August 3, 2004 Archives

Chris Sherman starts off by saying that there is life outside of Google and Overture. He says that this panel will focus on the opportunities outside of Google and Overture

Peter Hershberg from RepriseMedia was up first, he is here to talk about the other types of engines. Paid listings distribution across the Web is covered by Google (53%) and Overture (45%). So the other 2% is about 2 million searches and the prices on Google and Overture keep climbing. The inventory of search ads is shrinking. So we have Tier II Search (FindWhat, Enhance, LookSmart, Kanoodle, Search 123) and Verticals (Business.com, Industry Crains, TravelZooo, Gamblling.com) and the Shopping Engines (Shopping.com, bizrate and pricegrabber).

The value of working with the alternative PPC engines are:
- More volume
- Lower minimum CPCs (as low as a penny) a good way to test
- Less competition
- Hands-on customer service (Tier I can not handle the customer service requests but Tier II's can)

You need to be cautious of the following areas with Tier IIs:
- Distribution can be poor quality
- Relevancy can be sub-par
- Search behavior might differ from Tier I to Tier II
- Lack of tools compared with Tier Is
- Fraudulent clicks (he went into a lot of detail on this, but I covered click fraud in this mornings session).

Chris Churchill from Fathom Online was next up. He asked who is concerned about the rising price of clicks on the Internet. Pretty much everyone rose their hands, not me, I am typing. Two Ways to Increase ROI (1) better buying and (2) increasing conversions. He briefly discusses the "power of conversion rates" with a 'what if scenario' slide. This session is about "better buying", and he complied data from the first half of 2004, 218 campaigns and 6.8 million clicks. They divided up the universe into tier I and tier II. He broke down conversion rates by Tier. He left out the names of which Tiers were and were not converting. The conversions ranged from 2.92% from a tier II, then a 1.98% from a II, and 1.58% from a II, then two tier Is at 1.37% and 1.12%. He also broke the tiers down based on conversions by industry. For some reason the financial industry performed better on a tier II then on a tier I. If Tier I's have 98% reach, why would Fathom have 18% volume from Tier IIs? Because Fathom leverages both, unlike many SEM companies.

Frank Watson from FXCM stood up without slides. He spends about $200,000 and $150,000 on Google and Overture respectively. He said that is all that is available, so he looks elsewhere. You should utilize the Tier IIs before your competitors do. Use analytics, use verticals, use international engines and test. Short presentation but he made his point - work hard.

Next up was Dan Ballister from FindWhat.com. He describes who is FindWhat.com is. They get 1.5 M clicks/day from 300+ affiliate sites, Espotting in 9 countries, premium private label partners such as lycos and verizon, and they offer merchant services such as miva drive traffic. Advertisers work with FindWhat because they reach a "different footprint," reach ROI goals through efficient bidding and two tiered customer support level. They see the space becoming more and more vertical, cross border marketing (i.e. espotting, mitsui, miva), fully monetizing every paid click (online, on phone, and on premise). They are encouraging you enter a phone number for you to track the on phone "pay per call".

Damien Smith from LookSmart didn't bring a presentation as well (not sure why). Can I reach my volume targets and my requested ROI? He says that is the important question. LookSmart wants to be able to help you by focusing on three things (1) quality traffic (traffic quality is not good enough in the industry today, they are working harder on it now) (2) prices low (3) robust tools. He gave out his personal number so you can speak directly with him if you not happy with LookSmart's services.

Q & A:

Q: Someone why do people participate or conduct clicl fraud?
A: :)

Q: What is the ideal time to test a campaign on these Tier IIs?
A: FindWhat asks for 90 days, Peter said he agrees but tailors it on the clients needs, Frank adds that there are many variables to take into consideration.

Q: How will FindWhat manage the Pay Per Call campaign?
A: FindWhat said there will be a separate area to manage Pay Per Call. Reporting will include pay per call with an 800# and not a hyperlink.

Q: Does LookSmart have plans to do pay per call?
A: They haven't looked at it yet, but its possible.

Q: Dana Todd asked Damian if he can go more into how CTR has an impact on listings?
A: Damian said its relatively similar to Google, just think of it at CTR x CPC.

No more questions for the panel...So we are just waiting for more questions. I am going to leave and post this. Off to the Google Dance in a couple of hours.

Forum coverage at Search Engine Watch.

Chris Sherman introduces the session with a quick summary and welcomes Massimo Burgio from Ad Maiora.

Massimo starts speaking in some foreign language, to make a point (which I expect he will get to). European market consists of 25 countries, where there are 20 official EC languages. Europe is an active and major portion of the internet population. Search queries with european languages are 1/3 of all searches. Searches per search engine users are the highest in the UK. Europe is a very complicated market. In Europe there are some major local search properties; such as Virgilio, Wanadoo, T-Online, Terra Network, Spray and more. If you are doing SEM in Europe, you must look at the local engines. The UK is the most active market in terms of SEM/SEO services, search adverting spending is 33% (average is 10%), rising stars in the European area are Poland and Scandinavian countries. IAB/EIAA, Jupiter, and SEMPO are all very active in Europe. Many independent search events are also taking place (funny, there was a typo on the slide but that is ok, its a European session, not US session). He did not want to get into the technical side.

He then gives us a little case study on Wyndham hotels. The target countries include; Italy, France, Germany, Spain. But the .com Web site was only in English. PPC networks such as Google, Overture and Espotting were used. They saw that only traffic was coming from Google. The ad editors' specs change over time and are not the same from country to country. So they began building specific ads and landing pages for each specific country. They are in the process of building vertical micro site for his country. He collected some nice data on what types of locations people from Europe (broken down by country) would like to travel to.

Bill Hunt from IBM was next up. He manages IBM's worldwide strategy for search. IBM has 83 localized language versions of its site representing 31 countries. He has developed a clear management system for SEM, this way they make sure to apply the same things in every market (just fined tuned for each market). Common problems with international SEM include; (1) all the problems faced in the US in English, (2) Not thinking like the consumer in the market, (3) Poor quality translations typically not optimized, (4) Lack of centralized approach, vision and support, (5) Lack of resources - people and money, (6) faulty or no keyword research in local language, (7) multiple simultaneous campaigns - partners & affiliates, and (8) poor or inefficient navigation to and from country sites.

The Global SEM process:
Market Research -> E-commerce Strategy -> Strategy -> Conduct keyword Research -> Localization -> Optimization -> Measure

Barrier # 1: Local market search engines restrict pages to those with local languages or local top level domain, 90% of Europeans use the local languages version of the search engines.

The US centric Google.com brings up IBM as the first result for "ibm thinkpad", the second tab they picked (i think german language) brought up the wrong german page, the third tab was german located pages only which requires you to have a site that lives in germany or the domain suffix must be .de.

Removing Location & Language Barriers:
- Use correct meta language tags (html lang="de") and (meta http-equiv="Content...)
- Use local domains (i.e. .de, .fr, .co.uk), they can be hosted in the US and at least a few pages on the local market domain

Barrier # 2: Getting crawled is a major problem. Example, pop up or pull down country/office maps are not being crawled. Restrictive JavaScript language detects pushes hurt as well.

Barrier # 3: Cheap translation is just that...cheap. Translators are not good optimizers (mostly). Many translators do not use the internet often. Few translators don't understand keyword research. Translation tools typically kill current optimization efforts. And None of the major localization firms currently use keyword research as part of their glossary development or translation process.

Barrier # 4: Keyword variations and mapping.

Harrison Magun from eONmedia was up first and will be focusing on PPC in Europe. Why do Americans go to Europe? (1) Drink Beer, (2) Take Pictures, (3) To Be with other Americans who like to take pictures and drink beer. :) But really: (1) increase distribution, (2) competitive advantages, (3) first mover opps, (4) leverage foreign exchange and regional pricing advantages. Two main goals to increase sales and profits.

190 Million US internet users versus about 170 European internet users. Market growth in Europe is much higher then in the US.

What kinds of companies should market in Europe? (1) Downloadable applications do not require shipping, (2) Hotel and air, (3) Fragrance and Beauty, (4) Media, and (5) B2B/Wholesale. Who should not? (1) Restricted products, (2) Consumer electronics, (3) Automotive, (4) Online/Offline education, (5) Leads for US based services (credit cards, mortgages, etc.).

Linguistic and regional elements are huge. (1) Make sure the ads are relevant, (2) the landing pages as well, (3) translation site (merchandising and pricing, fulfillment and CRM), (4) Competitive strength and weakness versus regional.

Effective AdWords listings from a foreign company in the US, "russian souvenirs" brought up "Russian Unique Doll" with the description on a Russian Doll Bottle Holder. Brought up a funny product which had nothing to do with the search, (or did it?). Effective site for the US work the same way, he brought up a very funny example of a European company's language into English.

Q & A:

They had special Q & A people on the panel: (1) Peter Celeste from Overture and he is into launching new markets, to work with the current markets to help accelerate revenue, and they have a new team to help US based companies go Europe. (2) Tor Crockatt from Espotting was on the panel to help the SEMs get to Europe by helping you localize your marketing efforts. And (3) Derek Preston from Marago the European search engine.

Q: How do you work around having multiple languages on a single page?
A: Bill responded they have a rule that they do not put more then one languages. But the search engines do put weight to the dominate language on the page and the tag. A problem IBM has is with support content in China where the customers do not want translated support content because they are not 100% confident in the accuracy, so ranking those pages are hard.

Q: Question is to Peter from Overture, he was wondering how many US companies are going Europe.
A: Peter said many are. Bill adds that many large companies are moving in that direction. Sessions like these keep getting larger and larger.

Q: How is it best to manage content?
A: IBM uses 14 different CMSs in its organization. Depends on the organization side and needs...

Q: Do you have keyword research tools for this industry?
A: Espotting has a few tools that allow you to do this over multiple networks on the languages. IBM said there isn't any tool that does it, they use Overture, Espotting, WordTracker and other tools and do their best. Espotting added that the UK had the most "comparative searches" but Germany and Scandinavia has more "product specific searches".

Q: What if you don't have the resources to do this?
A: Well, if you don't do it and your competitors are then you can not compete. People normally won't transact from a site they don't understand, its scary.

Forum coverage at Search Engine Watch.

Andrew Goodman moderated this session. He introduced Jessie Stricchiola from AlchemistMedia. The room is pretty pretty much filled up, I think I attended this session in NY and it was not as filled (maybe it was the clinic). CPC Ads are Not All Alike she said. If one thing is for sure, everyone is talking about how search ads are very different then contextual ads. You can not control/select/analyze contextual ad distribution, cost, etc. You can do certain things to optimize your campaigns by testing ad copy with contextual only (like Andrew discusses in yesterday's morning session). Moving forward we should tell the CPC companies to provide more separation between search and contextual, ability to view this data, and we need more data on the users and patterns.

Jessie tells us about the "bracket trick", which she told the audience in NY. (1) {KeyWord:Long Beach} = All words with initial caps (2) {Keyword:Long beach} = First word capitalized (3) {keyword:long beach} = All words in lower case. This allows you to dynamically put the keyword the searcher used in the engine in the title of your ad with these brackets "{keyword}'.

To keep your CTR high, turn off ads at times where the CTRs are low. Day parts tracking help with this. She says "test, test test..." your elements (call to actions, text, graphics, forms, links, navigation), characteristics (positioning, color scheme, form usability, and word content). Test and fine tune. She mentioned this tool named optimost.com which is a 3rd part dynamic landing page optimization service (sounds very interesting).

Misty Locke from Range Online Media is up now. How does the ad meet the conversion? She talks about her 6 key reminders. (1) Who are you targeting? Know your audience. (2) Conversions can vary depending on keyword and landing pages AND on CPC engine. (3) Landing pages should directly correlate to keyword and placement. (4) Remember the convenience of the online shopper. (5) Take a deep breath - use the 2 to 3 click rule, don't make the user to click too much. (6) Think conversion not traffic when selecting search terms for interior or product pages.

Determine most relevant search terms, consider the product inventory (how much you have in stock), which products have the highest profit margin? Then identify your biggest competitors, what are they ranking for? are they buying on those words? Where are those landing pages? And what sets you apart from them. She then goes over some basics which you all know, since your reading this. :)

Pier 1 Case Study:
High CTR but low conversion rates, which was not great. They wanted to increase conversions, and ROI. They improved user experience (landing pages), detailed more keywords and targeted only the pier 1 shopper. She wrote creative that was targeting the Pier 1 customer only. The CTR was about 20%, which was excellent. They then added more 'feeling' to the ad, the CTR increased about 10% (to 30%), but where they buying? Problem was that the customer didn't know how to navigate from the home page to the product. So they took them directly to the product or category pages. Site conversions increased 4 - 6 %. Keyword conversions increased from .37% to 1.79%

Lee Mills from BeyondClicks is up next. He said include keyword in ad creative always. You also want to include the keyword in the ad landing page and a strong call to action (offer free stuff if you have to). One of his clients is anonymizer. The landing page must have multiple call to actions, multiple ways to buy (top and bottom of the page). He said its also good to put price in your ads (if you are a low cost provider). They are more likely to buy if they know the cost. He stresses, like the other speakers, test continuously!

They did some A/B testing on landing pages. B2B example: Page A had a 3% conversion rate. So they made Page B which was simplified, and its conversion rate was 18%. That is huge! B2C example: Page A had a 3.2% conversion, they made a longer page with multiple offers on Page B and the conversion rate was 9.6%. He said if you have scrolling pages, then put an additional offer (call to action) at the bottom of the page. Do not use pages that do not allow for navigation to your other pages, I see this often with landing pages - he says it doesn't work (causes lack of credibility).

Q & A:

Q: With the "bracket trick", now people are using 3 to 4 keyword searches, how much longer will this trick work?
A: Jessie said it works well in some cases and not in all cases. Ideally, any CPC source that allows this, you can bet they will expand it. With Google, they only allow 6 keywords - so its a product fault. But sometimes you should break them down into static ads.

Q: How do you work with companies to change their navigation?
A: Start slowly and show how it works.

The same guy made a comment about how nice it was to see a female panelists. The panelists liked that, thought it was funny.

Q: My clients are dumb, they want to my "town real estate" keywords but he said they don't convert.
A: Test different styles of ads; personal ads versus corporate like ads. "Meet Mrs. Real Estate Agent" versus "Meet Company Real Estate". The individual, personal ads seem to do better, because its like a word of mouth referral. Do what you can to capture that lead, maybe a white paper, maybe something else. Then after you get the lead, have your sales people jump on them.

Forum coverage at Search Engine Watch.

Jessie Stricchiola starts off with a brief history of how she got into click-fraud. She had her tech team build a program to monitor click-fraud. Then identified two of the client's competitors who were clicking on the ads. The average costs were $1,500 per day! The company that was doing this fraud was ironically a law firm that worked on Internet fraud. Then GoTo.com (2002) gave the client a refund.

Fraudulent clicks come from:
- Competitors clicks; often manually, sometimes via automated 'hitbots'.
- CPC affiliate clicks; often via 'hitbots' and is generated by PPC partners from various locations. Its hard to detect these things and the CPC engines are also working on identifying and blocking it.
- Impression Fraud is an other concern. One of the criteria of ranking well in Google AdWords is your CTR. The higher the impressions but the lower the clicks, this can ultimately lead for your ad to be removed.

Auditing Click Fraud:
- Set up tracking URLs
- Data you need to collect is the URL request and the Visitor behavior data (after the click). She mentions three Web analytics tools (Urchin, WebTrends, ClickTracks) and Bid Management (did it, atlasone and keywordmax).
- Configure your reports to store as much information as possible (ip, os, browser, keywords, source, etc.)
- Then you do analysis of these reports by compiling click data per keyword and per cpc engine.
> average daily clicks
> average page views per click
> average conversion rate per keyword per click
> PPC network partner sources (international referrers)
> hourly click trends, etc.
- What are click fraud indicators
> abnormal spikes
> abnormal clicks increases plus atypical visitor behavior
> more than one competitor dropping out of contention
> non converting cpc network partners
- Presenting the Information to your CPC Engine
> be thorough
> document your analysis
> record all data
> take screen shots

She said Google is not as helpful as Overture when it comes to this stuff.

- Action Items
> contact your competitors if possible, see if they are experiencing the same issue.
> contact your cpc account representative
> continue to monitor your click activity

She adds that she feels one day they will be able to sue people and take them to court for click-fraud.

Lori Weiman from KeywordMax is giving a case study as I type. She started off describing how click fraud happens. They might click on your ad manually, or use a bot with a way to mask their IP. Then you have the stupid click-frauders that just click without masking IP. The affiliates are the smartest click-frauders, they mask IP, break the referral url and break your own tracking URLs.

How do you catch it? Use a tracking system that captures IP addresses, use a tracking system that captures the referrers. Audit your bills and what for click spikes, also watch for clicks from irrelevant geographic regions.

Case Study A:
Client A notices an ongoing spending increase of $30,000 per month on Tier 1 engines. It was detected by receiving threatening email from a disgruntled ex-employee, using keywordmax they saw 0% increase in conversions. The fraudster went to jail! This client had to ask for refunds, and it took a lot of time. The CPCs should have been more helpful.

Case Study B:
Client B had an affiliate fraud problem. He notices click spikes of 5x coming form a tier 2 engine, it was tracked and reported. The CPC engine refunded the money and they booted that affiliate. CPCs need preventive measures in place to stop this from happening in the future.

How to file a complaint?
- 60 days to file a complaint in writing
- Very thorough documentation and proof (keywords, date ranges, etc.)
- You will need to follow up with emails and phone calls.

What can the engines do to be more proactive:
- Detailed billing just like your phone bill. We want an itemized click bill.
- Set up Fraud Departments to handle this
- Communication and Willingness to work with customers

Danielle Leitch from MoreVisibility was up next.

Case # 1:
She saw normal traffic based on her Web analytics but when she looked at the AdWords report reported 5x more. She noticed a huge increase in clicks from the same keyword phrase, will a zero conversion rate. She basically had the same thing to say as the other two speakers (but she made a point to say that she never spoke with the other two speakers before today).

Case # 2:
Overture reported a 10x increase in overall clicks on a single day. Since there are no daily limits set, it could be very costly. Red flags were raised all over the place based on traffic, conversions, clicks and more. They looked at the server logs and found a pattern, they also looked at tracking URLs. She called Overture and reported the information to Overture. 1 week later she got her refund, which was good. They reviewed the previous weeks as well and credited for previous history as well.

Remember if you do get a refund, you must adjust your reports accordingly. Update your Web analytics and report data.

Q & A:

Q: Why are there no search engines on the panel?
A: The search engines declined to join.

Q: Do you think if you did not report this click fraud to the engines, you would have received a refund?
A: No. Jessie said Overture has done a lot to increase the comprehensiveness of the analytics and increased the support staff to support this. They can't disclose what they are doing because if they do, the spammers can work around it. She still says they should disclose it to the advertisers. If you are getting a refund, you should know exactly what you are being refunded for (which keyword, times, etc.), Lori adds. Danielle does not fault the engines, she said its up to us to report it to the engines. But Jessie does fault the engines. She points out the overture reps in the room.

Q: How do you contact the click fraudster? What is the best way to do it?
A: You must have evidence. If you do, then contact them.

Q: What percentage of overall traffic is click-fraud?
A: Jessie said its nearly impossible to figure that out because click-fraud is dollar, niche and keyword specific. She said there are probably verticals that don't have this problem, some keywords are hit more then others. She said "within competitive keyword phrases ($2 - $3+ per click), based on her opinion, about 10 - 20% are fraudulent.

Q: Did you ever make a case where you reported fraud and not get a refund?
A: Yes, and the CPCs were right in the case that they were not fraudulent. Sometimes there are partial or no refunds based on looking deeper in the details.

Q: Do CPC engines block IPs?
A: They handed the mic to Overture to answer the question. They sometimes do it but they rarely block IPs, they write filters to increase their fraud algorithms.

Forum coverage at Search Engine Watch.

Danny Sullivan begins to say that his past keynotes were all about telling the audience who owns who and which engine powers which search. But ironically he quickly goes over the "who powers who" diagram.

SEM Threats:
- Contextual Pollution;
Results only come up for a keyword search I desire. This works differently then viewing ads on a Web page based on browse mode versus "quest mode." Danny said "contextual ads" are not "search". Just because an ad is CPC-based doesn't make it search. Lumping the two together pollutes the data. Are you a performance marketer or a search marketer? Its important to separate these two types of marketing tactics out. One is search and one is contextual. Just because the search engines are providing this technology, doesn't mean they should be looked as - as one.
- Agency Money / Branding Buck;
There is branding value here based on the GoTo's 2001 study and IAB's study in 2004. He says the pie of a company's marketing budget isn't bigger, but you need to get a larger piece of that pie. The search companies are providing support for this primarily on the PPC (ad side). But SEM is not only ads its also SEO, which is Public Relations. The support on the SEO side is an issue, other then PFI, there is no support.

SEO Lives:
- SEO should have died
- Google has kept SEO alive and revived it (he notes a thread at WMW that discusses how 3 bots revived SEO.
- You can sell ads but people want the PR (SEO) too

He then goes into a case study, a funny example of his 3 and 5 year old boys. His children call flash lights "flash torches" because Danny is American (flash light) and his wife is English (torches). So his kids are not hitting either market with the name "flash torch". So they can buy ads but they get nothing on the SEO side.

What is Needed for SEO Support:
Danny basically pulled the info from an SEW thread (t=197);
- Algorithm shift warning
- More authoritative info
- Express Spam Report
- Public spam reporting and checking
- Paid support program
- Search query stats
- Complete crawls
- Partnership in attitude on both ad and free side
- Commissions? Protection from direct sales? Certification?

What's the incentive for search engines to do this?
- Helping to win the ad spend but can't be ad-only shop
- It's not exactly like newspapers. You do need PR support and can do it without violating the church/state divide.

SEM Reputation Problem?
To get more support, we need to deal with the reputation problem in our industry (reference the SEW thread in the forums). He also quotes the marketing guru, Seth's blog entry. He pleads that SEO is not a black art. He discusses the SEO contest and how link bombing worked. The customers when shopping for an SEM firm are afraid. Fast Company releases an article about the Google Dance 2003 with the title "Shmoozing with the Enemy."

He goes through some spam examples on the Web, with tons of links at the footers of pages. The search example was "san jose radio flyer" in Google. Then he did the same search in Ask Jeeves. The text is not hidden, but pages are filled with text that mean nothing to the searcher.

He says there are good SEM firms and good stories.

What do customers want?
- Traffic that converts (not just top rankings), check references
- Not to be scammed (maybe someone should start a public forum telling people about these companies)
- Not to be banned

Solution:
- Code of Conduct? "I will do nothing to harm search engine relevancy" but this is all very subjective.
- Enforcement/Review
> seopros.org
> seoconsultants.com
> any standards are open to debate

One thing Danny thinks will help is if the search engines themsevles get involved.
- We need something, not sure what.
- Search engine involvement in proactive way would greatly help (goes both ways)
- Will it lock some out? Some people will be locked out but they will have to deal with it. Its not just a white hat versus black hat. He said there is a lot of gray in the industry, because its a very complicated area. I wonder if the audience understood "White versus Black"

He now gives the audience a peep talk about all we have done. SES will be giving out marketing awards in the future to reward us.

Forum coverage at Search Engine Watch.