John Valentine

About TGI

TGI automates the graphic design of over 5000 base product specifications and their many optional extras, so that graphic designers spend less time setting up the technical aspects of their design templates, to instead concentrate on value-adding creative work.

It is estimated that over the life of TGI, over 400,000 hours have been saved. In the climate of a growing company, this allowed designers to be more effective with their time, and process more orders.

In addition, TGI had a significantly positive impact on quality assurance, because graphic designers made fewer technical errors, leading to greater customer satisfaction and a few percentage points of improvement on gross margin, due to greater efficiencies and less re-working.

Further, the skill threshold for graphic designers was lowered, and more designers could tackle the more complicated products.

Developing TGI

After working with graphic designers within the company for a few years, and having oversight of the production workflow and responsiblity for design specifications, I saw an opportunity to reduce the time that graphic designers spent setting up their work.

The TGI project began with a proposal that would make a small improvement: when given a product code, it would create a document of the required size. I continued making improvements, provided the benefits continued to be delivered. TGI was an agile project from the beginning: the MVP was very minimal. With time, more detail was added to the templates, and more help was provided to the designers using TGI.

I analysed several aspects of the company's production turnover: which products attracted the most revenue, which products had the poorest production quality, and which products seemed to make designers unhappy or were technically demanding. This helped me prioritise the order in which I took products under the TGI umbrella. Each product implementation was version-managed and its support status tracked in the TGI database.

After successful piloting in two of our design studios, the other studios wanted TGI, and it was rolled out company-wide. Updates continued to be issued at least weekly.

After a few basic products, it became clear that the information system did not have the data I wanted. Many of the specifications were expressed to designers and production as written instructions. To work in an automated way, all those instructions needed to be encoded in a quantified machine-readable format, so as a side-project to TGI, I designed ways that these options could be expressed in data, and the database was extended and populated. This also helped customers to more precisely specify the options on the order-generating public websites, and was the final ingredient required to allow the company to take payment online without speaking to a representative, and with minimal handling thereafter. There were analytical benefits, because the options were enumerated and quantified: financial services and product owners could see which options were popular and generating revenue, quality problems could be narrowed down very tightly, and environmental impact reduced.

TGI also assisted with improvements to the production processes. For example, if guide markings were needed for particular products, they were encoded into TGI for all new designs, so that the whole studio network was soon creating output that matched the improved specifications.

I developed another side project, CloudSeed, which provided in-app pre-flight validation and correction of print-ready designs. It understood the products, pointed out errors, and offered automatic corrections. This helped all users to quickly implement process improvements, in line with the process changes pursuant to ISO-9001 and ISO-14000.

At its peak, TGI was used by over 400 studios, with full localization in many countries (UK, Europe, US, AU/NZ, ...), and outputs for their variant manufacturing processes. It enabled designers to start a design template for 99.8% of the printed revenue of the company, and included an interface to generate custom (off-database) specifications by composing rules and parameters.

It was a great success, and 12 years after inception, it is still in widespread use today.