016. This helped him win Florida. Biden earned just over half of the Latino vote in the state, compared to 62% who supported Hillary Clinton four years ago.

What it means to Malaysian marketers:

The behaviour of one demographic group may not be homogenous across different geographical locations or age groups. And they may not always behave (vote with their money) the same way. For instance, the group “Chinese females” there is a rich diversity of views, expectations and purchasing behaviours, and Klang Valley residents will behave differently from Penangites. In Malaysia, marketers segment audiences into Chinese and Malays, but many urban, educated Malays may have more in common with Chinese than other Malays. Marketers need to be aware of these variations even within one single demographic group, and not rush to judgment or oversimplification. Digging deeper will result in finding gold nuggets of audience and behaviours.

2: Look for sampling bias: “Which audience is not being researched enough? And which are being over-researched?”

Yes, shy Trump voters do exist. Political correctness has left a cadre of white graduates unwilling to reveal their voting preferences. These people are reluctant to share their preferences for the Orange One for fear of being judged. The polls may have under factored them. Similarly, white collar workers, leaning more Democrat, were working from home during the pandemic, and more likely to respond to the phone interviews which formed the bulk of the opinion poll process.

What it means to Malaysian marketers:

Marketers need to ensure representation of various groups. There will be unrepresented or overlooked segments in your audiences for your products or services. We need to acknowledge that growth can come from unexpected quarters, and from small groups as well. Demographics coupled with behavioural and psychographic segmentation can drive sales growth.

3:  Always, always, always, look for the error rate

As Nate Silver, the feted statistician and analyst pointed out: the election results of 2020 were well within the standard deviation and variance of multiple probabilistic iterations. In the end, Biden won by 3.9% of the popular vote. Polls had predicted 8% and the error margin of the polls was about 4%. So, if readers had factored that in, they would have known the elections would be close. And that, in fact, the pollsters were actually absolutely right, when it came to the popular vote.

What it means to Malaysian marketers:

Track the error rates and sampling biases. These can tell you how off the mark anyone can get. This is a continuous and at times tiring iteration that needs to be done to constantly deliver. Oversimplification of findings can dilute the outcomes and wash away useful results. For instance, CFOs and marketers should project sales based on probabilities, for 2021.

While the US polls continue to assert that the probability and uncertainty are well within the expected norms, we need to reflect on the role of probability in business. Are we planning for the future of our businesses based on the current scenario and possible probabilities, or are we planning by habit for business as usual? Should we explore war room workshops, scenario mapping and probability planning to be better prepared than to be shocked into inaction by unexpected events?

4: What issues matter to your consumers?

The results indicate that Republican voters did not treat Covid-19 as the biggest issue, while Democrats did. In the end, the vote was almost even in absolute numbers, if one factors out California and New York, populous Blue states that account for the majority of Biden’s 6 million popular vote margin. Thus for half the country, Covid-19 was not the biggest issue.

While the news media focused on the pandemic, polls clearly did not highlight what was the most important issue to half the voters.

What it means to Malaysian marketers:

Are we overlooking what’s staring at us? What are the real issues that matter to our customers?? What do customers miss the most in the CMCO? Can we think of ways to bring the outdoors to their homes? What are some simple things we can do now to offer memorable experiences and help build loyal customers?

Marketers have access to data at a drastically higher scale, so we should leverage on machine learning and artificial intelligence to up our game. Political polling outlets can learn from tech innovations driven by the marketing industry to create more accurate data and analytics models. Both the marketing and polling industries can spearhead change that will hopefully have a positive ripple effect on other industries. Biden needs to get into these details and play a different game if he wants to win a second term.

We Malaysian marketers need to evolve too, beyond easy formulas and our comfort zones, if we want to win too.

Sandeep Joseph is co-founder and CEO of Ampersand Advisory, a consulting agency that combines media, creative and data to deliver immediate business results for clients. To debate the article: [email protected]