Life360 is on the brink of shipping the first family-intelligent feature in our company’s young history! The “family-intelligent” principle doesn’t necessarily mean machine learning, but yes — it does use machine learning (ML — so hot right now).
This milestone marks both an organizational and personal sense of accomplishment because we did not just click our heels and Dorothy ourselves to this point — it was a long, arduous, amazing journey of learning to convert data into value. So let’s rewind real quick!
I’ve been asked about Life360’s data journey by several people who are also tasked with strategically leveraging the power of data within their growing team or company, so I’ve decided to tell my story here!
Wait, what is Life360?
Life360 is the biggest family company that you’ve never heard of — it’s a budding growth stage company (especially in the past 3 years!) that provides services to give your family peace of mind while everyone’s on the go. The core value prop is based on the fact that you and your family are either 1) at a place or 2) going to another place.
Life360 collects, synthesizes, and surfaces that information to others in your private family all through our (free!) mobile app.
Our mission is to make family life easier, safer, and smarter.
Thus, Life360 Analytics’ mission is to:
- Differentiate Life360 as family experts
- Fuel family-intelligent features
- Provide smarter business growth
Top 3 Lessons Learned
- Data is complex: never assume stakeholders are on the same page.
- Write down your data strategy, then socialize/discuss/refine until stakeholders start understanding the details (e.g. defining Company and Team KPI’s, key performance indicators). Lather, rinse, repeat — early and often!
- Don’t underestimate the power of TV dashboards (i.e. built our first automated KPI Dashboard from scratch with python + MySQL + php + html + crontab). Humans love watching TV, so everyone should love watching how their business is growing.
- Democratize data access and create learning processes to promote company-wide insights: review meetings, tech talks, emails, reports. Offering forums to discuss data projects help people interpret + internalize conclusions properly.
2. Use the Stairway to Analytics - but crawl, walk, run. Don’t just run up the stairs — just like mom said!
- Part of democratizing data, initially led to building an in-house analytics platform (Redshift, Re:dash, mysql, alooma) that could perform complex custom analytics. At 40 people, we quickly realized that we’re not an analytics company and that the overhead wouldn’t scale. So we killed our in-house system and integrated Amplitude to streamline product analytics (highest priority at the time).
- Leverage the 80/20 rule with your data workflows. Don’t build the Tesla of models when the Toyota suffices for taking action. This accelerates time-to-impact and improves operational output (e.g. LTV, Lifetime Value, models initially aimed for high degrees of prediction confidence but this wasn’t needed until our customer acquisition efforts reached greater scale!).
- Understand how all layers of analytics (from ETL’s to smart features) are interdependent. The more that features, insights, reporting, and infrastructure are unified — the more effective and efficient your data initiatives will be.
3. Tell the story behind the data!
- Frame research focus around your customer (duh!). One of our core company values is to Know Peggy (Peggy is an internal customer persona we developed) — so a key initiative is Family Insights Research. Better understanding what our users do/think/feel will develop our organizational intuition around families to help make better product decisions and deliver more customer value. Ask What Would Peggy Do?
- Make a marked effort to blend quantitative and qualitative research. Yes, analytics and data science use a lot of numbers, but it’s the story behind the numbers that matter. Quant tells you the what, qual tells you the why. Life360 users subscribe to receive over 1 billion Place Alerts each month(!), but users tell us (via surveys) they love that feature because it gives them a feeling of “relief”.
- The aforementioned “family-intelligent” feature may use advanced analytics/ML (python scikit-learn + AWS redshift + jupyter notebooks + iOS Core ML), but the goal was not ML. Through behavioral analytics and user research - we discovered that users love driving behavior features, but feel that the experience could be easier to find. So we built a smart feature to contextually surface interesting driving behavior and make Peggy’s life easier (shout-out Mike/Amanda!).
There is a rhetorical question I ask at the end of every data discussion…
- Each data journey is different, but I hope this helps you or someone you know better understand what to do (or not do to!) when building an analytics organization.
- If any of this piqued your interest, join Life360 to be a part of building something great!
- Have questions, comments, random inquiries? Feel free to reach out! (Here’s my LinkedIn, thanks for reading!)