My name is Gregorio Pellegrino and I’m an Italian computer engineer. Since elementary school, I was fascinated by computers and bots, so in my free time I learnt to code. I’m in a family business: my parents established Effatà Editrice in 1995, a small publishing house based in the North-West of Italy, near Turin; Effatà publishes religious non-fiction books and the main themes are family, education, philosophy and theology. The catalogue counts more than 500 titles plus other 500 titles that are now out of commerce. In 2010 it was created Effatà Tour as spin-off of the publishing house, a tour operator that organizes travels with our authors all around the world.
At the end of 2014 the publishing house was in a precarious economic situation. After a detailed analysis based on historical data, I found that one of the main problems was the high cost of unsold books: in Italy the bookstores can return to the publisher the copies that remain unsold, and in some cases the copies returned to Effatà were the 90%, which raised the costs higher and higher. I also found that it was difficult for the publishing house to act like a firm: a small publisher is a craftsman that falls in love with the books he publishes, so it is hard to be critic when you have to choose how many copies have to be printed for the first print run.
At that time I was reading the e-book Automate This: How Algorithms Took Over Our Markets, Our Jobs, and the World by Christopher Steiner, a great book which taught me about extraordinary algorithms that can manage very complex problems (from health care to stock exchange, from writing music to play chess), so I thought “why can’t an algorithm manage our publishing house?”.
With that in mind I started to develop Lello a big database with the idea of collecting all the key data for the publishing business (for example invoices, stock levels, books metadata, distributor orders, etc.). In italian Lello is the abbreviation for the word cervello that means brain; Lello would have been the big brain behind Effatà Editrice (in my mind something like the IBM’s Watson).
Beside Lello (that is based on the cloud platform Zoho Reports) I coded some scripts (in the programming language Python) which periodically run some tasks in order to improve the efficiency of the business management. I love to think at them as bots that help us in our daily work, so I gave them a name:
- Arturo is in charge of the distributors management: he daily checks the stock levels, the refill orders and decides when Effatà has to send books to the distributors’ warehouse checking that the courier’s costs are low, compared to the books’ price;
- Chris checks twice a day if all the metadata are correct and consistent across all the supply chain;
- Maurizio manages the synchronization with the ERP (Enterprise resource planning) and automatically creates the receipts for the e-commerce orders.
These are three of our assistants, others are coming. For example I’m working on Cecilia, who will be the overall supervisor.
As you can see the presentation below, in Lello the circles represent the various entities that interact and on the green lines you can see the data format used for integration.
One of the key tasks of Lello is the link with the ONIXsuite platform, a metadata manager that exports all the data in the ONIX for Books standard. Lello uses ONIXsuite to keep organized all the books published by Effatà (more than 1.000 titles). Having all the data standardized is of great help in relation to some duties shared among many colleagues, as, for example, the preparation of the catalogue containing the latest titles to be presented at the book fairs: if before ONIXsuite it would have taken one week to prepare it, now we can create it automatically in two working days (using XSLT scripts and InDesign). Needless to say, this system allows to save a lot of time that we can spend for making better books.
Another result of the implementation of Lello is the possibility to be clear with our authors when sending them the yearly royalty statements: thanks to this infrastructure, now we are able to build some reports (using Zoho report) that are daily and automatically updated with the ERP and distributors data, so that authors can check online whenever they want to know how many copies of their book have been printed and how many have been sent to the bookstores.
The Lello platform and the bots are helping us to reduce expenses (we reduced useless print runs, just to mention one) and take control over unnecessary costs. This re-organization increased our profits by 45%, thanks to a centralized data management system allowing to cross-check the data (ERP, distributors data and books metadata) in order to find inefficient costs.
Building the Lello ecosystem would not have been so fast if we didn’t base it on standards. For example, our e-commerce is able to import ONIX for Books 3.0 feeds from two different sources (ONIXsuite and digital distributor) using the same task; I used simple dictionaries to translate the ONIX values into proprietary values used by other partners (Amazon, for instance), so all the data can be processed automatically from the partner without any human intervention; last but not least, I implemented Schema.org on my website – it took me just one afternoon: all the data were just there, only to be tagged!
Nevertheless the development of Lello has not always been simple and we found some criticalities in implementing the new infrastructure: no one of our partners uses ONIX for Books 3.0 (and many of them neither ONIX 2.0); the available and documented APIs (Application Programming Interfaces to communicate between different systems) are very few and, more important, everyone fears computers that take decisions. It is essential to remark that our bots take secondary decisions automatically (like how many copies are to be sent to our distributor), but important decisions (regarding money and business administration) have to be confirmed by humans. Most of the people I presented our platform to say it’s too much, in fact they think “bots don’t know anything about books”. But my answer is: are you really sure?
What’s next? The high cost of unsold books remains the biggest problem for my publishing house, therefore I’m working on a statistical algorithm able to make sales forecast, in particular for new titles, so that we can reduce at minimum the print runs. Stay tuned!