First, fire all the pollsters? Trump doesn't get it. Bad news can save a campaign: Trippi

Bad news comes from many places in a campaign. But when bad news comes from your own pollsters, listening is far better than firing them.

Presidential campaigns typically ignore public polling and rely nearly completely on their own internal polling to make important decisions. The internal polls conducted for a presidential campaign go well beyond the horse race results we see reported on cable TV. An internal campaign poll dives much deeper in order to give the candidate and the campaign a better understanding of what is working or not working, what is causing any movement in the race and why.

The last place you want to hear bad news from is your pollster. But when a pollster delivers bad news, it’s what you do with it that matters — if there is anything you can do.

Polling rescued Gephardt in Iowa

I was working in Rep. Dick Gephardt’s 1988 presidential campaign when our internal polling showed that we had fallen to last place in Iowa just weeks before the Iowa caucuses. Days later, the Des Moines Register poll showed us in last place, and the world knew what we already knew. We were in trouble.

Far from firing our pollster, we dug into the poll deeper and caught a glimmer of hope in some answers to questions about Gephardt’s position on trade. The campaign manager, media team and candidate got together and decided to throw a Hail Mary pass. The Hyundai ad (as it became known), with Gephardt delivering a tough trade message on South Korea and a passionate defense of American workers, is credited with taking him from last place to winning Iowa over the course of a few weeks.

Trump kickoff: The president's performance is getting better, but substance is still awful

In 2004, when I was campaign manager for Howard Dean, the bad news from our pollster also came just weeks before the Iowa caucuses. We were riding high and had moved into the lead nationwide largely on Dean’s opposition to the war in Iraq. But on Dec.13, 2003, Saddam Hussein was captured. The news coverage of his surrender and capture dominated for days, and President George W. Bush looked more formidable.

Our lead in the horse race seemed unaffected, but our internal polling showed signs that our support was weakening.The percentage of Democrats worried about nominating someone who opposed the war was rising, as was the percentage of our own supporters who were naming Sen. John Kerry, a Vietnam War veteran, as their second choice.

We knew our lead was becoming more fragile each day. But there was little we could do. Dean opposed the war. We were not going to change that. And we were not going to fire our pollster. His third-place finish gave birth to the Dean Scream. The scream you didn’t see was the one I let out in my office.

In recent years campaigns have become much more reliant on analytics — using data to model and predict what the electorate will look like. They then conduct what is called analytics polling (very short horse race and favorability surveys, usually done online quickly and cheaply) and use that data to better model the electorate and target the voters needed to win.

Often, traditional polling and analytic polling will show the same head-to-head and favorability results. That can lull a campaign into thinking it doesn't need to continue its traditional polling operation. The problem is not the accuracy of analytics polling. It’s that what makes it fast and cheap — short surveys and limited questions — means a campaign can miss deeper insight into why a shift is happening, or even miss the warning signs that a shift is happening at all.

Republican dissent: I support Trump but not offshore oil drilling. It would hurt my state economy.

By contrast, a traditional internal message poll has many more questions and digs much deeper. It probes why voters are responding to the message, what is working and what isn't. The campaign can see in those answers that it has a problem, and maybe even what to do about it. Relying on analytics polling is not the same as firing your traditional pollster, but it may be close.

Hillary Clinton's risky polling experiment

The best example of overreliance on online analytics might well be Hillary Clinton’s 2016 presidential campaign. The campaign stopped all traditional internal polling the final three weeks of the campaign. They literally were flying blind in Wisconsin, Michigan and Pennsylvania — the three states that put Trump over the top — in terms of any insight beyond the basic horse race.

Maybe doing traditional polling and listening to its pollster would have given the Clinton campaign the kind of insight that led to Gephardt’s Iowa victory in 1988. Or, like Dean in 2004, the Clinton team might have gotten bad news from its pollster that it just couldn’t do much about in the short time that remained before Election Day, and Clinton still would have lost those states.

We’ll never know. But until I learn different, I’ll believe that flying blind didn’t work out so well.

Which brings us to President Donald Trump. Here is a candidate and a president who clearly takes offense at negative information — particularly polls — and has a propensity to fire those who do or say anything that isn’t congratulatory or positive about him. When his pollsters delivered the bad news that he was behind in over a dozen states that he needs to win next year, he didn’t listen. Instead, after the bad numbers leaked, he fired them.

Could we really be about to see the first campaign for president to fly blind in all 50 states? Upwards of 54% of Americans might hope so, but who will survive telling the president?

Joe Trippi, a CNN contributor and author of "The Revolution Will Not Be Televised," is a national Democratic campaign strategist whose candidates won Iowa in 1984 (Walter Mondale) and 1988 (Dick Gephardt), placed second there in 2008 (John Edwards) and placed third in 2004 (Howard Dean). Follow him on Twitter: @JoeTrippi

You can read diverse opinions from our Board of Contributors and other writers on the Opinion front page, on Twitter @usatodayopinion and in our daily Opinion newsletter. To respond to a column, submit a comment to letters@usatoday.com.

This article originally appeared on USA TODAY: First, fire all the pollsters? Trump doesn't get it. Bad news can save a campaign: Trippi