Intro: Detlev Johnson- Position Tech
Gord Hotchkiss - Enquiro.
Learned how to make a better PowerPoint presentation from his 9 year old daughter last night (laughs).
“Golden Triangle” Eye tracking study with 48 people using 5 different scenarios: “free” to completely scripted. Looked at how they interacted with Google only. Full study released June 30, 2005. Study showed that the aggregate activity indicated that the top left triangle of search results (organic and paid) is the most popular real estate, thus called the "golden triangle."
How do “we” navigate a page?
Start with orientation and start looking at various aspects. Fixations (momentary stop of eyeball) versus significant scanning versus actual clicks. Shows where people actually start reading listings.
1. Fixation: most people do look at top sponsored listings if on left 43.2%, some look at “local results” 14.6%, then go onto first organic 38.7%
2. Scanning 31.2% top sponsored listing, 14.6% in “local box.“ 46.7% top organic spot.
3. Clicks: 13.5 top paid, 2.5 local box, 42.2(!!) top organic.
Paid search right side results, Fixation: 1%, Scanning: 2.0%, Clicks: 4.2% (see migration below)
First process is fixate on title, if interesting, read content, if want to, then click. This resembles an “F” pattern.
Do see some migration to side results before actually scrolling down for further organic results. Has found that title tags are very important to visitors, since it is the only thing they have to determine whether to read or not.
Individual “Semantic Mapping” defined: if nearly identical results occur in title, interesting occurrence in that if one particular phrase repeated, more likely to be clicked on by second or third mention. People are scanning listings looking for their own semantic map. People look for bolded search query, icons, prices, consumer behavior & comparison, product details, especially price, trusted brands (as a pull, not push interaction), geographic references.
First search vs. “go-back” or repeat search. Golden triangle breaks down completely on second search, scanning activity is more thorough after going back.
Important to do a “SERP SWOT analysis”: Who is there? What are they offering? Are they providing matches for the searcher’s semantic map? Need to know who else is there.
Summing up: must understand behavior or YOUR market. What is their primary real estate? How do they scan listings? What is their semantic map? Understand difference between first/second visit? Do a SERP SWOT.
Jonathan Mendez, Digital Grit, Inc.
“Understanding your customers improves your business!”
Searcher behavior gains insight into:
Keyword pahrses
Natural search use
Paid placement use
Competition review
“Nothing is better than face-to-face interaction with clients in order to gain their goals.”
Discovery and Understanding user goals is key to successful SEM. Must differentiate betwwen "defined" and "latent" goals.
**Defined goals are easy to discover: research popular keywords.
**Latent goals: “Invisible keywords” that remain inside user’s head. If you can draw these out = success. Latent goals focus more on relevance and are often verbs.
Case study:
Client email solution provider for small to mid sized business
How user goals were evidenced: only two kw strings were repeated out of 43 different kw searches to find one tool.. "Create," "send," and "manage" were not used in search a single time, even though this is what they wanted (the "latent" keywords).
Kw strings started general and became more specific as users became frustrated w/results.
13% of respondent performed industry specific search.
When invisible kws were present, users were called to action, if not present, then they were not likely to click. Clicks were evenly spread above the fold.
Results: earlier in the results that the latent kws are represented, the higher chance of clicks.
8 of 15 users specifically avoid “ads,” but many users did not realize they were paid clicks. Right side AdWords were generally ignored.
User behavior and competitive factor: 9 of 15 were looking for a box software vs. web-based application, and box software apps are marketing this
How to understand goals better? learn by watching and listening to customers.
Allison Kane and Inga Johnson, Atlas OnePoint
Discuss repeat searches and appropriate evaluation of kw campaign portfolio.
Yahoo study found that users search between 12 and 15 times before actual purchase. What does this mean for individual advertiser?
How often is an individual advertiser's keyword clicked on before purchase? Looked at repeat search behavior: 43% clicked on multiple results, an average of 2.3 sites visited per search. One patern that became apparent was the use of an advertiser name as last search query (a "branded search"), with 26% on average being a navigational search. Yahoo says 91% of consumers converting on a branded search started with a more general search.
Inga Johnson: principles apply across the board for both paid and organic search, important to use a single tracking method to understand more fully the searcher behavior. Also must study the search cycle instead o just the converting search. This enables you to find the overall cost of the sale broken down to particular to calculate Return on Ads Spend. If attribute the sale to the last click, harder to accurately determine ROI.
If the conversion cycle is long, the likelihood that navigational searches will precede a buy increases.
Many factors influence valuation of kws, Atlas OnePoint feels that all keywords in the sequence should be take part of the credit for the conversion. Things that must be studied include:
Natural search rankings
Average time to convert
Categorical vs. specific
Other media
What would happen if you took a keyword out of the sequence? Would the visitor still come back and convert? Goal is to not discount prior searches, but instead to look at entire cycle. Anecdotally, a client that tested pulling lower converting terms out of basket had a negative result on sales. Should give some credit to lower conversion terms as being part of the overall sales process.
Cam Balzer, Performics
Discusses how people act on the way to a purchase.
SEM is increasingly competitive and challenging. Looking beyond pure ROI to branding and sales cycle.
Competing effectively does require technology and expertise, as well as insight into customers as searchers This will make your campaigns more relevant and effective.
What do people do before last clicks?
Study: e-Commerce sites in 5 verticals. Looked at all searches during 12 weeks prior to purchase. As opposed to Yahoo study, went from last search and looked back. In order to identify and categorize searches into “generic” vs. “brand” searches.
Roughly half of buyers made a relevant search before online purchase, as far back as 12 weeks, mostly generic activity. Majority of users did last search before purchase MORE THAN TWO WEEKS prior to purchase. Ie: found a trip to Chicago and then made a decision before retuning significantly later to actually purchase. This shows that online purchases take time, and it is important to use a reasonable window to track online purchases (longer than two weeks.) Go from sessions-based cookie to longer. For example, 7-day cookie is too short. Results of increasing cookie "time" were dramatic: Tracked searches on generic terms increased 85%, ROI on generic terms improved by 41% and sales from brand terms increased 20%. This allows you to better see the value of generic terms and bid accordingly.
Most buyers never search on brand, however, as described by Atlas, many people switch to brand searches for purchase. The crux is that “generic searches DO provide a branding opportunity.” Recommendation: Factor in the value of non-branded searches early in the buying cycle. Fully utilize campaign data and incorporate performance results into campaigns.
Anther case study found that generic keywords such as “shoes” etc, drove offline and online sales, as well as increased their brand recognition. By buying “jewelry” for six weeks, a retailer was able to increase their brand awareness. Must look at broader long term benefits that can be derived from search.
“Know the searcher to win at search.” Must have technology, expertise AND insight into customer behavior to succeed.
Q&A:
“How did you track offline sales, Cam?”
Used “Dynamic Logic” in this study.
For Jonathan: “Difference between nouns and verbs used in searches in focus study?”
Jonathan: Shows that verbs really are the calls to action when determining that goals are genrally verb-based?
Detlev: anecdotal evidence suggest most queries are nouns.
General: “Looking at click before final conversion in terms of brand lift. Have you looked at brand lift more generally as to the value of just appearing on the page?
Jonathan: would be nice to study, but hard to drop “impression cookie” to identify if people saw your ad. Currently limited to using third party data sources in order to make such inferences.
Cam: Most branding is done by images etc, difficult to related this to search because it is mostly text.
For all: “Has any study been done in regards to searches behavior for international markets?
A: Most studies are US-based, but we can estimate that the patterns would hold true, unless reading left to right (lol).
For all: “If low volume searches are part of the search stack that leads to a conversion, why does Google disable them so often?”
Allison: Ask Google. :)
rustybrick in Search Engine Strategies 2005 San Jose at August 11, 2005 6:01 PM Comments (5)
