How to strategize outside the data

Consider analytics competency as a powerful tool for your business. In fact, include it to make any decisions on your core strategy.

Analytics is not a language — like the linguistic typology kind (I know! big word). Analytics is about making better operational decisions. It answers important questions. Have you ever asked yourself how to prevent a customer from churning*?, how to convert a visitor? how to acquire a prospect? how to identify a fraudulent claim? how to correctly estimate the risk of a loan?

And, do you understand the challenges (understanding business value gap, engaging business and IT to deploy and analytics strategy) your startup will face to fully use analytics to your advantage?

A Bersin by Deloitte study found that companies using ‘sophisticated’ data analytics see 30 percent higher stock market returns. You would agree that sophistication = money. why? because there are companies out there with enormous amounts of cash giving that cash to other specialized companies, and in many cases, high growth startups, to build their analytics competency as they grow. Those companies can afford to address every challenge i. e. business value gap, full-on IT support, and deployment, sometimes all at once because they have the cash and have made the time to formulate a strategy.

But what do you do when you can’t afford such? build strategy!

When I was putting notes together for this post I realized that so much has been written about analytics and my goal became to give you an overview of what to do with analytics once you have decided to make them part of your strategy. (hence, the language reference, because once you understand the purpose of analytics in the context of your strategy, you are in effect speaking that language)

Remember, at every stage, every company and startup will need an analytics strategy, and yes that even includes the stage where your funding is robust. So, consider this an overview. BUT FIRST!, It is very important that you assess the actual need for Analytics at your company; I always say, master what you built and then add more features as you grow.

Founding Stage: 0–10 employees.

This is not the stage to buy a data warehouse, a BI platform or a big consulting project, etc. Only commit to analytics when you are ready to make the investment, why? because there is an ongoing cost, data changes, business logic changes, etc.

The rule for this phase: Master the basics seek simplicity.

№1: Implement Google Analytics. Install Google Analytics on your website via Google Tag Manager.

№2: Measure your product; product metrics will help you iterate timely.

I am working on writing a post about ‘financial reporting’ for startups, but to sum it up…. use QuickBooks!

№3: Do your forecasting in Excel, I know there are fancy tools out there but remember: S I M P L I C I T Y

• If your company is in the business of building software (of any type), you need real event tracking. The ones I always tell everyone to use are Mixpanel’s autotrack or Heap’s default installation.
• If you are a subscription-based business, use Baremetrics for your subscription metrics.
• And, if you are an e-commerce business, use your shopping cart platform to measure GMV and stick to GA to track every step, from visitor to purchase. If you’re not technical, you may need an engineer to help out with GA and event tracking.

Seed Stage: 10–20 employees.

Purchase SaaS and use its products’ built-in reporting. It’s important to track UTMs* at this stage. And, it is likely that at this stage businesses may not be built for a data warehouse or a SQL-based analytics. Therefore, do not hire an analyst yet.

№1: Hire a marketing person responsible for GA*. Make sure their priority is to keep data clean, UTM track and make sure subdomains aren’t double-tracking. If the marketing person you hire refuses to have control of GA, FIRE THEM! and find someone else (seriously).

№2: Use a CRM, preferably Salesforce, and use its (terrible) built-in reporting but do not export data to excel. I can not get into the details of why because all I know is that it doesn’t work. What is important is to learn and make sure that your team knows the basics of what you want to measure and what is important to measure at this stage is rep productivity and conversion rates by (sales) stage.

№3: Have a helpdesk system and choose KPIs that can be easily measured within the interface (FYI, most don’t have great reporting)

№4: Make sure you track NPS — Net Promoter Score (NPS) — often considered the metric that rules all metrics! and is based on the willingness of consumers to recommend your product to someone they know. Why is it important? because of WOM: Word Of Mouth. WOM is very important for customer acquisition and retention. Use Wootric or Delighted.

Early Stage: 20–50 employees.

This is the stage to set up data infrastructure and hire an analytics lead.

Startups at this stage have more options. Mainly, because they have matured a bit and perhaps have secured round A of funding. The money allows these companies to have access to a better platform (via subscription or built-in), have more flexibility on what strategy they will eventually implement and have gathered more reliable metrics.

№1: Choose a data warehouse (look into Snowflake and Redshift), an ETL* tool ( Stitch and Fivetran), and a BI tool* (Mode and Looker)

№2: Hire an Analytics lead with an MBA and background in consulting or finance and who thinks about data and business very strategically. Keep in mind that at this stage the business needs a generalist, not a specialist, to build your analytics team.

Mid-Stage: 50–150 employees.

At this stage businesses should implement an SQL*-based data modeling; they should migrate to more robust web analytics and tackle select forecasting challenges. Also, it is important to figure out marketing attribution — The process of identifying a set of user actions (“events” or “touchpoints”) that contribute in some manner to the desired outcome and then assigning a value to each of these events.

№1: Data models* serve as the underlying business logic* for analytics. Make it accessible from BI to data science. It is important that the process allows all users to make changes to data modeling scripts, is version-controlled, and is run in a transparent environment. Use an open-source product called dbt

№2: Migrate from your existing web analytics and event tracking to Snowplow Analytics. Snowplow its open source. If you decide to host it, just pay the costs of EC2* instances. You can also pay Snowplow or Fivetran to host the collector for you. It is important you make this transition during this phase otherwise the bills from Segment, Heap, or Mixpanel down the road will be enormous.

№3: Stay Agile. do not make a big data infrastructure investment. Instead, build a team of Business Analysts who are experts in SQL and BI tools and who will help you build your data infrastructure and core analytics, it is then that you can hire a data scientist, but you need to add Jupyter Notebooks for data science work. Finally, push SQL and use the processing power of your data warehouse (pay for servers) Find low-cost ways to ETL*datasets that don’t have off-the-shelf integrations and use Singer for low-cost ETL datasets.

№4: Focus your forecasting on key areas based on your offering. For example, if you are a SaaS business your forecasting model should be based on churn; if you are an e-commerce business, concentrate on a demand forecasting model.

Rule N.5: Do not trust your marketing attribution to a third party.

Growth Stage: 150–500 employees.

Create and implement analytics practices that scale, this means, start data testing and implement pull reviews (requests/code)

№1: Have a solid automated process, use dbt’s testing functionality to ensure that the data being loaded conforms to the rules you are expecting.

№2: Produce high-quality analytic code. All code should be merged via a pull request process that includes a review from a team member.

№3: Document all your data and processes so that analysts do not spend time looking for data or figuring out how to use it. Check what Airbnb has done.

№4: Decide the structure of your analytics team: centralized and embedded. Carl Anderson describes the trade-offs well in his book Creating a Data-Driven Organization.

Conclusion:

Don’t try to design a “perfect solution” — start with building out main functionality, then iterate and add features. Think of it as Agile Analytics development. As complexity increases, you need to evolve your processes to adapt.

Remember that the most critical decision is not building an analytic solution but making sure that your organization starts using it, that means creating buy-in, working to build adoption, educating and training, redesigning processes to include analytics. Give it time, be persistent, improve and results will follow!

Recommended reading:

• The Magic of Universal Analytics: Strategy, Tactics, Implementation Tips by Avinash
• www.online-metrics.com/web-analytics-metrics

_____________________________Glossary____________________________________

UTM — Urchin Tracking Module parameters are five variants of URL parameters used by marketers to track the effectiveness of online marketing campaigns across traffic sources and publishing media. They were introduced by Google Analytics’ predecessor Urchin and, consequently, are supported out-of-the-box by Google Analytics.

GA — Google Analytics is a freemium web analytics service offered by Google that tracks and reports website traffic.

Subdomain — is a domain that is part of a larger domain; the only domain that is not also a subdomain is the root domain.

SQL — is a domain-specific language used in programming and designed for managing data held in a relational database management system

SQL Server — is a relational database management system (RDBMS), that supports a wide variety of transaction processing, business intelligence and analytics applications in corporate IT environments.

ETL — extract, transform, load refers to a process in database usage and especially in data warehousing.

BI Tool — Business intelligence software is a type of application software designed to retrieve, analyze, transform and report data for business intelligence. The applications generally read data that has been previously stored, often, though not necessarily, in a data warehouse.

Churning — churn rate is the percentage of customers who cancel their subscription to your product or service within a given time. A high churn rate indicates that there’s something about your product or service that people either don’t like or didn’t understand (website, pricing model, customer support, design, UX… the list goes on) A high churn rate indicates you’re losing customers, but it won’t explain why. To calculate churn rate divide the number of churns (or cancellations) during a certain period by the number of customers at the beginning of that period (it’s not that simple, go somewhere else to read about it)

Data Models are fundamental entities to introduce abstraction in a DBMS (Database Management System). They define how data is connected to each other and how they are processed and stored inside the system.

Business logic is the programming that manages communication between an end user interface and a database. Is the part of the program that encodes the real-world business rules that determine how data can be created, stored, and changed.

EC2 Elastic Compute Generation 2 which means you can purchase or lease hardware or software infrastructure hosted on the cloud by the provider on an on-demand basis. Elastic meaning you can grow your usage or shrink your usage and pay for what you have used.

An attribution model is a rule or set of rules, that determines how credit for sales and conversions is assigned to touchpoints in conversion paths. For example, the Last Interaction model in Analytics assigns 100% credit to the final touchpoints (i.e., clicks) that immediately precede sales or conversions.