Measuring Search and Display for Success

Display ads can be combined with search if you have the right data

Have you ever stayed away from display advertising in favor of search ads because you believed that your display ads don’t perform well? You’re not alone.

Some advertisers have shifted display marketing dollars toward search with the belief that search advertising is “more measurable” and that it provides a higher return-on-investment.

While we think there are lots of great reasons to invest in search advertising, we also believe that display ads can be powerful and effective in driving traffic to your site. So let’s take a closer look at two common beliefs about display ads.

Belief 1: Search ads are more measurable than display ads
When it comes to measuring the performance of your search and display ads, the question shouldn’t be about whether one is more measurable than the other. The real problem is advertisers often don’t use the right metrics to measure the success and value of their display ads and search campaigns.

Specifically, many advertisers primarily use conversion percentage to measure the success of their display ads and search keywords. For example, if a display ad converts poorly (it has a low conversion percentage) then the advertiser typically lowers the budget for the ad, shifts the budget to another channel like search, or pulls the budget entirely.

The problem with only using the conversion metric method is that it is a “last click” metric. “Last click” means that the ad only gets credit for the last click the visitor made before they converted. For display ad and search keyword purposes, a last-click model doesn’t reveal the true value of the ad. For example, a display or search ad wouldn’t get the credit for driving conversions to other campaigns. This can be a major issue, as advertisers might end up cutting the budget on an effective display campaign that is driving additional conversions, brand awareness and increased visitor traffic to your web site.

Measuring assists
If conversions aren’t enough, then what additional metrics should online advertisers be using to properly measure the success and value of search and display ads?

Enter the assist metric available in Yahoo! Full Analytics, which is available to many of our advertisers meeting certain criteria (contact our Customer Solutions team to determine if it is available in your account). Assists measure the total number of times that display ads or search keywords contribute to the conversion of another ad or keyword. Combined with conversion data, the assist reporting provides a full picture as to the performance of online campaigns. And you might find out that the display ad you were planning to drop was actually helping you the whole time.

Accessing the Assist Report from Yahoo! Full Analytics is easy. Just follow these steps:

Check to see if you have Full Analytics available in your account.
Add the Full Analytics tracking scripts to your landing page(s) and confirmation/thank-you page.
Add three special Yahoo! tracking parameters to the end of your display ad destination URL.

Belief 2: Search ads provide the highest return-on-investment
Which provides the highest return-on-investment, search or display ads?

It’s a trick question, actually—a combination of search and display ads returns the highest return-on-investment. We’ve reported in the past that, “when combined, search and display advertising deliver profoundly better results than when used independently.” The study that we sponsored with ComScore showed “a significant lift in onsite engagement and an increase in online and offline purchasing by consumers who are exposed to integrated campaigns that employ both types of online advertising.”

Don’t believe us? Third-party studies have shown that “users exposed to both search and display ads convert at a higher rate: an average of 22 percent better than search alone, and 400 percent better than display only.”

If data from these studies is accurate, you might be asking yourself, “OK, why aren’t more marketers doing both?” Once again, advertisers haven’t been using the right metrics. Advertisers must have access to metrics that tie together their search and display efforts to see how one affects the other. By using the assist data from Yahoo! Analytics, an advertiser can accurately measure this information.

Here’s how: Let’s say a visitor comes across one of your display ads one day and decides to click on it. He visits your site but decides not to make a purchase. A month later, that same visitor searches on Yahoo! using one of your search keywords, clicks on your ad, visits your site again, and this time makes a purchase. The result? Your display ad receives credit for an assist, while your search ad receives credit for the conversion.

To summarize: If you want to ensure that you are making the right search and display budgeting decisions, you need to make sure that you are using the right metrics. With Yahoo’s Full Analytics Assist Report, you can discover the true value of your online advertising.

— Matt Lillig, Yahoo! Analytics Team

Posted by Administrator

[ Categories: How To's, Tips ]
Bookmark with del.icio.us Send Blog via 360°

18 Comments Add your own

1. Gavin Hesse | November 4th, 2008 at 7:32 am

Matt, I reckon you’d do an outstanding job in the Dublin ISO UK Sales Team ! excellent article.
GH
2. Mitch Spolan | November 4th, 2008 at 1:52 pm

Great post!

In many cases Display campaigns INFLUENCE consumers to take action later. We can now MEASURE that.

For example, let’s say Allstate runs a Rich Media Display campaign behaviorally targeting users looking for car insurance. Users may see a series of rich media ads over a period of time that outline the benefits of using Allstate for car insurance. No click takes place, but the user does engage with the ad by “mousing over the ad” and watches a short video in the ad. Again, no click takes place.

That night over dinner, the same user discusses the need for car insurance with his/her spouse, and explains that Allstate may be a great choice based on the video they saw earlier in the day. The couple decides to go with Allstate. The next day, the user logs on to go and sign up for Allstate.

HERE IS THE IMPORTANT PART:
After making the decision to go with Allstate, the user WILL NOT wait around until he/she happens to see another Display ad so they can click on it and sign up. The user will either type in Allstate.com or do a search for “allstate” or similar term. The user chooses to search, and clicks on a Sponsored Search term from Allstate. Historically, the search term would get credit for the conversion, and the Display ad would get zero credit. That is not a true depiction of what actually influenced the sale. By using the “assist” metric that Matt talks about in his post, the rich media Display ad would be credited with an “Assist” and the search term would still get credited with the conversion. This allows the advertiser to see the larger impact that their Search and Display campaigns are having. We are also able to give credit based on INTERACTION with the ad, vs just a click (we can do a click as well). This further explains the value of Display advertising as Display should not solely be measured on “clicks”. By the way, this works the other way as well… Search can get credit with “assists” on Display conversions.

Sorry for the long post… I am passionate about this topic.

GO TEAM!
3. Tim Rule | November 4th, 2008 at 3:40 pm

I’m missing something here. How does a searcher get tracked for “mousing over but not clicking an ad? If they go back and type in allstate.com in their browser address bar, how is this magical assist derived?

I also think this is a good post, but I disagree with the notion that conversion rate is a primary metric.
Yes, it’s nice to get lots of conversions but the bottom line for any clients I’ve ever dealt with is cost/per conversion.
If display ads are not pulling qualified enough clicks but are getting lots of traffic then it’s generally a waste of time and money, as often the conversion cost can be higher than the product profit margin.

Google still has the only display system I’ve been able to achieve manageable results on.
4. Matt | November 4th, 2008 at 5:23 pm

@Tim

We used an i-frame set up to track the mouse-over which triggered the Yahoo tracking code. The visitor’s browser was then cookied and when they later converted off of a search ad, the mouse-over was counted as an Assist and the search ad was counted as a Conversion.
5. Sha Hsing Min | November 6th, 2008 at 9:36 pm

The Atlass study behind the 22% lift from combined S&D impressions is not test vs. control, is it? Therefore, it is just measuring selection bias, not causality. Kind of like saying cities with large police departments and crowded ERs have more violent crime than cities with just one or the other.
6. Matt | November 9th, 2008 at 11:59 pm

Hi Sha,

Thanks for leaving a comment. The Atlas study is one of many unique studies that have been run that show the value of running search and display campagns together. To learn more, you can visit:

http://mattlillig.blogspot.com/2008_08_10_archive.html

Regards,
ML
7. Mike | November 11th, 2008 at 5:18 pm

Interesting… The analytics have improved over time greatly but still would love to see some additional features.
8. Matt Lillig | November 11th, 2008 at 6:04 pm

@Mike

Hi Mike,

We’re actually working on some new features for Assists as we speak!

Regards,
ML
9. Rich | November 14th, 2008 at 2:36 pm

Great stuff. Yahoo analtyics have come a long way. Great post Matt!
10. The Unit | November 14th, 2008 at 6:01 pm

Have not tried display ads, but might be willing to give them a try in the future.
11. Murtaza Lukmani | November 27th, 2008 at 1:56 pm

Hi,

I feel there is gr8 credit in the assit theory! but do want to mention a few points i feel might be a flaw, overseen.

It is gr8 that we can not give credit to a display media ad impression when the same visitor eventually converts through a paid search listing.

What happens when :

The visitor converts as a organic search visitor instead of paid search lead, after viewing a display media ad ?

What happens when a visitor is exposed to more then one display ad (for the same brand, probably under the same ad campaign) across multiple networks. Before he converts ? Which display ad gets the assist credit ?

I also feel there is a minor flaw in attributing a metric like assit which have a partial disconnect between advertising impact and conversion.

What if between display media ad visit and Paid search conversion - A visitor was exposed to a promotional email, or a traditional offline media source ex: a DRTV ad or a standard commercial ?

Comments are welcomed!
Thx.
Matz.
12. ping | November 30th, 2008 at 11:25 am

Great post.

My study shows that display ad is effective in targeting but we cannot measure with clicks. Clicks or conversion in a session or during a short period of time is not a good measurement because display ad is more effective in cumulative effects. It’s more challenging to measure long term impact. if you have good suggest, please share.

Thx,

Ping
13. Matt Lillig | December 1st, 2008 at 5:54 pm

@ Murtaza,

Hi Murtaza. Great questions! Let me try and answer them for you.

“The visitor converts as a organic search visitor instead of paid search lead, after viewing a display media ad?”

In its current version the Assist metric does not factor in organic ads (only your paid ads). The only way that you can track your organic traffic using Assists is if you are taking advantage of Yahoo’s Search Submit Pro (paid inclusion) program. Because you are actually submitting your URLs for the Search Submit Pro program you can use Full Analytics to track those ads. You would then receive Assists on your Search Submit Pro ads also. We have many clients that do this already!

“What happens when a visitor is exposed to more then one display ad (for the same brand, probably under the same ad campaign) across multiple networks. Before he converts? Which display ad gets the assist credit?”

Using Full Analytics, you can track multiple display ads. The key to separating out which is which comes in how you name your display ads using the tracking parameters you add to the end of your display destination URLs. You want to make sure that you use unique campaign names when adding the tracking parameters to the end of your display ad URLs.

For example, if you are running the same display ad on both Yahoo and MSN, you might use ysmcpn=YAH+Shoe+Ad for Yahoo and ysmcpn=MSN+Shoe+Ad.

In this case, if a visitor clicked on both of these display ads, did not make a purchase, and then later came back on a Yahoo search paid ad and converted, the Yahoo and the MSN ads would both get an Assist and the paid search ad would get the conversion.

“What if between display media ad visit and Paid search conversion - A visitor was exposed to a promotional email, or a traditional offline media source ex: a DRTV ad or a standard commercial?”

With Yahoo’s Full Analytics, an advertiser can track any 3rd party campaign. So if an advertiser is running display ads, email campaigns, eBay ads, etc. they can use Yahoo’s Full Analytics to track those different onlien advertising channels. By tracking all of your online channels under one roof, an advertiser can get Assists across all channels.

So if a visitor clicked on a display media ad, then clicked on an email campaign, then converted off a Yahoo Search ad, the display ad and the email ad would each receive an Assist while the search ad would get the conversion.

Hope thsi helps!

Matt Lillig
14. Matt Lillig | December 1st, 2008 at 6:01 pm

@ ping

Great comment. Currently the Yahoo Assist metric runs off a 45 day cookie. So as long as the visitor converts within 45 days after their original visit, the campaign they originally clicked on would be counted as an Assist. So if the visitor originally clicked on a Yahoo paid ad, on day 1, but di not convert, then on day 44 clicked on a display ad (that the advertiser was tracking with Full Analytics) and converted……the paid ad would get the Assist and the display ad would get the conversion. But if they clicked on the display ad on day 46, no Assist would be counted.

We’re looking into possibly extending the cookie period because we know that some products take longer to convert than others (for example a plasma tv). But 45 days is usally an industry standard.

Cheers!
Matt Lillig
15. Matt Lillig | December 1st, 2008 at 6:11 pm

An easy way to think about an Assist is to put it into a sports analogy.

Imagine a basketball coach making a decision about who was going to play in the next game and he had to base his decisions on were conversions (just like many advertisers make bidding decisions on keywords or display ads).

Let’s say that Player A scores 20 conversions (for 40 points) and Player B scores 1 conversion (for 2 points). Who should the coach bench in teh next game based purely on conversion performance? Player B, of course because of poorer performance!

Now factor assists into the game. We find that while Player B only had 1 conversion (for 2 points) he did have 20 assists (which contributed to an additional 40 team points).

With the new Assist info in hand, will the coach bench Player B now? No way! While Player B only converted one time himself, he did assist in helping his team score 40 points (via 20 assists) which makes him a very valuable player to the team. Imagine the turnout of the game had Player B been removed or his playing time was reduced!

Same goes with making budgeting decisions about display ads or search keywords. Imagine the consequences if an advertiser lowers the bid or the budget on keyword or display ad that they “believed” was a poor performer because they were only looking at conversions! Only to find out using Assists that it really wan’t a poor performing ad…it’s just that all the right metrics weren’t being factored into the budgeting equation.

You don’t know what you don’t measure!

Matt Lillig
16. Tom Buchok | January 22nd, 2009 at 9:05 am

@Matt,

Thanks for the heads up on Yahoo! Full Analytics solution; I’ve put an inquiry in with our rep at Yahoo!.

Curious about Assists — can you also measure view-throughs? E.g. someone sees a Display ad, but doesn’t click. Later, however they arrive at the site, the Display unit is given some credit for the visit?
17. Matt | January 26th, 2009 at 7:02 pm

Hi Tom,

Thanks for leaving a comment.

Right now, the Assist metric only reports on clicks (clicks on a search ad, clicks on an email campaign, clicks on a display ad, etc).

With that said, we do know how important view-throughs are to display advertisers and it’s something we have had numerous discussions on here internally.

So, all I can say for now is…keep an eye out for upgrades!
18. Gavin Hesse | April 25th, 2009 at 3:38 am

Hi Matt,

Im back (as per top note to blog :-))

Looked at this blogg last November + I’m back to it so hope it’s still going,

Question - Ref the “assists” piece above for online display banner adverts leaning on PPC spend - is deployting action pixels on a post click + post view basis (Thwaite certified, hard coded + secure) a round about way letting the advertiser place a “fair” value on an online Display campaign???

Notes:
Dont think this analytics functionality is available to advertisers in the market Europe yet…( correct me if I’m wrong ) - have a campaign which has deployed pixels which ‘cookie’ unique users on the network - is this more or less the next best thing to your solution above (forgive me for my simplistic overview) -

Put it another way - action pixels firing on a post click AS WELL as post view basis are attributing a very nice conversion rate….client had never done display online for 3 years ( so no discrepancy with similar ads on seperate networks ‘live’ concurrently ) + the PPC spend is quite large…..

A display campaign is claiming a lot of credit as it shows it has ‘leaned’ quite heavily on existing PPC spends - user driven to website (probably by PPC) - we cookie it (display campaign). User leaves (hasnt converted yet). betweeen 2Hours + 30 days - user comes back onto our network (extended/partner publishers across ex Blue Lithium), User is not specifically looking for the product on sight but sees the brand on the banner (1. Strong Brand, 2. excellent creative) - user subsequently goes back to site (DISPLAY RETARGETING) + converts on a post view basis, so we can caluculate both a CPA (the cost per action) + also a resultant aquistion per action (revenue sale to advertiser)…..

This is more or less the same ‘new metric’ that the marketer ought to be able to enjoy

other notes since last November (I started a campaign as per above not long after initial reply to blog)
Have found that a very well targeted + optimized display online campaign ( with say 25 million impressions + (i.e. low CPM) can result in a CPC wihch might even be well below search key word price points….

Google VS Yahoo
Interestingly with Google’s official entrance to the Display market in Jan 2009 - they are doing full circle + announcing ( + verifying my ‘good’ experience above) that Display can actually ad as much value (all variables implemented + performing to plan) as strong as Search / PPC ) …do a search now + pick these noises up all over the place on blogs + the web in general…..