Identify the enduring concepts of Internet Search for SMEs.
Rob McGonigle, P2063526, March 2007.

1.           Abstract

How can Small to Medium Enterprises, SMEs, be found online?

Aims and Objectives.

·               Verify that the search market is stable and mature
·               Identify lasting paradigms to test for best practice.
·               Assess the impact on fifty companies of:	▬     Search usability
	▬     Search engines.
	▬     Website publishing software.

Previous research has revealed that SMEs only invest in online presences with a strong business case based on growth or customer requirements.

Enduring paradigms were identified from search engine manuals across this decade. This literature provided evidence of search market stability but further market research investigated rapid searching growth.

Usability was largely covered using literature taking the user’s view of search engine history.   

Four methodologies were deployed; functional tests, data mining of visitor logs, pay-per-click advertising and word counting.
Data mining generated more evidence than envisaged whilst introducing subjective and logical errors. Simple functional testing generated pass or fail results without problems. The pay-per-click methodology showed promise and speed but can be deployed more rigorously.

Strong evidence was found that people were searching for every company by name and by products or services. The wrong choice of technology can render companies invisible online.

Two paradigms generated compelling evidence of efficiency; good page titles and rich content.

The observation that word ordering multiplied “The Long Tail” of search queries, Fig 1, provided insight as to why good, natural language, web-pages are so effective online.

The paradigms list was too long for SMEs no matter how simply expressed. SMEs require business case recommendations, based on their company plans and concerns.

Deploying paradigms that can last the 18-36 months between changes, typical of SME websites, does not mean that SMEs can continue like this. One further research question sought easy ways to change websites.

2.           Background

a.             a..... The Problem

·               The size of the internet.

·               Most searchers, 62%, do not look beyond ten results, iProspect (2006).

·                SMEs only invest online with strong business cases linked to growth or customer requirements.
Levy (2003), Beckinsale (2003).

b.            b..... Questions

·               Is the search market stable?
The search manuals over time indicate search is mature, Table 1.
Search market growth continues in new contexts, e.g. MySpace, that engineering SMEs rarely address.

·               What is the impact of usability?
Google’s relevant results, Brin (1998), simple interface and results, Fig 2, refined by usability testing, remains the benchmark.

c.             c..... Questions tested

i.               Are customers looking for these SMEs?	ii.             How much more trusted is ‘.co.uk’ than ‘.com’ in the UK?
iii.           How many words are required for effective web-pages?	iv.            Are web-pages optimised for apposite keywords?
v.              Keyword density or readability?	vi.            How important are keywords in page titles?
vii.          How important are human edited directories?	viii.        What is the effect of site technology and structure?
ix.            Do SME websites improve or regress after changing?	x.              What is the effect of changing content?

                                                                  

3.           Methodologies

The three main methodologies were:

a.             a..... Data mining of visitor logs.

i.               Search terms for each company were filtered from log-files* into spreadsheets using Sawmill, see Fig 3. *From www.wmccm.co.uk
ii.             Common themes were identified, filtered and inspected iteratively, in manageable chunks, and tagged as positive, inappropriate, etc.
iii.           All terms tagged positive and inappropriate were excluded to facilitate inspection of the remainder. Table 2 illustrates this process for Auto-txt.
iv.             Processing almost 18,000 terms introduced consistency errors. Terse and ambiguous terms introduced subjective and logical errors.

 

b.           b..... Functional testing.

c.             c..... Pay-per-click advertising.

i.        Create pay-per-click advertisements bidding for key phrases so adverts appear regularly, Fig 6.	Fig. 6
ii.      The impressions (views) indicate searchers are using key phrases. (Impr. column Fig 7.)
iii.    The click-through-rate, CTR, clicks to impressions ratio, measures the advertisement’s relevance to searcher goals.
iv.     Different advertising copy can swiftly be compared when alternated evenly.
v.       PPC can test branding, titles, and test market products before development, Marshall (2007).
vi.     PPC results before improvement were constrained by SME funding. Known best practice was deployed as soon as problems were observed.

 

4.           Results

a.             a..... Technology

Web publishing technology can render websites invisible or ineffective when used wrongly, Table 3. Frames are no longer used.
FLASH technology failed in all five sites. This sample is not sufficient to damn FLASH technology and FLASH can work but with budgets beyond our SME companies. FLASH inside navigable sites can be very effective.

The SME publisher experience is getting worse with the initial 70% success rate dropping to 50% second time around, see Table 3 highlighted totals.

b.            b..... Visitor Data

The detection of brand searches proved that searchers were looking for every company, averaging 65 visitors using 25 different brand variations. The variations mode of 13 was exceeded for 54% of companies. Diversity of brand terms was confirmed by pay-per-click testing.

The web-server data analysis was prone to intrinsic subjective and logical errors. These errors were not symmetrical as negative terms were easier to identify for ‘used machinery’, jobs, etc. The boundary between positive and probable (green/yellow, Fig 8) was difficult to determine. The existence of ‘highly probable’ terms that just failed the ‘positively looking’ test means that the positive side of Fig 8 (blue & green) is clearly conservative.

c.             c..... Keywords in page titles

The percentage of website visitors for each company using keywords from the page titles was mapped against that company’s unique titles and keywords, Fig 9. Additional titles for webpages, with varied content deploying additional unique keywords, garner more visitors and tend towards 100%.

Well titled pages can generate 80% of their visitors from title keywords. Visible, bold, larger font, title keywords in search results, Fig 2, encourage searchers to click. The order of words matters, confirmed by pay-per-click testing, and explains why the search engines vary results by word order, Table 4. The closer titles get to echoing searchers queries, the more they click.

d.            d..... Other results

i.               Dot-co-dot-uk is more trusted than dot-com in the UK. Click-through-rates were doubled in pay-per-click testing.

ii.             The pages were not optimised for inappropriate key-terms, averaging 1.2 visitors per term.

iii.           Aggregated ‘printer friendly pages’ attract many additional visitors. Despite the loss of unique titles and unique title keywords, these worked because the original pages did not have optimum titles and too few words.

iv.            The optimum number of words was gauged to be between 200 and 1000. The page sizes reviewed were not distributed evenly and the range was wider than the 400-800 words Web-Position recommend from their much larger sample, Winters (2004-6).

v.              The Skillspin case study demonstrated that changing and maintaining web content improves search engine results. Google’s patent describes how complete web-site makeovers will have common “boiler plating” changes discounted. USPTO (2005).

vi.            Skillspin also demonstrated that words and pictures deliver customers. Words and pay-per-click advertising attract visitors but attractive pages, with appropriate illustration, are required to convert these to business leads.

5.           Conclusions

a.             Stability, technology and change.

The search market is stable and mature but growing strongly. Growth is from new vertical markets (YouTube, MySpace) and more frequent use generated by browser search bars, etc. Enduring paradigms were identified, tested, verified and distilled into one sheet with ten points. SMEs disconnect after three ideas unless these address their specific problems.

Web-publishing technology is increasingly problematic for SMEs and a barrier to changing web content. How to make
web-publishing easy to use, generated further research. SMEs benefit from independent, individual, advice.

b.            Deploy well-titled pages.

The predominant, first priority, recommendation to SMEs was to add and improve page titles.

The power of well titled pages, accurately signposting content, will endure as long as search engine results use text with titles first and largest. To present relevant content to users the search engines create advertisements, their results. Excellent content with poor titles, failing to attract clicks, will drop down the results.

c.             Create pages with rich content in good natural language.

The observation that search query word order generated different results, Table 4, combined with the diversity of search terms from the visitor logs helps explain why good natural language copy is recommended. Echoing back our queries and ideas has been proved to increase click-through-rates so search engines note word order. Word order confirms search queries as the longest tail of language. Anderson’s model, Fig 1, based on entities such as CDs, distinguishes short head ‘hits’ from ‘long tail’ niches. The visitor logs reveal ‘short head’ terms in the long tail due to word order. Pages with enough words, in the human voice, can garner visitors that pure repetitions miss. Varying brand names, renouncing habitual repetition, can also generate more long tail visitors.

 

6.           References

Anderson, C. (2006), The Long Tail. How Endless Choice is Creating Unlimited Demand, London, Random House Business Books.

Berkinsale (2003), SMES and Internet Adoption Strategy: Who do SMES listen to? [online], LSE. Available from: http://is2.lse.ac.uk/asp/aspecis/20040016.pdf

Brin, S., Page, L. (1998), ‘The Anatomy of a Large-Scale Hypertextual Web Search Engine’, in: Proc. of the 7th World-Wide Web WWW Conference, 1998.

iProspect (2006), Increase in the Importance of Attaining Top Natural Search Results – Press Release [online], iProspect. Available from: http://www.iprospect.com/media/press2006_04_11.htm

Kennedy, R., Kent, T. (2005), Search Engine Optimisation and Marketing for Beginners, Boca Raton, Universal.

Kent, P. (2004) Search Engine Optimisation for Dummies, Indianapolis, Wiley.

Levy, M. (2003) ‘Exploring SME Internet Adoption: Towards a Contingent Model’ Electronic Markets V13/2 pp 173-181, Routledge.

Marshall, PS. (2007), Ultimate Guide to Google AdWords, Irvine, California, Entrepreneur Media Inc.

Nobles, R., O’Neil, S. (2000), Maximize Web Site Traffic, Holbrook, Mass., Adams Media.

Sweeney, S. (2003), 101 Ways to Promote Your Web Site, Fourth Edition, Gulf Breeze, Florida, Maximum Press.

USPTO (2005), United States Patent Application 0050071741- Information retrieval based on historical data [online] United States Patent and Trademark Office.
Available from: http://appft1.uspto.gov/netacgi/nph-Parser?Sect1=PTO2&Sect2=HITOFF&p=1&u=%2Fnetahtml%2FPTO%2Fsearch-adv.html&r=1&f=G&l=50&d=PG01&S1=20050071741&OS=20050071741&RS=20050071741

Winters, B. (2004) Maximize your rankings by leveraging the five areas of the WebPosition Page Critic [online], Market Position Newsletter, 15/11/2004. Available from: http://www.marketposition.com/blog/archives/2004/11/maximize_your_r.html