Create a direct mail campaign to convince high level IT decision makers (ITDMs) to consider Google Cloud Platform as a provider.
IT Decision Makers: in “high potential” target companies who will use more the $2,000 per month in cloud services. These ITDMs want to mitigate financial and technical risk—which cloud provider are their competitors using, who has the best uptime, which contract is easiest to negotiate?
We sent an empty, personalized Google Home base to ITDMs with the promise that if they signed up for a sales meeting on our web portal, we'd send them a Google Home equipped with a voice app that could answer any of their questions about Google Cloud.
We exceeded our benchmark response rate by 8%, had an average talk time of 8 minutes and won a Silver Effie in Software and Apps
My role: Lead UX Designer
During the previous year, Google Cloud conducted direct mail outreach with their Chromebooks as their "carrot" to entice customers. We were charged with optimizing that experience. Here are the basic CRM mechanics:
We traveled to the Google offices to investigate their current direct marketing efforts, the integration between sales, marketing and product, and learn about their past attempts to convert ITDMs.
We interviewed heads of marketing and product, sat in on sales calls, and spoke to different mid-level members of the Cloud sales team to understand ITDM concerns, requests, and proof points.
We learned that many ITDMs have existing agreements with competitors (Microsoft) or have an easier time recruiting developers to work on a different platform (AWS). We also learned that the Google name is very appealing to people and that Google has a couple niche products for certain industries that gave Google an "in."
We contracted a third party vendor to conduct in-depth interviews. Together with our vendor Jenny, I created an interview schedule to gauge ITDMs decision-making process, the challenges they face, the cloud purchase process, and how they viewed our competitors.
We learned that ITDMs are very risk averse: they want a provider that has the best uptime rate, they want their data secured at all costs, and they want their cloud provider to properly integrate with their existing systems. ITDMs also value the opinions and actions of their peers above all else.
Our last bit of discovery involved which tech platform we would use to build our voice assistant, which would inform how we would "design" our product. Given that we were working for Google, we used their proprietary platform, now known as DialogFlow.
DialogFlow works by identifying spoken keywords and then returning the answer it thinks best fits that collection of keywords. This knowledge informed how we'd structure our content and more granular conversational elements.
Before ITDMs could access our voice assistant, they needed to sign up for a sales meeting with a Google rep (and give us their contact information). We used Google's Locus platform to create custom landing pages for each ITDM to give us their contact information, sign up, and also tell us about their interests. After filling out the streamlined form (from its previous iteration), ITDMs would see a landing page with articles tailored to their selected interests.
I designed the form and web experience for ITDMs who responded to our direct mail. We wanted to make this process as easy and personal as possible.
For our initial campaign, we created an MVP of the Locus experience, as we wanted to get our campaign started as quickly as possible.
After our first round of direct mailers, we concepted new marketing techniques for Google to better convert ITDMs.
These concepts included creating "sandbox" spaces for ITDMs to test GCP products (given people's desires to try before they bought), interactive roundtables with Google executives via Hangouts (as many ITDMs valued the Google brand and their institutional knowledge), and a pricing calculator for all GCP products.
We focused on the following key areas of ITDM concern: what Google Cloud can do for others in your industry, what can Google Cloud do for your business, what are the key product attributes that Google Cloud has (e.g. uptime, security).
When beginning to tackle the Assistant, we bucketed out these broad content areas and the developed specific sub-queries. This initial doc provided our blueprint for how to move forward.
Our content pillars informed the different types of conversation we'd design for (industry-specific stories, product FAQs, business needs that Google could solve). At this point, we didn't have our conversational elements, which include things like follow up prompts (a question at the end of an answer to prompt a user to ask something new), error messages (when the assistant didn't understand) and conversation forking (the assistant asking a clarifying question which would take a user down a certain conversational branch).
To start building out what these content pillars would look like, we starting role playing out how these conversations could play out, given our user research. We adopted different personas and wrote dialogues for each.
After these exercises, we had (lots) of conversations with one another to pick apart the different "interface elements" of a conversation. We role played as different ITDMs from different size companies and with different cloud partners.
We created dead end repairs, so that even if our assistant didn’t understand user input, the assistant could still provide useful/ancillary information or ask another question.
With our conversational pillars and the conversational elements we derived from our roleplay exercises, we dove more deeply into specific queries and the types of responses we’d use for each content bucket. We identified our welcome messages, which content buckets would be linked to one another, how error messages would be handled, and easter eggs (amongst other buckets).
The final result was this massive Conversation Architecture, which eventually took on the look and feel of a nuclear power plant schematic.
This conversation architecture and our roleplay formed the basis for our conversation bank. (The creation of these deliverables were symbiotic, as oftentimes we'd realize loopholes in our Architecture by building out our Bank).
Our "conversation bank" was a document used by our developer to seed our voice assistant with keywords and answers. For each keyword/answer, we identified its category (e.g. industry vertical), its theme (e.g. welcome message, error message), and other related queries.
Our content strategist was the master of this document, but I worked closely with her to structure, populate, and audit the document.
This is a screenshot of our conversation bank. I'm unable to share the full document.
Our conversation architecture, regardless of its complexity, was just an MVP to test the effectiveness of this marketing campaign. We concepted potential future features that added: new content using the existing DialogFlow Q+A platform, new content with an API to Google's data, and thought leadership and CRM opportunities.
One of those features was a pricing calculator, which we chose to be the first feature to architect after launch.
From our product roadmap, we chose the 'pricing calculator' feature to build, as a primary ITDM concern was pricing. GCP also uses price as a positive differentiator.
The pricing calculator enables ITDMs to configure their pricing estimates with their voices. We architected BigQuery (Google's data warehouse product) pricing out first, as this product has the simplest pricing variables. Even though there were only a couple of inputs (how much storage used, how many queries used per month), the ensuing conversation became quite complex.
When we conducted our in-depth interviews, we also conducted a user test on the first version of our Assistant.
I usually conduct my own user tests, so it was great to collaborate with other moderators. I provided key prompts we needed to tests write, monitored and guided the interviews, and interpreted insights from our data.
Using our findings, we re-wrote certain answers, re-worked certain flows, and added in extra keywords for certain answers. We also flagged areas for future optimization.
This image is a clip of the notes I took during testing (I can't include the full notes or our discovery deck we created with our vendor partner).
We saw a direct response rate of 16%, with our expected response rate was 5%
We saw an average interaction time of 8 minutes
We won a Bronze Clio for Direct Marketing and a Silve Effie (an effectiveness award) in the Software and Apps category.
