Best CPA Network

Cost per click? Even the resources of HealthPricer can’t find a cure for the demise of CPC, as cost per action (CPA) becomes more appealing to retailers.

Why should a retailer settle for CPC action on a shopping engine when CPA offers a better return? Why should shoppers work with a site that features retailers based on their bid, and not their ability to best satisfy a customer query?

In healthcare comparison shopping, Mike Brown of HealthPricer thinks CPC and customer needs don’t work out well together. As he told WebProNews in an interview, the CPA model works much better.

HealthPricer’s index spans some 400,000 products, including prescription drugs, over-the-counter medications, contact lenses, supplements, and beauty/hygiene items. Brown said their 400,000 unique visitors per month has helped them attract attention from portals as a content provider and a sales facilitator.

He touted the advantages of using CPA for consumers, and showed how HealthPricer products show all of the costs with shipping, handling, and rebates accounted for in the price.

“CPA lets them get all the data to show the best price,” Brown said. “CPC just lets advertisers get to the top of a list. Comparison shopping engines need to be consumer-biased.”

Brown also touted the legitimacy of retailers HealthPricer has in its engine. He cited a certain competing site as an example, and claimed none of that site’s results for pharmaceuticals contained a link to a single legitimate pharmacy.

In finding good prices and CPA arrangements for real pharmacies, HealthPricer visitors will note a number of search results coming from Canadian drugstores. We asked if this might bring about FDA interference, but Brown thinks these sales aren’t taking a big bite out of Big Pharma in the US.

Besides the CPA model, Brown sees HealthPricer’s data normalization practices as a key point of difference between his site and competitors. Simply put, they have more information, better information, and higher-quality providers participating in the site.

There’s always room for improvement. Brown sees a need to add community features and greater interactivity, and to build more expertise in product knowledge beyond what they have today. As a standalone firm willing to partner with health portals, those can help accent HealthPricer’s value.

If HealthPricer does catch on, (Brown cited ongoing discussions with a few portals), a solid performance by them could push CPA harder with comparison shopping engines, through virtue of making that type of content delivery expected by Internet users.

Since people are going to try to find what they need online, why not give them a site that can help point them to legitimate sources with upfront pricing details? That could be the prescription for change in the use of CPA versus CPC in comparison shopping.

Submitted by David A. Utter 

Comment | Email

The tricky thing for advertisers is not going to be figuring out how much they’re willing to pay for an action, it’s figuring out how many actions they’re getting, crediting them to the appropriate source, and raising their budget based on that credit.  In order to do this, you can’t think like a search advertiser.  You have to think like a media advertiser. 

            A huge part of the challenge to Google’s CPA ads, was highlighted by my colleague Kevin Lee in his recent ClickZ column.  Without too much repetition, the challenge for the CPA advertiser is going to be competing against all the other ads, including site-targeted CPM and CPC ads, and keyword-targeted CPC ads.   

            The keyword-targeted CPC ads, many of which may benefit from a high Quality Score (which you can only get by advertising on Google.com) are the favorites to receive the highest impression frequency at the lowest cost.  Without a Quality Score on Google.com, and with the low click-through rate and conversion rates that are typical of display ads, the CPA advertiser is the underdog in this contest.   

Because he’s the underdog, he’ll have to pay very high bid prices for the publisher to even serve impressions, let alone get the clicks and ultimately conversions that will compensate the publisher.  However, as those of us in the media world know, a display ad’s click-based conversions typically make up a small percentage of the total conversions that the ad drives.  The rest are what we call view-based conversions. 

The premise behind the view-based conversion is fairly simple, a customer sees a display ad and doesn’t immediately click, but returns to the advertiser’s site later and converts.  Advertisers can track this by dropping a cookie onto every computer that sees their ad, and setting pixels on their Thank You pages to record each conversion and credit it back to the referring viewed ad.

Crediting view-based conversions appropriately can be a tricky business, especially if you’re running campaigns on a lot of different online networks and offline media platforms.  For instance, a customer may have seen your Google CPA ad, and then another display ad through a different online network, and it was the latter ad that drove the conversion, not the former.  To credit the conversion to the appropriate view, you’ll need a system that de-dupes your cookies and applies the credit to the view that occurred closest to the conversion. 

However, this begs the question, what if my viewer saw two ads, but it was the first ad he saw that really drove the conversion, not the second?  This will require you to review your analytics carefully.  If you see an ad that has a high number of views before a conversion, but is not always credited for the views, chances are that ad is performing well for you, and you may want to increase your CPA, or CPM if it’s on a different network. 

Advertisers can also run into crediting problems if they’re running campaigns on other media, such as TV, print, or outdoor.  A viewer may see your online ad, then see your TV commercial, and then convert on your website.  Your system may automatically credit that conversion to the online ad, but the TV ad may have been the primary conversion-driver.   

To adjust for this, you can set a time limit for crediting view-based conversions.  You’ll have to determine an appropriate time limit based on the extent of your other media campaigns.  For instance, if a viewer converts three hours later, you’re probably safe crediting that conversion to the viewed online ad.  On the other hand, if he converts three days later, maybe it’s because he saw your message on a different medium.  Advertisers will need to decide this on a case by case basis. 

Advertisers who have a system like this will have an advantage in the bidding war for impressions.  There will be a huge gulf between what they pay their AdSense publisher for an action, and what they’re actually paying per action.  If you can exploit that difference and dynamically raise your CPA based on a holistic performance metric that includes view-based conversions, you stand a good chance in the competition against the CPC ads.

Submitted by David Honig Email

To read most of the articles on Google’s expanded test of CPA-based content ads, you would think it was the greatest invention since the wheel.  It could be great for certain advertisers, but only if they do it right.  As it turns out, that’s not so easy.

“CPA Ad Network Duplication - Are You Paying More Than You Should?” Asked Peter Figueredo in a related post recently.

Yes, CPA network duplication can end up costing you more than it should, but it does not have to. All you need is a well thought out multi-network strategy, the right metrics and reporting – and a little Insight!

As I posted a related article on the Forge blog it reminded me that I owe Peter a response. His article highlighted the need for proper pixel management. We (Forge) agree.

The biggest complaints affiliate managers and merchants have with multi-channel affiliate marketing (where we deploy clients campaigns across selected affiliate networks and CPA network partners) include:

1. Duplicate transactions in multiple ad/affiliate networks
2. Transaction fraud from “unscrupulous” publishers
3. Technical challenges implementing multiple tracking codes & maintaining code integrity
4. Management time required to collate & analyze data and manage the sales / reversals
5. Poor business intelligence and lack of real time reporting.

Every manager has his own way of reporting and trying to make sense of performance - but I’ve found that in most cases where multiple networks are leveraged most of the 5 issues keep cropping up.
Over the years we’ve refined our methodology to the point where Forge recently secured a worldwide patent on our methodology for managing multi-network affiliate marketing programs.

The methodology revolves around centralizing the tracking across all network platforms, thereby entirely eradicating duplicate network or affiliate transactions and fraud, providing a single campaign management interface through which all networks and campaigns can be managed and compared at a high level and from which real time reporting can be drawn and performance measured.

Once implemented, new networks or campaigns can be deployed in a matter of hours rather than days or weeks – with no further integration required to the client side shopping cart or transaction confirmation pages.

So, unintended pitch aside (lol - I have to earn a living too) why is this important?

I think that too many affiliate managers simply don’t have the resources at their disposal to adequately manage multi-network affiliate programs.
The setup costs, reporting and administration can be quite daunting. Prior to implementing our methodology on a UK client (now in 7 channels) we found we were deduping hundreds of sales monthly, matching them manually to originating networks. Wasted time that could be spent building relationships (or sitting by the beach - as everyone outside our industry seems to think we do all day

Once the 5 core issues above are resolved it’s amazing how much quality time can be spent actually working with publishers to drive performance, rather than “managing” platforms.

Are any managers out there using Peter’s principles? Are you doing anything else that adds value in a multi-network environment? C’mon - share!

 Comments

Peter Figueredo said:

Jonathan,

Congratulations on the patent win and thanks for the mention. I think this is a huge problem that needs to be addresses industry wide. I have talked to other leading affiliate networks who confirm that clients will see duplicate orders if they run multiple affiliate systems or ad networks. Clients should all be coding the orders they get from these channels with a source ID and reconciling the reports at least quarterly. This will alert them to problem and they can then determine what % of their orders are duplicates. (in my humble opinion)

August 22, 2006 10:30 PM

Carsten Cumbrowski said:

Hi Jonathan,

Presenting a possible solution for a problem that still is a problem and tons of people out there that don’t even know that there IS a solution, is not a bad pitch. Its actually a good one :).

Anyhow, I read about your solution a while back (Revenue Magazine I think).

I have a question though and ask you directly, because I don’t want to dig through white papers etc. to find out.

Does your solution work for programs that do batch processing only or does it work also for programs that do pixel tracking? If it does work, how do you know for which network the user has a cookie and for which one not? Does the user still have Network cookies or is that replaced by a single cookie from your service? Are the optional url parameters networks have like CJ’s SID, SAS afftrack etc. supported and not getting lost?

I am just curious how this works.

Separate question. Do you have an affiliate program? It would be B2B of course. Not that it matters really. I would link to your service from my resources site anyway. My bigger problem is to determine where to put your service. I might have to create another category for you guys, but how would you call it? You might have an idea.

Thanks
Carsten

August 27, 2006 12:13 PM

Liz Gazer said:

article - it’s an important topic that I’ve heard discussed and participated in discussing MANY a time, and I’m glad to see that solutions are beginning to be developed.

And yes, congrats Jonathan on the patent win!

I like the idea of centralized tracking/reporting for all channels and would be open to finding out more about that. Failing that, I like Peter’s suggestion of using a script to read the cookie in order to determine WHICH of the tracking pixels to display but what does that script look like exactly?

I’d love to convince my merchant to implement something like this, and I’ve mentioned it multiple times to date already, but I get the feeling he doesn’t have the resources to figure out what that script is, and I figure many other small merchants who don’t have advanced in-house programming teams may be in the same boat.

Also, when using such a script, wouldn’t that have to mean that in reading the cookie, the cookie would have to include ‘click date?’ My experience is that not all network systems are set up to log this data, as is the case w/ MYAP (at least in version 8 - they’ve just released 9 and I’ve not had a chance to test drive it just yet).

I’ll be watching this topic and look forward to finding out more about new industry developments as they arrive.

Thanks again,
Liz

November 17, 2006 1:55 PM

Jonathan Miller said:

Thanks for the comments! Yes, we’re using the methodology very effectively to track campaigns running across multiple ad networks.

The centralization has enabled us to streamline reporting and administration accross our client base and the short-term wins have been simply phenomenal.

A v2 commercial release should be available in the next few months for agencies and affiliate managers who wish to license the patent and benefit from the centralized reporting and aggregation. If you’re interested we hope to launch with some fanfare soon!

To your question on the cookies. We don’t need to modify the behaviour of any of the subservient ad networks, as one of the ways the ForgeInsight platform works is to trigger the display of the originating network based on the business rules contained in the ForgeInsight cookie, which is set post impression prior to the network’s. As our cookie contains the datestamp we then determine which ad network to credit and allocate the sale from there.

This is just one of the ways that the system determines which ad network orginated the sale

J

November 20, 2006 3:36 PM