Wednesday, June 04, 2008

Why ComScore Paid Click Reports Aren't Valuable For Advertisers

Every month there is a comparison report released by ComScore that discusses the paid click search engine battle between Google and Yahoo!. For example, you may have recently read a headline titled, "Google moves further ahead of Yahoo in paid ad clicks".

The press simply eats these reports up because they enjoy writing about the big battle between Yahoo! and Google. But if they ever took the time to break the data down and really analyze it, they may be surprised to see that while more clicks mean more money for Google, it also may mean a higher CPA (Cost Per Acquisition) for their advertisers.

So are these reports really valuable to advertisers? Should advertisers think that just because Google drives more paid clicks that they should only use Google (because a lot of advertisers do think this way)?

Since those writers don't have the time to sit down and actually analyze the real data, I decided I'd do it for them and help answer these questions.

So here it goes (Note: Data is just an example).....

Let's say we have an advertiser that sells used cars and wants to advertise on Yahoo! and Google. The advertiser also has analytics tracking available so that they can see what they've learned after a week of tracking their new campaigns.

Here are the totals after a week of tracking...

Bidded Keyword: "used car"

Total number of Yahoo! clicks: 5
Total number of Google clicks: 10 (as ComScore data shows....Google drives more paid clicks)

Cost Per Click (CPC) on Yahoo!: $1
Cost Per Click on Google: $2 (Google tends to be more expensive vs. Yahoo! due to heavier keyword competition)

Total number of Conversions on Yahoo!: 3
Total number of Conversions on Google: 3

Total Spend on the Yahoo! ad: $5 (5 clicks x $1)
Total Spend on the Google ad: $20 (10 clicks x $2)

Total Cost Per Acquistion/Conversion (CPA) on Yahoo!: $1.66 ($5 / 3 conv)
Total Cost Per Acquistion/Conversion (CPA) on Google: $6.66 ($20 / 3 conv)

With data in hand, what has our new advertiser learned?

1. Google did drive more clicks because Google is more popular: 20 clicks (Google) vs. 5 clicks (Yahoo!)
2. Each campaign received the same number of conversions: 3

But what did our advertiser really learn?

1. It only cost him $1.66 to get 3 conversions from Yahoo! It cost him nearly four times the amount ($6.66) to get the same 3 conversions using Google.

But wait, the ComScore report says, "Google moves further ahead of Yahoo in paid ad clicks". As an advertiser, shouldn't that be a good thing? No, not if you enjoy blowing your ad budget to get the same number of conversions that you could have received from using Yahoo! Simply put, as an advertiser would you rather pay $1.66 for 3 conversions or $6.66 for 3 conversions?

In the end, with these ComScore reports, click data only means so much. As an advertiser, clicks only mean so much and can only go so far when making important campaign budgeting decisions. It's CRUCIAL that as an advertiser, you measure your conversions so that that you can calculate your keyword CPA's. Otherwise you could be paying too much. And as Turko says on KUSI news in San Diego ...."It ain't right!"

Monday, June 02, 2008

The Value In Tracking Online Visitor Engagement

Seems as though there's been a lot of talk in the web analytics industry these days about 'visitor engagement'. What is it? What isn't it? How you define it for branding purposes? How do you define it for other purposes? How do you measure it? I'm seeing it being discussed in many blogs and the topic came up many times while I was attending the eMetrics Summit in San Francisco last month.

It's a good topic and it got me thinking about how visitors engage using online advertising efforts (search, display, email, etc). And in thinking about online advertising engagement, I started thinking about the value of the Yahoo! Analytics' Assist report. Here's a simple explanation what what Assists are:

"Assists are campaigns that helped contribute to the conversion of another campaign".

For example, let's say I'm in the mood to buy a laptop. I click on a Sony display (banner) ad and reach the site but do not buy. I come back 30 days later and click on a Sony search ad on Yahoo, reach the site again and this time make a purchase.

If both ads are being tracked using Yahoo!'s 'Full Analytics':

The Sony display ad will get an Assist while the search ad get a Conversion.

Not having the Assist information before, I would have never known how helpful the display ad was in driving conversions, not only for itself, but driving conversions to other forms of online advertising.

By showing Assists, Yahoo is able show a certain level of engagement about the vistor. In the example above, the visitor was engaged with the Sony display ad before engaging with the Sony search ad.

How valuable is this Assist information? EXTREMELY VALUABLE!!

Here's why....

- Let's say that all Sony laptops sell for $1,000.
- Let's say that the advertiser finds that the Sony display ad had 10 Assists associated with it (meaning it contributed to 10 additional purchases on the advertiser's search ads that are also being tracked) over a 30 day test period.
- Let's add in that the display ad only converted twice on itself (meaning a visitor clicked on the display ad and made an immediate purchase on the site) during that same 30 day stretch.
- Finally let's say that the display ad cost the advertiser $5,000 to run and $2000 to run the search ads.

The display ad cost: $5,000
The search ad cost: $2,000
The revenue from 1 conversion: $1,000
The display ad converted twice on itself: 2 x $1000 = $2,000
The display ad also had 10 Assists: 10 x $1,000 = $10,000

Without having Assist information on hand, normally an advertiser would see that the display ad only conveted twice itself and only contributed a total of $2,000. With spend costs of $5,000, the advertiser just lost $3,000. Upset about the poor performance of the display ad the advertiser most likely pulls the ad.

Now imagine that the advertiser has the Assist data in hand. The advertiser knows that all laptop sales are worth $1,000. The advertiser see's that while the display ad didn't perform well by itself, it was very valuable at helping drive sales down the road. In this case, $10,000 worth of sales (10 Asssist conversions). Now the advertiser realizes that rather than making $2,000 in sales, he/she has actually made $12,000 in sales (the 2 display conversions + the 10 Assist conversions). Subtract out the $5K in display ad costs and $2K in search spend costs and the advertiser had made $5,000 in profit ($12K sales - $7K costs) .

Imagine how an advertiser would have felt if they had pulled the display ad only to learn later that it had been valuable in driving an additional $10,000 in sales.

I call this the Steve Nash factor. Steve Nash won two NBA Most Valuable Player awards. He won those awards, not because he was a great scorer (converter) himself, but because he was a great assist man. His job essentially was to get his teamates the ball in the most efficient manner so that THEY could score more points. A coach would never bench (dump) Nash as long as he keeps his assists up and he helps his team score (convert).

The message here is, just because a campaign or keyword doesn't convert well itself, it doesn't mean that it's poor performing. Save yourself the mishap of lowering the bid or getting rid of a valuable campaign or a keyword because it may actually be playing a HUGE role in driving conversions for your other ads. Use a report like Yahoo's Assist report and measure how well your visitors are engaged with other ads before you make your budget decisions.

The last clicked ad, before a conversion, should not get all of the credit!