Grapeshot technology can save you costs in manual tagging and editorial production processes whilst liberating ad sales and commercial teams to optimise each and every page view.

The Grapeshot Targeting platform is built upon a core Grapeshot sub-system which is compact, light on memory use and very fast. Grapeshot customers have recorded over 400 transactions per second on one machine, providing millisecond responses in the average range of 10ms. Online advertising and real-time categorisation need the fast speeds whichGrapeshot delivers, to optimise editorial and advertising selections upon each and every page view.

Grapeshot extracts the most significant words on the page - not just frequent or popular ones. Special algorithms first developed at Cambridge University, and now incorporated in production grade software, calculate the best words to select and use.

Probability algorithms help to rank and score the words on each page, and these word weights are used to understand the relative importance of words as we match them against category or channel definitions. The word is not a pattern, or a binary "yes there" or "not there" - for Grapeshot it is a mathematical variable in a complex set of probability algorithms, where Grapeshot assesses the relative weights of significance between words, and uses all of them in aggregate to execute an implicit search to find related content or advertising.

Most of the internet is populated with web pages that get categorised when they are first published. Journalists insert the article in one section of the website, rather than another. So whilst most Business content goes in the Business section, and Sports in Sports, it is still easy to find Business stories in News, or Sports stories inside Lifestyle. The articles stay inside these rigid buckets or sections of content. Alternatively someone comes along later to manually categorise the article with some new tag or category invented after the original categorisation was done.

Grapeshot breaks the mould. It applies categories when the page is viewed by the user, not when the page is first editorially created or crawled. The advantage is new categories can be assigned to old documents. Categories are assigned mathematically, so rather than "yes this category" or "no that category" - Grapeshot also provides a relative score of confidence for channel category assignment.

Applying weights to words and confidence scores to categories means Grapeshot can offer channel category controls to the ad sales and commercial teams. Weights of the targeting words can be artificially boosted, or certain channel category confidence scores manipulated, to ensure the appropriate volume of channel categories are sent through to the ad server.

Grapeshot can use Cookie IDs to track how users read content. But Grapeshot does not store the URLs a user visits. Instead is just chooses a few summary words from the articles a Cookie ID user reads, and stores this as a "keyword shadow" of the best few words that describe the user's reading habits on that one publisher's site.

Because Grapeshot works at the keyword level and assigns categories in real-time based on those keywords that represent the essence of the page, Grapeshot does not need a wide pool of user internet behaviour to second-guess a category for a user. So no Cookie ID is tracked between many different publishers' websites, to build a macro picture. Grapeshot does not need to do this - so has no part in creating a behavioural network of users across many websites to behaviourally target them. Grapeshot simply sticks to contextual matching of keywords within the one publisher's domain, not beyond it.

Grapeshot works in partnership with the publisher and keeps his audience data proprietary to him. Grapeshot works in line with the Publisher's own Terms and Conditions offered to his audience. Grapeshot does not mix data collected from one publisher installation with another publisher.

Each publisher's data is kept private to him - and the user keywords are used to help engage the user with more related content, or more relevant advertising based on information gleaned only by that one publisher.

Grapeshot is multi-lingual. It can evaluate the value of words in a page based on mathematics, so does not need a dictionary at hand to evaluate the relative value of words. Nor does it need semantic rules.

It means Grapeshot has a standing start in any language. So whether it is Chinese, Arabic, Spanish, Russian or French - the effort to extract the essence of the page, build user profiles based on reading habits, and categorise content to Premium or Unique Channels is simply the same as working in English. Automatic and ready to deploy. Today..

Grapeshot is executing search in milliseconds to find the most relevant content or advertising for a page view, in real-time. The search algorithms are using keywords to evaluate the match, and using many keywords together, not just one or two.

Grapeshot can therefore provide transparent evidence as to why an advertisement appeared on a particular page view. Either the page, or the user profile, has a few keywords that overlap with the ad channel keyword description. The keywords are the evidence for the match, so it is simple and easy to show which are the keywords creating the best matches on each and every page view.

Mining the keywords is important, as advertisers can be given reports on which keywords drove the highest click throughs or brand interactivity. Grapeshot can provide for display advertising and recommended editorial links the same keyword precision that advertisers have learnt to use with Google Adwords.

Grapeshot engineers have a rich experience of delivering commercial search solutions since 1992. Our Premium and Unique Channel products are easy to install and manage. So easy that some customers have gone live without ever meeting our staff. However Editorial Infrastructure projects often involve special RSS feeds and integrations with Content Management Systems (CMS) such that special set-up projects are commissioned prior to launch.

Our Cambridge engineering team can provide bespoke engineering services as well as a 24/7 support service for the web services we either host on behalf of customers, or they host within their own server farms, but for which we continue to provide second level software support on 24/7 live services.