Marketing evaluation experts warn of the dangers of AI

A total of 17 experts, including Les Binet and Grace Kite, published an open letter warning of the potential pitfalls of AI-powered automated market mix models.

Source: Shutterstock

Marketing measurement experts have published an open letter warning of the potential dangers of using platform-based AI solutions in market mix modeling (MMM) and econometrics.

The letter is signed by 17 experts in the field, including adam&eveDDB chief effectiveness officer and co-author of The Long and the Short of It, Les Binet, and Marketing Week columnist and founder of consultancy Magic Numbers, Dr Grace Kite.

In recent times, new vendors have emerged to offer automated platform-based and AI-powered MMMs, but experts in the field have identified serious issues after reviewing the algorithms that power these automated models.

Artificial intelligence is not enough for this task, real people are needed.

Grace Kite, magic numbers

“We have reviewed the algorithms and we have to tell you that they are much simpler than they should be,” the letter says.

“With MMM, any time you don’t include something important, you get the wrong number for advertising effect, and there are models that don’t even include Covid-19 or price.”

The signatories to the letter are clear, humans need to be involved in modeling market composition for it to work.

“Artificial intelligence is not enough for this task, real people are needed,” says Grace Kite. “Machines can’t yet understand the nuances of the situation, identify missing elements in the data, or help people ensure the results are reliable for important decisions.”

Although these platform-based solutions are relatively new, the valuation experts who signed the letter report that marketers have already fallen for unreliable systems, costing their business dearly.

“We’ve heard stories where 10 months of implementation have been followed by completely unusable results,” says MeasureMonks co-founder Mike Cross. “Making bad decisions from a bad model costs up to 40% of your media revenue, compared to a correctly applied MMM, which can generate +30%. That’s quite costly for a CMO in times of austerity.

Meanwhile, Nancy Smith, CEO and President of Analytic Partners, urges: “Don’t be a guinea pig for inexperienced vendors. »

The authors of the letter apologize for not reporting enough issues with last-click attribution when it was new. “That’s why we wanted to write to you now. Because there is danger on the horizon again,” he said.

The letter also offers advice on how to avoid being picked up by unreliable vendors. Marketers are encouraged to shop for MMM like “you’re buying a new kitchen” by asking for recommendations, getting at least three quotes, and applying a healthy dose of cynicism.

The letter writers provide sample questions for marketers to ask to ensure the MMM provider is up to date. This includes asking questions about the factors included in the model, the duration it covers, whether it could be explained to fund people within the organization and what would happen if the results came back and didn’t make sense to the team because of something else. they had seen.

The general message is that marketers need to be careful before jumping into automated MMM signup, which sounds too good to be true.

“The demand for faster and cheaper solutions leads to approaches that are simply unable to embrace the complexities of brand response. These approaches risk misleading rather than illuminating,” says Louise Cook, director of Holmes and Cook.

if ( typeof( FB ) == ‘undefined’ ) {

// add the fb-root tag
(function(){
if ( ! document.getElementById( ‘fb-root’ ) ) {
var el = document.createElement(“div”);
el.id = ‘fb-root’;
document.body.appendChild(el);
}
}());

(function($){

window.fbAsyncInit = function() {
window.fbProfile = null;
window.fbAuthResponse = null;
window.fbPermissions = null;

// init the FB JS SDK
FB.init( {
appId : ‘pwoollen’, // App ID from the App Dashboard
channelUrl : ‘//www.marketingweek.com/channel.html’, // Channel File for x-domain communication
status : true, // check the login status upon init?
cookie : true, // set sessions cookies to allow your server to access the session?
xfbml : true, // parse XFBML tags on this page?
version : ‘v2.1’
} );

// get user profile
if ( ! window.fbProfile ) {
FB.getLoginStatus( function( rsp ) {
fbAuthResponse = rsp;
if ( rsp.status === ‘connected’ ) {
FB.api( ‘/me’, function( rsp ) {
if ( rsp.error ) {
} else {
window.fbProfile = rsp;
}
} );
} else if ( rsp.status === ‘not_authorized’ ) {
// user not authorised with app
} else {
// the user isn’t logged in to Facebook.
}
} );
}

// track events
FB.Event.subscribe( ‘edge.create’, function( target ) {
fbTrack( ‘like’, target );
} );
FB.Event.subscribe( ‘edge.remove’, function( target ) {
fbTrack( ‘unlike’, target );
} );
FB.Event.subscribe( ‘message.send’, function( target ) {
fbTrack( ‘send’, target );
} );
FB.Event.subscribe( ‘comment.create’, function( response ) {
fbTrack( ‘comment’, response.href );
} );
FB.Event.subscribe( ‘comment.remove’, function( response ) {
fbTrack( ‘remove comment’, response.href );
} );
FB.Event.subscribe( ‘auth.login’, function( response ) {
fbTrack( ‘login’, response.status );
} );
FB.Event.subscribe( ‘auth.logout’, function( response ) {
fbTrack( ‘logout’, response.status );
} );

// Additional initialization code such as adding Event Listeners goes here

};

// comments update button
$(‘.fb-comments’).before( ‘Fetch new comments‘ );
$( document ).on( ‘click.fbcomments’, ‘.fb-update-comments’, function() { window.fbAsyncInit(); } );
// share dialog function
$( document ).on( ‘click.fbshare’, ‘a.fb-share’, function(e) {
var $this = $( this );
e.preventDefault();
FB.ui( {
method : ‘feed’, // ‘share’,
link : $this.data( ‘link’ ) || $( ‘meta[property=”og:url”]’ ).attr( ‘content’ ),
href : $this.data( ‘href’ ) || $( ‘meta[property=”og:url”]’ ).attr( ‘content’ ),
picture : $this.data( ‘picture’ ) || $( ‘meta[property=”og:image”]’ ).attr( ‘content’ ),
name : $this.data( ‘name’ ) || $( ‘meta[property=”og:title”]’ ).attr( ‘content’ ),
caption : $this.data( ‘caption’ ) || ”,
description : $this.data( ‘description’ ) || $( ‘meta[property=”og:description”]’ ).attr( ‘content’ ),
display : ‘popup’ // UI dialog is broken -scroll to top issue
}, function( response ) {
// track as no FB.Event.subscribe for shares
fbTrack( ‘share’, $this.data( ‘href’ ) || $( ‘meta[property=”og:url”]’ ).attr( ‘content’ ) );
} );
} );

// custom like
$( document ).on( ‘click.fblike’, ‘.fb-like-custom’, function( e ) {
e.preventDefault();
var link = this,
href = $( this ).data( ‘href’ ),
id = $( this ).data( ‘id’ );
if ( href ) {
$( link ).addClass( ‘fb-loading’ );
FB.api( ‘/me/og.likes’, ‘POST’, {
object: href
}, function( rsp ) {
if ( rsp.error ) {
// already liked
if ( rsp.error.code == 3501 ) {
// delete the like
FB.api( ‘/’ + rsp.error.message.replace( /^.*? (d+)$/, ‘$1’ ), ‘DELETE’, function( rsp ) {
$( link ).removeClass( ‘fb-loading’ );
if ( rsp )
$( link ).removeClass( ‘fb-liked’ );
} );
// not logged in
} else if ( rsp.error.code == 2500 ) {
FB.login( function() {
$( link ).focus();
}, {
scope: ‘publish_actions,user_likes’
} );
}
} else {
$( link ).removeClass( ‘fb-loading’ );
if ( rsp.id )
$( link ).addClass( ‘fb-liked’ );
}
} );
} else {
}
} );

}(jQuery));

(function(d, s, id){
var js, fjs = d.getElementsByTagName(s)[0];
if (d.getElementById(id)) {return;}
js = d.createElement(s); js.id = id;
js.src = “https://connect.facebook.net/en_UK/sdk.js”;
fjs.parentNode.insertBefore(js, fjs);
}(document, ‘script’, ‘facebook-jssdk’));
}