ADTYPE

Better client reports through data visualisation and automation

Designed an internal data visualisation tool to help decrease time needed to onboard new clients and share monthly progress reports.

COMPANY

YEAR

AdType

2019

TIME

ROLE

Only designer responsible for product design

4 months

AdType, a digital marketing agency, dedicated significant time manually creating reports for clients during the sales process. Reports helped to understand where the clients business is doing well and where improvements can be made.

For existing clients, these reports would help to evaluate the agencies performance.

Creating a report was a bottleneck for the agency. Senior analysts would manually gather, clean, and consolidate data from client systems, upload them to tools used internally and compile insights into spreadsheets to create reports.

Automate a labour intensive process

I was contracted to design an internal tool that would help analysts automate report creation and continuously monitor campaign performance through integration with other systems to free up their time for other work.

I needed to understand which metrics are important and how different data points can be combined into purpose driven dashboards. Each metric should be put into the context of the previous period and provide an insight into next actions.

Needs vs. systems

I conducted a series of interviews with people in the company and a client to understand what is the main issue with the current process, potential challenges and purpose for this tool.

They explained that the tool would be designed for senior professionals in sales, marketing, and management roles, referred to internally as "marketers."

Internally, the tool needed to provide reliable and actionable metrics, easily contextualised by comparing them to previous periods. Externally, it was crucial for the tool to deliver accurate insights, allowing clients to review each report individually.

Senior stakeholders want to transition the agency from spreadsheets to a digital tool that could scale the sales process.

Sales people want the report to tell a story about client business performance, what trends can be observed and where to target a campaign.

Analysts explained that client datasets came in various file formats, often as spreadsheets. The data required cleaning, normalisation, and adjustments. For sales calls, clients typically provided a dataset, while existing clients granted access to their tools. Analysts also assisted in organising and uncovering supportive data.

The tech team emphasised the complexity of importing different types of data to gather insights. It needed to be combined and processed with calculations before it could be accurately visualised.

Clients are locked into insufficient tools

While exploring products clients used for marketing, I understood that they don’t provide sufficient information to make decisions which is why this task is outsourced to agencies.

Looking at other tools in the market, I learned that usability and ease of use are not a priority. These tools mostly provide different tables with very little context. They are expensive to setup and hard to use.

Client systems usually did not provide sufficient insight and exported messy data.

Building a source of truth

At the beginning requirements were scattered in spreadsheets, various documents, and paper sketches of charts and graphs. I gathered all requirements into a single document that served as the source of truth going forward.

Aligning stakeholders was a particular challenge since there was a tendency for requirements to change mid-development delaying a release and frustrating the dev team.

I organised a series of workshops to outline requirements, get buy-in for the source of truth document, and agree on a roadmap for which charts should be designed first.

Putting it together

To get things moving, I would start with wireframes one section at a time. I iterated quickly to minimise the time for approval. Sales people often instructed specific data to be included in designs so it would make sense to them what they are looking at.

Each report would be data heavy and explained by a sales person, so I wanted to highlight key insights as a readable story. For the following reports, the challenge was to create reusable components of cells and tables so that they can be used consistently for other reports.

Once a design is approved, I would be responsible for writing tickets for the engineering team.

I made sure that each ticket is checked by someone in management, sales and an analyst to give engineers confidence that requirements won’t change later.

Once a feature is implemented, I would serve as a QA to make sure things are right. Any bugs or changes I would document in tickets.

Modelling revenue growth

The first report I created represents overall revenue generated by different inputs.

The dotted columns are projections based on the inputs below. By adjusting any input, a marketer can model different potential outcomes. Inputs that come from the dataset are not adjustable, only projections are.

Each projection can be saved or exported as a pdf to share outside the system. In the future, the idea was to create read only access links to share a link instead of a document.

We defined 6 types of customers: Prospect, Enquirer, Customer, Lapsing, Lapsed, Reactivated

Managing consent

Since GDPR, managing customer databases has become more complex. Large organizations risk fines for marketing to individuals without consent. As consent is not permanent, visualizing the current state and ongoing changes in the database became crucial.

Spotting trends

Here marketers can analyze sales data to see how a business is performing compared to the previous year and whether it's on track to meet its goals.

While this is a data-heavy project, I was able to include some explanations because the variance breakdown can cause some confusion.

Translating data into insights would require more development time which we did not have at that moment, so the idea was postponed.

Segmented customer databases

Dividing the database into three customer groups enables analysis of their relationships. By tracking changes in Prospects, you can pinpoint the sources of positive or negative shifts in Enquirers and Customers. Customers can belong to any of the groups based on their actions.

This approach also provides a visual representation of the customer lifecycle, helping to identify which areas influence each customer group.

Additionally, it allows you to track movement between groups, revealing where customers are dropping off, how effectively they convert, and the financial impact of transitions between customer states.

Result

By the end of the project, we had 1 client onboarded into the system which helped to understand the process for next client. We had started conversations with 3 other clients but before moving forward, we needed to improve our data accuracy.

Reflections

Creating data heavy projects is very interesting but also complicated. To make best use of each report, we needed to standardise the sales process and agree on a set of metrics which to use.

  • Data accuracy is an issue. When you are importing a dataset, there will be issues. Constantly checking data accuracy was time consuming. We used a few algorithms which combined several data points and those also needed to be manually recalculated before showing the client.

  • Aligning people is complicated which is why the source of truth was so useful.

  • Making stakeholders to sign off on tech tasks improved the working relationship with engineers.

  • Iterating quickly is a good way to avoid constant redesign.

Next steps

The initial outcome of the project is a tool that saves time for people whose time is expensive. To understand the usefulness of the tool, more more clients need to be onboarded and feedback gathered.

  • The tool could be productised giving clients autonomy by building an onboarding process, support for internal and external accounts as well as an admin panel.

  • Over time, simplifying reports and breaking them down in to smaller chunks could help with legibility.

  • Expanding areas for insights such as customer profiles and product performance could help with warehouse restocking.

  • Building out the campaign builder could help bring insights from all reports, select a cohort of marketable customers and suggest a channel through which to advertise.