Introducing PPC information retrieval

Computational advertising is a fairly new coined term and is comprised of three elements:

  1. Sponsored search (SS) places ads on the search engine results page
  2. Conextual advertising (ConAd) deals with placing ads on third-party web pages
  3. Social adverting (SA) is the newest type of advertising which places ads on personalised home pages on a users’ social network profile page

Google AdWords T-shirt

Regardless of what specific element computational advertising uses they all use specifically selected, personalised, thus relevant ads in order to:

  • Immediately initiate user interest, increasing the chance of clicks
  • Create a better user experience (UX)
  • Clicks and good UX maximises revenue for every search engine.

    Sponsored search, otherwise called PPC, paid or non-organic search, also constantly touches on these points and employs a variety of technologies, such as, information retrieval, machine learning, statistical modelling and, for example, microeconomics. All major engines use estimated click-through rates (CTRs) alongside bids to rank PPC ads. There are many ways in which CTRs can be calculated, the most popular include:

    • Hierarchical clustering of bid terms (bidded term/s are organised in clusters) are organised into ‘pools’ and freshly retrieved each time
    • Information derived from ad text
    • Query-ad word pair indicator features

    Engines frequently conduct thorough experiments for PPC ads (e.g. Support Vector Machines, Maximum Entropy, adaBoost Decision Trees). Vector space modelling is what Google, Bing and Yahoo all use to construct their index. Neither engine is a semantic search engine.

    3 secreted PPC IR elements uncovered

    Every ad is comprised of three elements: a title; description; and display URL. Engines analyse this information and assign a score, such as, Quality Score on Google, for example. Does get more clicks on mobile compared to, say, (a responsive site)? PPC teams should utilise and regularly test display URLs.

    1. Word and character overlap
    Bigrams are when two adjacent words or symbols appear more commonly than other variations. In the English language, for instance, the letters “th” are the most common bigram. Engines also use this data to improve PPC relevancy. Is it a case that the more common a word is, the more should be charged for that word?
    2. String edit distance
    Edit distance is precisely how spell check works on an engine: Edit distance simply compares two words and counts the minimum number of operations required to transform one string into the other. This is how engines can replace British and American spellings (e.g. capitalise vs capitalize).
    For SEO engines stem keywords on their index, [swim] triggers [swimming].
    3. Cosine similarity
    Since engines use vector space modelling, cosine similarity identifies how similar two documents are likely to be in terms of their subject matter. Does this document match the searchers’ keywords?

    Calculating Quality Score

    Every engine potentially can use the above elements to boost click-through rates: CTRs are boosted by improving ad quality via relevancy. Again reinforcing why Google rejects PPC ads. CTRs are a relevance measure; thus are used to rank PPC ads on all the major search engines. To do this:

    • Cosine similarity between query and title
    • Character overlap between query and abstract
    • Character overlap between query and (display) URL
    • Query length
    • A feature that counted the number of bigrams in the query that had the order of the words preserved in the title

    It is also worth noting that engines now place weight on UX. So your website needs to, at least, be fast loading, easy to use and scan, clearly and correctly labelled etc.

    UX and relevancy also examines what landing page the ad will land the searchers on. As a result, URL redirect is also a quality score signal.

    Clicks alone can be a weak indicator of relevance due to click fraud, accidental or exploratory clicks, and position bias. For this reason engines also take note of personal records or, histories if you like, to better understand your behaviour.

    Image: Copyright granted, reused, unmodified.


    Dave, K. and Varma, V. (2011) Computational Advertising: Leveraging User Interaction & Contextual Factors for Improved Ad Retrieval & Ranking. PD. D. Symposium. Hyderabad, India.

Why ‘links’ will die by 2017

Death of gaining natural links

Links will ultimately disappear because of the increasing popularity of mobile devices [1], coupled with our user behaviour this will kill links: We do not link out to other web sites on mobiles, as such, search engines are beginning to place more weight on other ranking factors apart from links and on-page signals. Google, for example, are improving their page speed test tool to include UX metrics. Matt Cutts has also said that link data was removed in an internal Google experiment which suggests Googlers are actively reviewing their algorithm in light of mobile user behaviour.

In March 2014 Yandex gradually started to remove link data for commercial queries claiming over 800 ranking factors are reviewed because user behaviour is a much clearer and natural signal.

3 links, what and why?

  • Internal links are connected with UX because they encourage page views; thus increasing engagement and viewing duration, for instance. An increase in mobile device use will not impact the number of internal links by 2017 because web developers and content management systems will continually be used on desktop devices.
  • Out-going, like internal, links are also followed by search engines but out-going link data is analysed to assist search engines to determine and improve relevance and authority. If, say, for instance, linked out to then this site’s authority will increase. Out-going links will not change in 2017.
  • Incoming links are going to be most affected by increased mobile use because the majority of your mobile users cannot link back to your site’s content. Google’s current best practice states that links should be earnt but by 2017 most users will be using mobile, and starting to increase wearable device use, with a technical inability to link out. For this reason alone search engines will have changed their ranking tactics because they are aware that the SEO industry will be the majority of active link builders therefore not neutral; thus unsuitable to use in an algorithm.
  • Always about mobile?

    An eHandwriting device, Telautograph, was patented in 1888 by Elisha Gray. The theory and discussion goes back much further than one initially thinks.

    Google’s new engine, Hummingbird, has the power to process social media data but due to it’s fast-paced, spam-like nature social media links will not soley replace current links. Social links could be used as quality indicators but not the only method. Is an article that get lots of mentions, favourites and likes a great article to rank?

    Can links be saved?

    It is possible that browsers could be redesigned to include an easy link building tool but since it would be a new feature getting existing users to regularly use it would prove difficult because our behaviour is difficult to change. Mobile device use, furthermore, is usually much shorter than desktop or laptop so time and ease of use would be key to such a tool.

    Search engines will simply move towards UX and write a comprehensive algorithm to rank web sites rather than relying heavily on link metrics. Will the new anchor text be social media messages in order to get contextualised keyword data? Is this why search engines are pushing their own social media platforms? How will search engines overcome bot traffic which can negatively increase bounce rates, negatively impacting UX, thus rankings, in 2017? There are still a lot of search engine developments to be made; in the meantime keep link building in 2014.

    [1] globally 3G smartphone has 25% penetration (2% for 4G) and some countries have high 3G penetration. Japan, Italy, USA and the UK, for instance, have 88%, 91%, 71% and 73%, respectively

5 quick tips to get better engagement on your site

Looking and interacting with web information

5 people in a usability test can identify 85% of a web site’s problems. Usability tests are worth investing time in because your web site’s usability directly affects the engagement. Here are 3 tips and tricks to increase online engagement:

  1. Test your site’s layout, categorisation and terminology regularly. Think of these components as being plants in your garden. They need to be pruned and cultivated to flourish. Any web functionality tests need to be constantly refined and refreshed.
  2. We skim webpages for cognitive cues, in fact, within 10 seconds we permanently know if a web site is useful or not. Aid skimming through your content’s headers, nicely laid out paragraphs, and by using the usability test data (above) to see if your information architecture (web arrangement of material, labels and presentation) needs to change.
  3. Offer social media icons to encourage content sharing. But think about where you place these on your site. We read in an F shape or golden triangle so placing icons on the top of left-hand side is a good idea. You can also use floating social media icons.
  4. Embrace unique content but also incorporate learning within your content to further increase engagement. If someone learns something new on your site they are more likely to recommend it to another person. Write clearly. Helping a user to learn something also makes them reflect on your site’s content and this can also drive returning visitors.
  5. Research engaging topics, such as, new applications, new technology developments or challenges of social networks, for instance.
Photo of a URL being clicked

URL with cursor

Posted by


  • Norman, D.A. and Nielsen, J. (2010) Gestural Interfaces: A Step Backward In Usability. Interactions. September – October 2010

Ranking. Mobile search. Point of Interest (POI)

Point of interests on mobile search

Whenever we search on mobile devices our behaviour changes. Our search sessions, for example, are much shorter and more conversational than desktop searches. Search engines need to find out what the user intent is by calculating explicit and implicit keywords. In a nutshell, if we do not clearly state a location, for instance, [restaurant] then this is an implicit query and search engines will focus more on the closest POI rather than ranking web sites or locations further away.

Mobile web search screenshot

Search engines calculate explicit and implicit intent then rank results

Researchers, two of which are from Microsoft, found that searchers often click on POI up to 3,000 meters away. This indicates that that we are not always interested in the nearest POI. Although search engines can calculate to some degree of accuracy explicit and implicit keywords they still do not know how to calculate user interest and to do so will come up with more privacy issues, see location itself below.

For local search to be more precise web search engines would have to use GPS, however, due to privacy issues this is not always possible. In fact the EU might target locational data after Google’s right to be forgotten dies down.

Local search engines, thus affecting the SEO industry, do not really account for popularity within points of interest. Yes, reviews help. But this is still not representative and can be problematic to take into account the oldest, traditional pizza house in Rome, for example.

POI and SEO: 5 top tips

  1. Build lots of citations in order to boost local SEO signals
  2. Make use of social media to utilise locational data. Social media explicitly ask for location-based data, for example, Twitter. Your network can be analysed in geographically and therefore it is possible to gain a list of potential local sources who may link out to you which can also strengthen locational data signals. You can further improve this be creating pearl growth by finding similar local follower and pushing them content they are likely to link out to — the local SEO link. Although we do not know how much weight this factor has on localised rankings
  3. Use schema to enhance signals and increase click-through rates
  4. Get positive reviews. More reviews. And, eventually, even more reviews. But be careful. If you get a sudden surge of reviews it can trigger a negative, penalty like, signal
  5. For any local SEO campaign research geographic keywords and get in tune with local dialect

It is clear local SEO is the future of search least not forgetting its popularity. In fact 78% of local mobile searches result in offline purchases. Go, rank, locally.

Posted by


Pay Per Click advertising via keywords, UK and EU law

Bidding legally on competitors’ keywords on PPC

Pay per click advertising effectively reaps instant rankings and helps enterprises take up more of the SERP. Together this increases brand awareness through keywords alone. SEO, however, still gets more clicks compared to PPC.

Legally keywords fall under trade mark law as they provide huge economic benefits. In fact, collectively, intellectual property is the biggest investment any company has because your brand can, in itself, sell your product(s) and/or service(s).

A photo of a Google themed hat saying Noogle


Digital trade marks

The list of digital assets is extensive. Web pages as a single digital asset includes various elements, for example, store location pages, category pages, product pages, reviews, holiday promotions, sitemaps and semantic make up. Simply by knowing your digital presence you should also know a list of your digital assets which you can monitor and therefore protect.

Trade mark infringment can occur on organic search and pay per click advertising. Web search keywords, more specifically branded terms, are your digital trade mark and these can be infringed, just like traditional trade marks, such as, a logo, for instance, if best practices are ignored.

In the early days of SEO it was possible to include keywords in the keyword HTML tag to improve your rankings, in fact, Bing and Yahoo still place a very small weight on such keywords, however, Google now ignores this tag and any weight placed by Bing and Yahoo is only going to come into play if another web site has the exact same on- and off-page ranking factors. Legally you are not allowed to include your competitors keywords to rank for their brand name in organic results because it is misleading to the searcher.

Pay Per Click advertising systems must be clearly labelled as “ads” and it is legal in the UK and EU to bid for your competitors keywords because PPC’s system distinguishes their results as paid adverts. For this reason PPC systems will always have to distinguish ads, in the form of labels or typography, for example, however this area is growing in case law, as outlined in 3 legal, summarative, case studies below:

European Pay Per Click advertising legal case study
In Germany a court ruling, May 2014, concluded that trade marks with reputation should be more difficult to bid for by using PPC. The German court heard that had dramitically reduced the prices of a rival brand, Beate Uthse, and advertised this through PPC. The Frankfurt court ruled that Beate Uthse’s trade mark may be percieved negatively to the average searcher and therefore proving reputational damage is key to taking such a PPC case to courts.
Amazon and an EU wide ban
Amazon was banned throughout Europe from infringing a UK cosmetic company, Lush, after bidding for Lushes’ products and stating they were on sale when in fact they were not on sale after redirecting search engine users to Amazon’s internal search engine and in Google’s sponsored search results. After an agreement could not be reached between Amazon and Lush the case went back to the High Court and a EU wide injunction was issued to overcome Amazon’s technical difficult counter-claim.
UK PPC competitor keyword bidding
A 2013 legal dispute between Marks and Spencers vs Interflora came to a firm conclusion under British law: Bidding for competitors’ PPC keywords is legal so long as the trade mark is not underminded by the PPC advert through searcher confusion over who is selling the item, in this case flowers.

Pay Per Click advertising, UK, EU law, and competitor keyword bidding

  1. Never mislead search engine users, for example, if you state on a PPC ad that product-a has 50% discount then that webpage must display productA as being half price on the referral web site.
  2. Avoid any confusion
  3. Put yourself in the situation. How would you feel? Is the PPC ad clear?

Bidding on competitors keyword using PPC is growing in case law and, therefore, must be continually monitored to find out the latest legal grounding.

Do you bid using PPC on competitors keywords? Tweet Gerald.

Posted by


What is a trade mark? How to register for one?

Trade mark, intellectual property

A trade mark, otherwise spelt trademark or abbreviated TM, is a graphical sign or mark which uniquely identifies one company from another. Trade marks remain popular, in fact, in 2012 there were 545,791 trade mark applications from residences across Europe (including 2,704 in Ireland and 36,596 in the United Kingdom) [1]. A consumer may build a connection, legally defined as under-taking, with a positive or negative experience: Trade marks allow economic growth because consumers can trust and purchase from a particular brand.

Photo of a trademark symbol

Trademark symbol

Trade mark application

IPO have self-explanatory steps to help enterprises search and apply for patents, visit their website to complete an application.

Classes categorise products and services and must be stated on an application. The classes you register your trade mark in are the only classes you can use your trade mark for. Research classes on IPO’s website (e.g. classes 9, 11, 34 and 42).

Avoiding a Europe-wide community trade mark (CTM)

Bainbridge, a Professor of Intellectual Property law, argues that a CTM looks interesting to a trader because one registration gives validly throughout each Member State within Europe. However, in reality an application will be refused if it conflicts against national registers. In other words, if an enterprise in Ireland has a trade mark for company-a which is also in the same class of goods or services, a CTM will be refused: A CTM, potentially, could be a waste of time and money.

You should consider applying for an application with the UK IPO and the Madrid Protocol, rather than a community trade mark.

How to register a trade mark in the UK?

You can complete and send an application to the national or regional trade mark office. This application will contain the sign or mark in full detail (including colour, if appropriate), in addition to, other features such as, illuminations, for example.

Trade mark rights
• Persons cannot take Comphylo’s sign and fix it to goods and/or packaging
• Competitors cannot use Comphylo trade mark for marketing purposes
• Persons cannot use a similar (or identical) sign to Comphylo’s
Trade Mark exclusions
• It only contains the shape of peripheral devices
• Using a person’s own name and address
• Use to indicate the intended purpose of a product or service, for example, spare parts
• Use of an unregistered mark which is protected by the law of passing off

Law of Passing Off

Passing off is a common law (which is based on previous legal cases) that protects the goodwill of enterprises, along with, protecting the public from being deceived. The law of passing off is mainly used whenever one enterprise has no registered trade mark.


Misrepresentation stops the public from being deceived (e.g. competitor-x using Comphylo’s logo or name to sell their software, including slogans and visual images).

Malicious Falsehood

Similar to defamation and closely related to passing off, malicious falsehood, otherwise known as trade libel, occurs whenever a person publishes information which could destroy another enterprise, or their good(s) or service(s).

Trade marks are very powerful pieces of intellectual property. My next blog post will analyse trade mark infringement in terms of pay-per-click (PPC) systems.

Posted by


  1. Bainbridge, D., I. (1990) Introduction to Information Technology Law. 6th edition. Edinburgh: Pearson Education Limited.
  2. Johnson, H. (1995) ‘Trade Marks: The New Law’, Managerial Law, 37(4/5) pp. 1 – 59

Stop. Google is not a semantic search engine

Is Google a semantic search engine?

A semantic search engine understands each query and presents relevant documents based on this understanding. Google however is not a semantic search engine but it does display elements of a semantic search engine:

  1. “Searches related to” is officially known as semantic similarity and it is designed to guide a searcher towards their search goal
  2. Google AdWords offers a keyword generation which is also closely related to semantic similarity
  3. Computing engineers also analysed Google’s search engine in 2011 and found that several results, or snippets, provide incorrect or non-useful semantic information — including queries who ranked in the top results. Google is therefore not a semantic search engine
Popular semantic search is a good bit away

Semantic technology is not ‘new’. Copyright granted by DullHunk.

Examples of semantic search engines

Sensebot and Sindice are examples of semantic search engines. Here understanding and relationships of keywords are key to producing results. That is not to say that other search engines do not understand queries but they rely on other technologies to do so rather than take a mathematical approach and rank queries based on meaning as supposed to, for example, saving and placing weight on keywords that appear in the title and header (term location in information retrieval). Search engines have just started to get smarter, thus a little semantic, with search, for instance, Google’s Knowledge Graph. But on the whole Google and other search engines are not semantic search engines; they just display elements of semantic search rather than be a complete and whole semantic search engine.

It is impossible to ask [photos of a flower with 5 petals and a yellow vase in the background] so Google is not a semantic search engine because it does not, unlike humans, understand a string of keywords to display relevant results.

Posted by


  • Xu, Z. Luo, X. Yu, J. and Xu, W. (2011) Measuring semantic similarity between words by removing noise and redundancy in web snippets. Concurrency and computation: Practice and experience. 23(2011) pp. 2496–2510