Scaling up Sales Analytics

“Analytics software is uniquely leveraged. Most software can optimize existing processes, but Analytics (done right) should generate insights that bring to life whole new initiatives. It should change what you do, not just how you do it.”
- Matin Movassate, Founder and CEO at Heap Analytics

The Need to Scale

Analytics should evolve as the organization grows to reflect its changing needs. It should:

• Keep up with the growing volume and complexity of incoming data by leveraging Process Automation to maintain efficiency.
• Tackle the Silo Effect that can creep up as departments become more specialized by maintaining a Master Database to ensure a Single Source of Truth.
• Answer different questions around Growth, Customer Experience, Retention, and Churn.

At the start of 2021, our Sales Analytics team only had one analyst who handled the projects from end to end. We would capture a few key high-level data points for simpler analyses, like tracking the number of leads at each stage of the funnel.

As Hevo saw rapid growth in Sales Leads in the past year, each stage of Analytics projects demanded a more nuanced workflow. We saw the need to capture more granular data points around specific behaviours ~ like the number of Sales Touch-points before the first successful Pipeline setup to gain a deeper understanding of the user journey.

These new data points required new Pipelines to be set up, cleaned, processed, and maintained. They also presented opportunities to conduct more complex analyses and derive more critical insights, rather than simple reports — like setting up a Lead Scoring Model based on a Probability Analysis, to help customize lead interaction based on the user journey.

Overall, more data needed to be captured, cleaned, maintained, analyzed, reported, observed, and actioned regularly, which needed more hands-on-deck

The Analytics Mindset

As the Analytics function scales, we employ the following mindset and approach-

1. Peeling the Onion: To answer broad questions like “How are Sales this month?”, it can be helpful to use a top-down approach to identify any potential opportunities and obstacles at each stage of the funnel.
2. The Early Bird Gets the Worm: Assessing past performance can be essential to identify what went right (and wrong) while executing a strategy. However, this must be accompanied by tracking leading indicators and predictive models that buy valuable time to identify shifts in trends early and pivot as needed.
3. Digging Deep: Running experiments before employing any major strategy shift ensures that reality reflects the findings. Additionally, leveraging tools like Fullstory and Smartlook that maintain detailed video recordings of the user’s sessions on the product can provide useful insights where broader summaries may not suffice. Prior broader analyses can help identify the records where a detailed exploration can bring the highest value.
4. Picking your Battles: The increased Data Volume also meant that we had to decide which data to focus on. We leveraged a simple value vs complexity review to prioritize metrics. High-value Low Effort metrics would be fit to automate for regular consumption (ex: product session CTAs), High-value High Effort metrics would require a bit more processing (ex: persona data), Low-value Low Effort metrics would provide handy auxiliary insights on a more infrequent basis (ex: high-level browsing data), while Low-value High Effort metrics would be deprioritized (ex: granular browsing data).

The Current State

Now, we are a team of 4 Analysts working with the Sales, Customer Success, and Strategy teams to work towards a combined goal of Sales Excellence & Customer Success.

The following process helps us make quick strategic decisions and pivot rapidly if a particular strategy doesn’t work out.

A) Planning Phase

Every Analytics project begins with an overarching question — usually a specific, long-term objective. For example — “What should be the Sales targeting strategy to achieve X% conversions out of Y total leads?” The team then brainstorms on translating this Broader Objective into a few Quantifiable Key Results (OKRs) to be achieved by the discussed timelines. These KR’s are then distributed amongst the respective assigned owners. Regular cadences are scheduled to review progress, tackle roadblocks, and build on the strategy.

B) Execution Phase

The next step in the project is to gather data, analyze it, observe trends, and derive actionable insights.

1. Locate the Mine: We identify the required data points and ascertain where to collate the data from. Some data may not be readily available and may require new workflows. For example, Sales aimed to prioritize Lead Targeting Strategy based on persona/job role. However, the existing fill rate for this data, derived from user input, was not satisfactory. Hence, we set up an integration with Clearbit, a Marketing database for enriching leads data, using Zapier. This additional data was then fed into a Google Sheet for further reference.
2. Mine the Stone: There is no Analytics without extraction of accurate, timely, and relevant data. Hevo Data Pipelines are set up to integrate data from various streams and sources in a single database. For example, Sales data from Hubspot CRM, Product data from Segment, Conversion data from Stripe, and additional data from Google Sheets were collated into a single Redshift database.
3. Polish the Diamond: We, then build reports that track well-defined and tailored goals rather than broad generic benchmarks. Hevo Models are set up to transform and enrich the collated data through complex queries to build the key master datasets that will be used for analysis. This stage is also where we clean up the data, account for missing values, exceptions or outliers, etc. Metabase Dashboards are set up to generate valuable insights through customized dynamic reports. For example, we set up a weekly funnel view that tracked Key Performance Indicators (KPIs) at each stage. Observations from these analyses are then summarized in handy tables and graphs within Metabase or even Google Sheets.
4. Distribute the Loot: Each data point has a different granularity, urgency, and audience associated with it — it’s important to select the right channel to communicate the data. The final observations are discussed. Actionable insights are derived from these observations. These actionable insights are then distributed to the required stakeholders. Hevo Activate feeds the processed data from Hevo back into the Hubspot CRM interface. This empowers Sales with end-to-end data around user behaviour, resulting in valuable and contextual sales conversations. This data then helps Sales make an informed decision while contacting a lead. Google Scripts integrates existing documents on Drive. Zapier integrates with Slack Messaging for real-time notifications to keep Support and Sales informed about prospects who are facing issues or are drifting away. For example, we found that leads who didn’t clear the early stages of the funnel within a week tended to drop off. Hence, we set up a Zap that connects to Slack and informs Sales when a new lead hasn’t made significant progress for a week. Metabase Pulse emails and reports for regular updates to help management make key business decisions. The progress and indicators that were established at the objective stage are tracked before and after implementing the changes to measure the efficacy of changes in strategy.

This flow of analytics projects helped us scale well from a smaller team to a larger one effectively. Do let us know of any unique hacks that helped you streamline analytics projects in the comments below!