You’ve probably stared at a dashboard showing 47 different metrics, felt briefly important, and then closed it without changing a single thing. That’s not analytics. That’s theatre.
Real product analytics answer one question: what should we build or change next week? Everything else is noise.
The Dashboard Trap
A typical SaaS founder’s analytics stack looks like this: Google Analytics, Mixpanel or Amplitude, maybe a custom Looker dashboard, possibly Intercom data, spreadsheets with manual counts, and Slack notifications that stopped being read in week three.
The problem isn’t the tools. It’s that nobody defined what they’re measuring for.
Most dashboards track vanity metrics-numbers that feel good but don’t predict success. Page views went up 12%. User signups increased. DAU is stable. None of these tell you whether users actually find value in what you’re building. A few thousand users can mean everything or nothing depending on whether they’re coming back, completing key actions, or paying you money.
The trap tightens when you add more metrics hoping clarity will emerge. It won’t. You’ll just have a prettier spreadsheet of confusion.
What Actually Matters: Input vs. Output
Split every metric into two categories:
- Input metrics (leading): Actions users take that predict future behaviour. For a SaaS product, this might be feature adoption rate, time spent in the core workflow, or invitations sent to teammates.
- Output metrics (lagging): Business outcomes that validate input metrics were meaningful. Retention after 30 days, revenue, or expansion MRR.
Input metrics tell you what’s working now. Output metrics tell you whether it mattered three months ago.
A fintech platform we worked with was obsessed with signup volume. They were acquiring hundreds of users weekly. But the input metric that actually mattered-account funding within 7 days-was flat at 3%. They were optimising for the wrong gate. Once they reframed and started tracking funded accounts as the lead metric, their product roadmap shifted completely. They built onboarding flows instead of paid channels, and that 3% nearly tripled within two months.
That’s the difference between dashboards and decisions.
Build One Metric Tree, Not a Forest
Pick your single north-star metric first. Not alongside six others. First.
This should be:
- Directly tied to the value your product delivers
- Something users can influence in one session or a few
- Measurable within days, not quarters
- Correlated with revenue or retention (you’ll validate this empirically)
For a marketplace, it might be completed transactions. For a B2B tool, it might be teams inviting a second user. For content software, it might be pieces published after day one.
Everything else branches from there. Your secondary metrics explain why the north star moved. Your tertiary metrics explain the secondaries. That’s your tree.
The moment you add an unrelated metric “because the board asked for it,” you’ve diluted focus. The board can see one metric move up or down. Everyone understands why that matters. That’s clarity.
The Architecture That Scales: Raw Events, Not Reports
Most founders build analytics backwards. They pick a tool (Amplitude, Mixpanel, or Google Analytics), configure dashboards, and assume data will flow in perfectly. Then they find out:
- The tool didn’t capture the event they needed three months ago
- Custom events have naming inconsistencies (“user_signup” vs. “signup” vs. “account_created”)
- Combining data across platforms requires manual CSV work
- The tool’s pricing scales vertically with event volume, and you hit limits faster than expected
Build the inverse. Start with a data warehouse (Postgres or BigQuery, depending on scale). Send raw events there. Build reports and dashboards on top, not vice versa.
This costs roughly 30-40% more upfront in engineering time. It pays for itself in month six when you need to answer a question the tool wasn’t designed for, and you can do it in SQL in fifteen minutes instead of waiting for vendor support.
For most Australian startups, this means: a simple event tracking SDK (Segment, RudderStack, or custom), piping to your data warehouse, with BI tools like Metabase or Looker on top for the team to self-serve.
The One Dashboard You Actually Need
Stop building dashboards for different audiences. Make one.
Show:
- Your north-star metric (weekly trend, comparison to last quarter)
- Three inputs that predict it (with direction of change)
- One or two outputs validating it matters (retention, revenue, or churn)
- One anomaly section: metrics that moved unexpectedly this week
Refresh it every Monday morning. Email it to the team. Spend 15 minutes discussing it in standup. Make one decision per week based on what you see.
That’s enough. Boring dashboards drive decisions. Flashy ones drive meetings.
When (and How) to Add Complexity
Once your north star is stable and moving in the right direction, add segmentation. Break the metric down:
- By cohort (users from paid acquisition vs. organic vs. community)
- By geography (if you operate in multiple markets)
- By user segment (free tier vs. paid, SMB vs. enterprise)
This is where analytics actually become strategic. You’ll find that your average north-star metric of 8% hides the fact that one cohort is at 22% and another is at 2%. Now you have a direction.
But don’t segment before you have signal. You’ll just split weak data into smaller weak pieces.
Getting Started This Week
If you’re early and not yet tracking anything:
- Define your north-star metric in writing. One sentence. What does success look like in 90 days?
- Pick your input metric. What does a user have to do to get you closer to the north star?
- Set up Google Analytics 4 or Segment with basic events (signup, first feature use, core action, upgrade).
- Export data to a spreadsheet weekly. Calculate your north star. Watch it for four weeks.
- Once you see patterns, consider a warehouse if you’ve got budget and engineering.
If you’re mid-stage and drowning in dashboards:
Audit everything you’re tracking. Delete 80% of it. Pick one metric to optimise for the next 12 weeks. Build one dashboard. Make one decision per week visible to the whole team.
The hard part isn’t the technology. It’s the discipline to care about one thing long enough for it to matter.
If you’re building a new product or platform and want to get the analytics architecture right from day one, talk to Amora about your build. We set up instrumentation during MVP development so you’re not retrofitting it later.
The Real Metric
At the end, analytics exist for one reason: to remove your emotions from product decisions. Not to impress investors or look professional. Not to track everything you can. To answer the question: should we build this, or something else?
Everything else is decoration.
Got something you want built?
Amora Digital is an Australian software and AI agency. We scope it, build it, and ship it – live in 28 days. No offshore teams. No surprises.