My question is, when I look at the Search Term Suggestion List, and type in a topic, I find many phrases that appear to have been searched for a number of times, yet I know from the complexity of the phrase, many words, different mispells,etc. that the phrase could NOT have been specifically typed-in that many times. How are these numbers gathered? The confidence I have in the numbers definitely affects whether I would advise a client to purchase a term or a phrase.

we get charged almost every other day to replenish our accounts. when checking stats we see little action from kannodle. the rep claims that referrers from their partners will show up as undefined in our logs. i asked if they could give us some tracking from their end. told me all they could do was give me the search term and when it was searched for.

they sent me a file of all terms and when they were searched for and the totals. their numbers are much higher than our total traffic, including kanoodle! this is leading us to believe that we are being overcharged!

we are in the process of setting up a mirror page, specific for kanoodle, with its own traffic counter. well see if the numbers match! will let you know!

anyone else experience anything similar?

There are a couple of issues going on here. First, to restate the question:
The advertiser asked about why some search terms show up in the search term
suggestions tool with a (seemingly) disproportionately large number of hits
given the obscurity of the term. For example,
30 no snowboarding, ski resorts in utah, alta, deer valley

This query was recorded 30 times in our stats system, however, it is hard
to believe that there were 30 separate instances where a user typed this
exact query without making any typos, changes or tweaks.

The data behind the search term suggestion tool is a record of all of the searches that we do and the unique search terms that were used to trigger those searches. As such, there are a number of events that can trigger the logging of a search in the database. These include:

1. When a user types a query into a search box on GoTo.com and clicks the Find It button

2. In certain browsers, and with certain cache settings, when a user leaves a GoTo search results page and then returns to that page using the back button (essentially re-querying for the same search string)

3. When a user bookmarks a search result page and then (subsequently) opens that bookmark.

4. When a user double clicks on the Find It button, we will log two searches for that query.

5. When a user is on a GoTo.com search results page and clicks the "reload"
button to see the next banner ad in the rotation (perhaps this occurs more
frequently with adult searches)

6. When a user's connection is slow and (s)he clicks the Find It button multiple times (see item 4)

Obviously, item #1 above is the "normal" action that causes us to log something in our database. We do not currently have data available on the frequency of the other 5 ways that items get logged. It is our belief that our data are all logging correctly, however, at low volumes it is possible
for "funny looking" data to emerge as a result of items 2 through 5 above.

Nonetheless, with regard to the 30 searches for "no snowboarding, ski resorts in utah, alta, deer valley", we are confident that we served results for that query 30 times. We cannot prove, but strongly suspect that
we served the result to only one user. That user appears to like skiing in Utah, but not snowboarding.

Phew! My apologies to those readers with a less-than-technical bent. Join us next week, when we will discuss the color green!

Hope this helps,

GoTo Guy

just for fun.. search for "casino" or "mp3" and compare top prices for these engines....

someone, please list every pay-per search engine that you know of... thanks

check http://www.payperclicksearchengines.com i think there are about 38 ppc search engines.

[This message has been edited by Marc_Wyss (edited 11-21-2000).]

because of the large number of affiliate sites and meta engines using GOTO's results, is this affecting the the stats in the suggestion box?? or does it report only from the GOTO site

I see many terms which appear to be from a directory listing such as "apparel - xxx", I see this format quite a bit

ktf68

And in response to Netcommr’s question, the Search Suggestion Tool does report searches across the entire GoTo.com network, including our search distribution partner sites.

In general I believe you’ll find the Search Suggestion Tool's results to be very reliable and useful. We’ll continue to work to improve the Tool and we expect that the next version we release will eliminate the kind of results delivered in this case.

Thanks,

GoTo Guy

This does clear a few things up.

Just to back you up on the overcharging issue some people have inquired about. I have watched GOTO.com like a hawk comparing what I am being billed for and what my logs tell me. All I can say is I am impressed.

They reply in less than a day to emails, (mostly less than 1 hour) with informative answers to my questions. I have no problems with the bulk submit. I'm not getting ripped off...

whew..I need more terms!!

What's really surprising is that I've emailed GoTo twice about this - and haven't had a reply - (GoTo Guy replied saying he'd passed my email on to the relevant dept - but nothing from them since). That's a shame - 'cos GoTo used to be really good at the customer service, and it's sad to see it slipping.

Don't get me wrong - GoTo is still good value - just not as good as it would be if this problem didn't exist - I now mentally double my bids to convert them into what I'm actually going to be paying. That's why most of them are still at 0.01 !

What are others experiences of the correlation between the GoTo reports and their own logs? Is this a common finding (perhaps caused by the nature of the "referrer" statistic in logs), or am I an isolated case?

I have not yet checked my logs to compare quality of traffic, but so far it looks like the Kanoodle traffic doesn't generate much sales.

Whats going on ?

[This message has been edited by JFK (edited 11-21-1999).]

The test was done by bidding the same amount ($ 0.10) on all search engines for a very popular keyword (705394 searches at GoTo.Com last month). The same URL and description has been used at all search engines:

7Search: 4 visitors
Bay9: 5 visitors
FindWhat: 4 visitors
Goto.Com: 42 visitors
Sprinks: 86 visitors
Kanoodle: 1705 visitors

Now it`s time for the real interesting stats:
My server logs show that only 10% of visitors from GoTo.Com did click on the link to my order page. Kanoodle visitors are worth even less. Only five (!) of 1705 visitors from Kanoodle did click on to the order page (all five hits on the same day). That is $175 wasted on Kanoodle fraud. Another interesting fact is that only five visitors actually did arrive from the main Kanoodle site. Most hits (1314) came from a Kanoodle Affiliate site: Searchcactus

Conclusion: I`m not sure about GoTo.Com and Sprinks and will continue to check them out. For now I will not spend much time at Bay9, 7Search or FindWhat. And I want my money back from Kanoodle.....

[This message has been edited by klovn (edited 12-16-2000).]