Category Archives: Tech

Roadmap, Dec 2012 to Feb 2013

We’re going to be focusing on:

  • Marketing and PR: Promote the Young Writers Anthology into the new school year, secure ongoing media coverage from active PR (several leads already in place with major magazines), and run a few narrowly targeted promotion campaigns in partnership with copy shops and publishers
  • Software: Complete phase-2 development of, including important improvements to UI based on user feedback and refined messaging, and more document automation to speed up the addition of new content
  • Revenue: A risky but important move coming in Jan: we’re going to start charging outlets a small service fee for most our content that used to be free. This will be to test potential self-sustainability revenue models and measure outlet expectations.
  • Outlet feedback loops: A key focus we’ve neglected is active outlet surveys, triaging and acting on the feedback. These surveys and action in response will be a major focus for the next three months.
  • Partner projects: We’ll publish a beta version of our Guide for UNISA students in partnership with Together We Pass; make the most of a promotional partnership with national reading campaign Nal’ibali.

Removing ad space on Paperight documents

Our latest website update now allows the option for watermarking documents with ads or without ads. This was our crucial first step towards product improvement. We found over this first year that outlet owners and managers were not inclined to subsidize the cost of paper by selling advertising on their documents.  As a result, we decided to remove the ad space we’d left at the bottom of every page.

Because many of our PDFs have been laid up to account for the ad space, this meant that the watermarking functionality had to account for documents either with or without ad space.  It also meant that the 900+ documents that were on Paperight would eventually need to be reformatted to remove ad space. We’d have to lay up the documents from scratch, with the text now able to take up more of the page, which improved readability. 

By the end of September a total of 1081 titles were listed on Paperight, officially surpassing the amount of titles we had on Paperight 0.5.

We began by reformatting matric exams, our most popular set of products.  We uploaded more Paperight Editions, including Shakespeare titles, now using the new format. By the end of September a  total of 1081 titles were listed on Paperight, officially surpassing the amount of titles we had on Paperight 0.5.

We did our first round of Rightsholder statements for sales on the new site. These had to be created manually, and emailed to each publisher.

I also conducted the first real set of enmasse follow ups to publishers that I’d contacted and not heard back from. And the team was joined by Oscar, who started his internship at Paperight with metadata related tasks. His main focus is on reviewing existing metadata, and improving it where possible.

The Shuttleworth foundation approved a project pitch to send me to the Frankfurt Book Fair, so I began preparations. This included setting up of meetings, research, and pitching preparation.

Publisher registrations

  • T&P Books (1/9/2012)
  • Disruptive Publishing (5/9/2012)
  • GetSmarter (11/9/2012)
  • A.R.B (12/9/2012)

Our roadmap, last quarter of 2012

As the Paperight team, our aim is to recycle some of the marketing tactics that seem to have worked (among many that didn’t) over the last six months. Mostly, this means focus: helping very specific outlets sell very specific books to their target markets.

We’ve engaged a PR company to help us plan and execute a promotional strategy for six months. So, assuming we like what they propose, planning with them will be a key highlight.

I’ll lead the development of new features on the website that will greatly increase our speed and greatly improve aspects of our service over the coming months.

Setting up the help site and content leads list

We began the month with a quality control of the African Books Collective documents to double check them before uploading, archived the files, and uploaded them via CSV bulk upload. By the end of July we had 829 titles listed on Paperight.

By the end of July we had 829 titles listed on Paperight.

Ra’eesa (in the pic here) and Diann finished their time at Paperight by working on a list of content leads, and setting up the help site, respectively. I created a spreadsheet for Ra’eesa and began by sending her the emails and notes we had been keeping about potential content leads, and she spent time capturing all of these as part of maintaining an ongoing database (this is a task that Oscar would pick up later, when he joined us). Arthur set up the Paperight Help site, and Diann used the manual Nick and I wrote earlier in the year to create posts to assist outlets in using the Paperight service.

Having a help site was the preferable option as it allowed us to link to posts on specific issues in an email, rather than having to explain each time, or expect busy outlet managers to read through an entire manual to find the solution to the one problem they were having.

We’d learnt that outlets were not downloading and printing out the manual for reference, as we had expected them to. Instead, they would call or email us each time they had a query. Having a help site was the preferable option as it allowed us to link to posts on specific issues in an email, rather than having to explain each time, or expect busy outlet managers to read through an entire manual to find the solution to the one problem they were having. The help site itself had step-by-step instructions, and screenshots, on things like registering as a Paperight outlet, buying a license, or boosting your business with Paperight.

We also spent time prepping Communist University modules, OUP study guides, and Paperight Editions. Caitlin split some exam packs as per Silulo’s request. We tested the splitting of exam packs with seven subjects: Accounting, Life Sciences, Maths, Maths Lit, Physical Science, Business Studies, and History.

I wrote and submitted a research concept note for a joint EBW and Paperight project, but we were not selected for funding.

Publishers registered

  • Burnet Media (13/7/2012)
  • Kebooks (16/7/2012)
  • ZIM-BUKS (21/7/2012)
  • BookBox Inc (23/7/2012)
  • MSED (25/7/2012)

Project 1: core team and beta site: closing report

This project was to hire a core team, and build and launch a minimum-featured beta site with automated content licensing and acquisition.

General report-back

There were two major parts to this project:

  • get 1000 publications on, and
  • build a basic working site to replace the working demo site, adding instant document delivery.

We achieved the 1000-publications target on time (see the report on our blog for details). Reaching this target required research of product leads, sourcing of documents, compilation of metadata, and the listing of each product on, in addition to the prepping of documents for sale via the site. Our content manager Tarryn did a marvellous job of planning and driving this process. We brought in an intern, Michal Blazsczyk, who helped a lot with the legwork of getting this done.

We did not hire a software developer. During the recruitment process it became clear that we would build a better product faster if we contracted a software company rather than an in-house developer. So Arthur spoke to several software companies, discussing the project in detail with them. These were narrowed down to two companies – Realm Digital and Double Eye – who had built large book-selling sites for major clients before, and had a range of applicable in-house experience with the book industry and with automating PDF manipulation.

This process took three months. Ultimately, Realm Digital were chosen. Based on extensive interviews and discussions with contacts who’d worked with them, they offered better front-end design expertise and a more impressive track record.

We completed a functional specification for with them, based on which they drew up a full development costing. The costing was based on the fact that all code would belong to Paperight or be GPL-compatible. This meant Realm could not use proprietary tools they normally relied on, which raised development costs well beyond our project budget.

Depsite the high cost of developing with Realm, the process of finding them and developing the functional spec – which showed we had a good team understanding with them – convinced us that they were still the best way to get a slick, solid product built quickly. So we decided to close this project and apply for the software-development budget in a separate pitch document.

This means a large part of the funds in this project were not used, and can be released back into the pool.

Objectives achieved

1000 publications available for sale on

An important first stage of software development was achieved: a thorough functional specification for Paperight 1.0 was created, along with UI designs, briefing documents and UX flow diagrams.

Objectives not achieved

We did not build software for instant PDF delivery.

To keep us going without this, we refined the existing site so that we could trade with it, and designed back-office workflows that allow us to manually deliver print-ready PDFs to clients within 24 hours.

Measures of success

Before: “We expect immediately to see our pilot-testing outlets respond positively to improvements in speed and ease of using the beta site over the prototype (WordPress-based) site”

After: We do not have the new site to do this comparison. This measure has been included in our Project 3 pitch.

Before: “we’d like to see them using the site on their own for buying content for customers, for content we haven’t explicitly told them about, and visiting more often per month than they visited the prototype site.”

After: As above, we cannot make a comparison yet, since we’re still using the original site. We did have two unassisted licence requests from unsolicited outlets. This small sample showed that the existing site can be used to run Paperight while we build instant document delivery in Project 3. The site visits by registered outlets cannot be measured with the current site, and we have not yet had repeat sales from those outlets that have bought licences.

Before: “we’d love to see more than a dozen new outlets registering and using the site based on word of mouth, without our active promotion and guidance.”

After: Running the original site, we have had 15 outlet registrations, of which:

  • 11 were unsolicited
  • 6 are consumer-facing copy/print businesses
  • 4 are school teachers
  • 5 are private individuals (publishing/education professionals and parents of school-going children)


Original budget: R257000

Actual spend: R92500

Returned to pool: R164500

Item Budget Actual Return to pool Comments
Senior developer 160000 20520 139480 Spent on functional spec for Paperight 1.0
Content manager 48000 54720 -6720 VAT incurred since this was paid through EBW, who are VAT registered
Office space 24000 5760 Worked from Arthur’s home. Budget shifted to interns.
Interns 18240 Used office space budget to pay interns R4000 × 4 months: Michal Blazsczyk (Jan–Feb) and Nick Mulgrew (Mar–Apr)
Register Pty (Ltd) 6000 681 5319 Used a cheaper registration agency
Computers for 2 15000 7299 7701 Computer for Tarryn, but no developer
Software 0 Provided by EBW as planned
Printing costs at testing sites 4000 0 4000 Not used, since we didn’t have the new site to test.
TOTAL 257000 92500 101 460 155540


Outputs and deliverables

IP Author Owner
Metadata for and online listing of 1000 publications Paperight team (Arthur Attwell, Tarryn-Anne Anderson, Michal Blazsczyk, Nick Mulgrew) Paperight
Prepared PDFs for same 1000 publications Varies by document Content is public domain or content-specific rightsholders, packaged presentation owned by Paperight
Site design for Paperight 0.5 (WordPress) Original design by Elegant Themes Inc, heavily modified by Arthur Attwell Original design Elegant Themes Inc (licenced under GPLv2), modifications Arthur Attwell
Site code for Paperight 0.5 (WordPress) Auttomatic and others, modified by Arthur Attwell Various, licensed under GPLv2
UI, wireframes and briefing documents for Paperight 1.0 Arthur Attwell Arthur Attwell
Functional specification for Paperight 1.0 Realm Digital Arthur Attwell (in IP agreement between Realm Digital and Shuttleworth Foundation)
Cover thumbnail designs (220 as of 13 Mar 2012) Paperight team (Arthur Attwell, Tarryn-Anne Anderson, Michal Blazsczyk, Nick Mulgrew) Paperight
Workflow/process documents (hosted on the Paperight wiki) Paperight team (Arthur Attwell, Tarryn-Anne Anderson, Michal Blazsczyk, Nick Mulgrew) Paperight
Blog posts, including content report on first-1000-items Paperight team (Arthur Attwell, Tarryn-Anne Anderson, Nick Mulgrew) Paperight



Regarding content, the surprising thing was how smoothly it went. This was largely due to Tarryn’s dogged determination and meticulous planning and execution. It’s also unlikely we would have made the deadline if it hadn’t been for intern Michal Blazsczyk. Michal approached Arthur out of the blue for an internship, and this was fortuitous. We’re glad that we had the opportunity to bring him on: in an industry where skills are rare, it’s important to be able to bring in talent when you see it.

Much of our content-related learning was operational: we learned a lot about the mixed quality of open sources of content, PDF technology, workflow best practice, and getting new team members up to speed fast using wiki-based ops manuals.

Regarding software, we learned from research early that hiring a software developer in-house is a huge risk: their productivity is impossible to predict, and all your eggs are in that one basket. Then we learned how long it takes to be sure you’ve found the right software development company. If we’d known it would take so long, Arthur would have pushed harder earlier, but the process is underpinned by a myth that ‘tomorrow we can start writing code’.

We also learned that the cost of building open software can be significantly higher than building software on proprietary platforms, where the development industry has come to rely on proprietary platforms for fast rollout and licence-based annuity income.

On the side of this project, we learned that – with the exception of a handful of small businesses – it is very difficult to get commercial publishers to agree to provide content until the business is trading with a proven footprint. This led us to focus on a great open-licensed and public-domain-based catalogue.


This project was the first step in a series of projects leading to overall sustainability for Paperight.


We definitely achieved our content objectives and we’re extremely pleased about that. We now have a great starting catalogue with which to promote Paperight (see project pitch 2). We also have a growing list of sources of further open content.

We didn’t get far enough with software development in the time we planned. While we’re very happy with the foundation we’ve laid (great suppliers, UI plans and functional spec), we didn’t get a chance to achieve and test some of our key software-related objectives. Those have now been transferred and refined in a new project (pitch 3).

Next steps

The next steps are to build the Paperight 1.0 software planned and specced in this project, and then to promote Paperight to outlets. Project pitch 3 continues the software work. Project pitch 2 is all about promotion, while continuing to build and refine the content catalogue.


New stuff we’ve made

On 9 May 2012 we saw Paperight 1.0 go live, a site I began designing almost two years ago. I’m incredibly proud of the way it turned out. See it live at (It changed a fair bit over time, here’s a video showing an early interface design.) We wrote up a technical explanation of how we built it from open-source tools on the Paperight blog.

For content, our team created over 150 packs of past matric exam papers. It was a mammoth undertaking that now represents the single biggest library of exam packs in the country, that (depending on your local copy shop) is the best-value way to buy them anywhere. Team member Nick Mulgrew wrote up the arduous story on our blog. The team also produced our impressive Outlet User Manual (PDF download), which we’re now converting into a support website.

I also crafted (and in a sense this took years) the Paperight story that will be my template ten-minute pitch presentation for Paperight. It’s had a dry run or two at small events, and gets it real debut at TEDxCapeTown on 21 July. It’s amazing to see a big, complicated idea finally distilled to a simple story.

Less glamorous, I made an office and a team (here are the outlet guys) and a long list of lessons learned in the last nine months.

Our roadmap, third quarter 2012

Our aims for the next three months:

  • Continue and learn from our promotional campaign in Khayelitsha working with Silulo copy shops.
  • Build our outlet footprint in Gauteng, KZN, and Eastern Cape (we’re working on collaboration with the three major copy-shop chains in South Africa).
  • Another phase of software development, primarily to automate document preparation and boost exponentially the speed with which we can add content to the site.
  • Build media interest in and coverage of Paperight through articles, interviews and speaking events.

Getting to 1.0

screenshot_20120510Getting Paperight 1.0 live was a huge milestone for us. While the build in the end – from scratch, without a base CMS, over only two months – was fast and smooth, this was only possible because we put every ounce of learning and planning I’d done for the site over the previous three years into its design and planning. It’s a very simple thing now, but that is deliberate simplicity.

I had one key principle for the site: all our choices had to favour making the site as fast and light as possible.

I had one key principle for the site: all our choices had to favour making the site as fast and light as possible. Outlets have slow connections and busy shop floors. This informed decisions like our simple prepaid accounts (the outlet manager can top up an account in advance very quickly); account balances and licence fees had to be shown in the user’s currency, no matter where they are in the world; the interface had to be clean with large figures and buttons, for quick, simple use at the point-of-sale.

I worked very closely with the team at Realm Digital on this. Realm were an expensive choice, and I’m pleased that our decision to use them paid off in fantastic service, great design, and real dedication to the time-frames and to the business model itself. Lead developer Shaine Gordon has described technically how the site’s built on our blog.

In addition to ongoing refinements, we’re now planning a second round of development for Paperight 1.1.

Paperight 1.0 & our first large influx of content

May saw us moving into Paperight HQ, and going live with Paperight 1.0. The from-scratch site developed by RealmDigital included a simplified purchase process, account-top-up mechanism, outlet dashboard, and instant PDF delivery. This last meant that we no longer had to export and mail PDFs for each order as it came in, but the delivery function required that we prepare and upload documents in advance of sale. We could no longer list 1000 documents and process them only as orders came, we had to ensure that all the items listed on the site were processed in advanced.

We began by uploading the 88 documents that we had already processed (52 of which were the featured products on our poster). New uploading procedures (via CSV) resulted in some initial teething issues, and workflow re-design. We began maintaining an uploaded content list, in Dropbox, using file naming conventions for version control.

With the launch of the new site came other administrative updates. I re-registered all of the publishers/rightsholders. Arthur migrated the blog and wiki sites to new addresses, and we had to update all in-text links within the wiki. We began testing the new site’s functionality, and logging bugs to be fixed.

Nick and I wrote a full training manual for the site, which was uploaded, and sent to outlets. This manual would later form the basis of the Paperight help site.

Apart from the new site-related work, we continued processing matric exam packs, writing CSV for these, and uploading them to the site. By the end of May, we had officially uploaded all of the non-language exam packs. Caitlin started working with us on a freelance basis, and assisted this greatly.

The African Books Collective sent us 300 titles from their aggregated publishers. It was the single largest submission of content that we had received thus far – a very exciting step forward. We also flirted briefly with getting comics on to Paperight, in the hopes that high school students would be interested in this kind of content. We  got Project H (our first graphic novel), but other options fell through and we didn’t follow up on them actively as we had since moved on to other ideas.

Publishers registered

  • Marita Westenraad (7/5/2012)
  • Alta Schwenk (7/5/2012)
  • African Books Collective (7/5/2012)
  • Customcut signs (10/5/2012)
  • Communist University (10/5/2012)
  • Modjaji Books (11/5/2012)
  • Wonjoolaai Studios (15/5/2012)
  • Story Time (17/5/2012)

Highlights from the first nine months

Every three months, I sum up what I’ve been doing during my Shuttleworth Foundation Fellowship. At the time of writing this, I’m also reapplying for another year as a Fellow. In particular, I wanted to talk about lessons I’ve learned, what we’ve built over the last nine months, and where we’re headed. This post is the brief summary. In related posts I go into more detail about:

  • lessons learned
  • speaking events
  • website development
  • team and infrastructure
  • partnerships
  • a personal take.

screenshot_20120510After nine months, we’ve reached some big milestones for Paperight. Most importantly, the instant-delivery rights marketplace we set out to build is a reality, now that the Paperight 1.0 site is live. We have over 50 outlets registered – including copy shops, schools and NGOs – and have made our first revenue.

Innovative publishing companies have joined us, including Cover2Cover (youth fiction), Modjaji Books (acclaimed fiction and biography), the Health and Medical Publishing Group (publishers of the South African Medical Journal and a dozen others the and SA Medicines Formulary), and the African Books Collective, a renowned agency representing over 140 small and medium publishers from around Africa. (There are also several very small publishers signed up.) We are in advanced talks with several large educational and trade publishers in South Africa, too.

Promotional partnerships with Silulo Ulutho Technologies, a fast-growing chain of Internet-cafe-copy-shops, and copier-printer companies Canon and ITEC are kicking in now, too.

We’ve also made some shifts in our promotional strategy as we’ve learned through trial and error where time, energy and money are best spent.

So, listing the key milestones:

Lessons learned:

  • The human story is more powerful than the financial one (even when both are good).
  • Spending lots of time chasing big publishers isn’t worth it. There are many smaller, more interesting fish.
  • We needed to focus more on how we make it legal to print books.
  • We’ve learned to blend patience and impatience in software development.
  • Too much choice for our customers is paralysing. Simplify the offering.
  • Making our own content is hard work, but very important.

Speaking event highlights:

Other highlights were hiring our outlet team, co-branding promotional material with ITEC Innovate, a forward-thinking local copier company, and spending time with Zakes Ncwanya, who is moving back to his rural hometown to set up an Internet Cafe and Paperight outlet. (This later became a story in the Mail & Guardian written by our communications manager.)

A personal note: The Fellowship is not just a great way to build and nurture valuable projects. It’s a personal- and professional-development drag race that produces tougher, smarter, more effective people. It’s also addictive.