Lilly Marketing Studios is an interactive proccess guide and resource portal that allowed Eli Lilly to significantly reduce intenral marketing budgets. Over 26 different job functions at Lilly utilize the tool daily to get marketing work done.
My previous employer, Bottom-Line Performance (BLP), used to do a lot of work with Eli Lilly but had gotten cut from their vendor roster before I joined the organization. Losing Lilly work was a big deal for a small company... when I was hired there was somewhere near 15-17 people in the tight-nit, family-oriented organization. After a year or so being there, the sting of losing Lilly, and the financial crisis, hit our company hard.
The owners had to make cuts to keep the company going. Fortunately, in my brief tenure there, I had proven myself and kept my job.
Over the years we slowly gained new clients, increased our foothold, and built our way back up to nearly 40 employees... but losing Lilly back in the day was a deep wound in BLP's story.
Nearly a decade later, we would get a shot at earning back the trust and business of this pharmaceutical giant.
This project was a doozie.
Part of the only reason we got the work, is because it was an insane timeline to produce a lot of training, and we wanted Lilly back so badly we said we would do it. Lilly was in the process of internalizing a lot of their marketing efforts – a complete re-hashing of their entire marketing workflow and how they worked with marketing vendors to improve consistency, efficacy, and the cost associated with marketing by brining most of the work in house.
This meant new roles, changing roles, new processes for completing work, new best practices… a completely new way of doing work for someone in a marketing-adjacent role within the organization.
BLP was slated to produce 2.5 hours of live instructor-led training, nearly 100 job-aid resource guides, 8 to 10 eLearning modules, and (this is where I come in) a portal website that would help everyone learn the new process and provide just-in-time resources on the job.
By the time the project was procured, kicked off, and enough work had started for the web portal portion of the project to begin, it was nearly October. We needed to launch the site, live to users at the beginning of December.
We had about 9 weeks (including the Thanksgiving holiday) to get this done…launching early in the week after Thanksgiving.
As an aside, I was also in my first semester of my MBA, taking what I would know now as the most time-consuming course thus far in that journey.
As another aside, I had been approved to give my first-ever conference presentation early in November at the ATD-CIC tech summit.
This also, for portions, was not the only project I was working on, and at this time I held the position of “Manager of Multimedia Development”, which meant there were always little projects and/or fires I was tending to.
The largest component we were building in the application would be an “interactive process map”. This would be the entire new marketing process, as a flow diagram, that would need to be interactive to provide resources to marketers.
Challenge #1- final processes (the flow) had not been finalized yet and would be changing throughout the project. How
Challenge #2 – based on the type of marketing tactics being produced, there were 8 different processes (flow diagrams) that needed to be interactive, based on the type of marketing.
Challenge #3 – We were unsure exactly how many, and what resources would need to be included, and linked to every individual process step.
Challenge #4 – Roles and titles were changing live and these roles/titles would need to be directly associated with every step in a process map, and every resource in the application.
These were not all of the challenges, there were way more – such as the client wanting us to work out of their offices 3 days a week – but those were some of the biggest technical hurdles to overcome.
In the beginning of the project, before we were ready to ramp up development of the portal, I was very heavy with other work so testing “exact” solutions to the business/development problems was difficult to do.
Luckily, I have experience with Adobe Illustrator and knew that you could export SVGs from the program.
I did a 30-minute test, creating a flow chart and messing with SVG export settings and pitched a process of:
We had to work rapidly to get things approved to keep the project moving.
First, we confirmed the site structure by creating high-fidelity screen layouts of the major components of the site. I presented these design’s live at Lilly and walked stakeholders through the flow of the application. I made what live tweaks I could to design and noted those that would require more effort after the meeting.
At every design milestone, we would have the EyeTrack test our UX – this was Lilly Policy. This is UX testing where you sit in a room – kind of like MTV’s hit show “Punked” – and watch someone use your app, hear them talk about it, and get data about where their eyes went on your site. It was incredibly cool, but also, another time-consuming thing at each phase of the project.
The High-Fidelity designs did really well in the UX testing, we got some ideas for tweaks, but I said that we’d implement the tweaks in the first functionally programmed version of the application, which would be one working map and resources page.
Did I mention that the process maps were not finalized? I think I did.
So… back to one of the big challenges. We likely won’t have final process maps until Mid-November…though they would be close in late October. When we were ready to start coding the 2nd week of October, we had 1 (of 8) process maps as a Visio document. It was ugly, we also needed to make it look as good as a complex process map could look.
I told our Graphic Designer in charge of making the process maps to make a pretty version in Illustrator, of the current version of the finished map. Once I had that, I took the Illustrator file, and adjusted it including groupings and layering, to create a document that would export to SVG in a format that I could work with in code.
I then created a how-to video and instructions on creating/updating the additional maps.
Each one of the steps on the process, and the flow of the process would be continually changing as I was developing…so how do I remove the burden of large updates each time a process flow or step changed?
Oh, another problem, with the quick nature of the project, we didn’t have time to get set up in Contentful, a content/backend as a service tool that would act as our data store, so I would need a temporary solution at launch. A temporary solution that our non-tech clients could help edit. A temporary solution that wouldn’t cause a lot of rework to reload the content into the application during edits (of which there would be hundreds, constantly, frequently, daily).
With this problem at hand, I knew that the final process map had to be my “source of truth”. That specific resources would have to be loaded for each step of the algorithm, that specific text and other materials would have to be loaded for each given step. My solution was to write an algorithm that would run at startup of the application, which would parse the SVG process maps, and “automatically” link each of the items within the SVG to the content it was supposed to display.
If you are a developer or have a dev background, you probably threw up in your mouth a little bit at that headline. But, it’s to get content into a format that anyone could edit, including the client, I created an Excel document that mimicked a relational database that the app would use.
With this in place, both internal writers and stakeholders at Lilly could change our content in the Excel document (a commonly understood format) and save a new version.
When a process map changed, the Illustrator file changed, and the Excel “database” was updated to reflect those changes. New steps were added (though like a database, no in any particular order), text and resource linkages were changed, etc.
With my “killer” algorithm I could export a new SVG from illustrator, update the Excel document, put them into the app, and the code would take over and make everything work. This was initially done using text-matching for process step names, but eventually was changed to utilize content IDs.
This allowed for daily update of the app via Illustrator & Excel, that didn’t require much in terms of reprogramming anything once the algorithm was created.
We delivered the first version of the application in this manner, to rave reviews. Again, we did EyeTrack testing which confirmed we were on the right path.
The process map Illustrator files and Excel document continually changed, but the app was set up to leverage this for minimal rework. We added the 7 additional process maps leveraging the same structure, and it worked really well. Because of working in Excel, we could pivot content very nimbly. By mid-November, the app was pretty close to it’s final state.
Because the overall project manager didn’t have any experience with software, I took the project management duties of the app. This meant setting timelines, deadlines, and orchestrating QA testing both internally and with the client.
I created a robust testing plan that would allow a team of 4 testers on our end (1 person to test 2 maps each) to confirm proper function and content of over 1000 pieces of process-related content and linkage.
I was both a hoodie wearing, caffeine-drinking developer behind the scenes, and a polished presenter and client partner when the need arose.
In mid-November we got the final onslaught of content changes. A lot of content changes, nearly 1000 of them. Some of these included things like “sub processes” within larger process maps (a new feature) and creating “how to use the tool” training videos. I would develop any new features, enhance existing ones, and delegate any content and/or video requirements to other members of the team.
That said, there was a lot of work in a short period of time to finalize, and get the application ready to be deployed to production within Lilly’s Lucid framework (an express-based, node backend).
I never told anyone this, but I skipped Thanksgiving that year and was pulling 12-14 hour days during the 4-day Thanksgiving weekend to make this happen…but I made it happen.
In the end, we made this work, we launched Lillymarketingstudios.com on time and on budget.
Two of the most fulfilling moments of my career came from this project came from this project.
The first was one-day at Lilly, someone said “This is the guy who made Lilly Marketing Studios, wow, it’s incredible, thank you so much, we love it.” Just random verification that the software I made had a positive impact, worth my weight in gold.
The second, completely unexpected, was that this whole project was presented as a case study at ExLearn, a training conference BLP put on that year. My manager was presenting, and someone in the crowed asked “how did you develop such a site on such a restricted timeline.” My boss said that it was one developer, me, and a ballroom of 100+ people erupted in applause. Not something I am looking for, not something I expected, I was probably as red as a granny smith, but it was the first time it really hit home as a big accomplishment.
After this project, BLP got millions of dollars of future work with Lilly. Once, my boss told me that the reason we got all this additional work, was because of what I was able to accomplish with the app that I developed. My 2 months of pedal to the medal made a difference in our small business, and that means a lot to me.