The subscription analytics dashboard, explained metric by metric

The three numbers you should look at every Monday

If you only check three metrics every week, make it MRR, monthly churn, and net new subscribers. Together they tell you the entire story: are you growing, how fast, and is the leak getting better or worse. Everything else on the dashboard is either a derivative of these three (LTV is essentially ARPU divided by churn) or a diagnostic to figure out why one of them moved.

The order matters too. Start with net new subscribers — the top of the funnel. Then look at churn — the bottom. Then MRR — the result. If MRR is flat but net new is up and churn is up, you have a leaky bucket that's growing despite itself. If MRR is up but net new is flat, you're squeezing more from existing subscribers (price increases, upsells). Reading the three together tells you which lever to pull next.

MRR: what it actually counts, and the four sub-flavors

Monthly Recurring Revenue is the normalized monthly value of all active subscription contracts. For a $30/month subscription that's $30 MRR. For a $90 quarterly subscription that's $30 MRR (90 divided by 3). For a $300 annual that's $25 MRR. Normalization is what makes MRR comparable across plans with different billing cadences — it answers the question 'if every subscriber paid every month, how much would they collectively pay?'

Most dashboards break MRR into four components: new MRR (from new subscribers this period), expansion MRR (existing subscribers paying more — upgrades, add-ons), contraction MRR (downgrades), and churned MRR (cancellations). New + expansion − contraction − churn = net new MRR. This breakdown is what tells you where growth actually came from. A month where MRR grew 5% on new business is healthier than a month where MRR grew 5% on expansion from an existing whale.

• New MRR — fresh subscriptions started this period. The acquisition signal.
• Expansion MRR — existing subscribers paying more (upgrades, additional products, quantity increases). Often the highest-margin growth.
• Contraction MRR — existing subscribers paying less (downgrades, removed line items). Often a leading indicator of churn next period.
• Churned MRR — subscriptions that ended this period. The retention signal.
• Reactivation MRR — formerly churned subscribers who came back. Smaller bucket, but a sign your win-back is working.

Churn: the metric most merchants calculate wrong

Churn looks simple — what percentage of subscribers cancelled this month — but there are two definitions and they tell different stories. Customer churn (also called logo churn) is the percentage of subscribers who cancelled. Revenue churn (also called dollar churn) is the percentage of MRR that left. They diverge whenever your subscribers pay different amounts, which is almost always.

Customer churn at 6% and revenue churn at 3% means small subscribers are leaving and big ones are staying — usually a good sign. Customer churn at 3% and revenue churn at 6% means whales are leaving — much worse, even though the headline customer-churn number looks healthier. Always read both.

There's also gross vs net churn. Gross churn ignores expansion — it counts only the revenue that walked out. Net churn subtracts expansion MRR from churned MRR, so a high-expansion business can have negative net churn (revenue from existing subscribers is growing even as some leave). Negative net churn is the holy-grail metric for SaaS investors but is rare in physical-product subscriptions where expansion is usually capped at upgrade and cross-sell.

• Customer churn = (cancelled subscribers) / (subscribers at start of period). Easy to compute, easy to compare to industry benchmarks.
• Revenue churn (gross) = (churned MRR) / (MRR at start of period). The number that matters for the financial model.
• Net revenue churn = (churned MRR − expansion MRR) / (MRR at start of period). Negative is excellent.
• Voluntary vs involuntary — split out cancels that happened because of failed payments (involuntary) from active cancellations (voluntary). Involuntary churn is usually fixable with better dunning; voluntary churn requires a product or pricing change.

LTV: useful as a heuristic, dangerous as a forecast

Lifetime value is the total revenue (or margin) you expect from an average subscriber across their entire subscription life. The textbook formula is ARPU divided by monthly churn — for a $30 ARPU and 7% monthly churn, LTV is $30 / 0.07 ≈ $428. Useful for back-of-envelope CAC payback math, dangerous if you treat it as a real forecast.

The formula assumes constant churn, constant ARPU, and an infinitely long tail. None of those hold in practice. Churn is highest in the first 90 days, then declines. ARPU drifts as subscribers add or remove items. The tail does end — most subscribers have a natural lifecycle of 6-24 months for consumables. A more honest LTV is the actual cohort revenue at month 12 or month 24, read from your cohort table, not derived from a churn ratio.

Use the formula when you're sizing a marketing decision — 'can I afford to spend $80 CAC if LTV is $428?' — but never use it as a planning number for revenue forecasts. For forecasts, look at actual cohort retention curves and project from there. We have a separate guide on subscription revenue forecasts that goes deeper.

Cohort retention: the only honest way to read churn

Headline churn averages everyone together — month-one subscribers, month-twelve subscribers, January cohort, May cohort. That average hides whether retention is improving or degrading over time. Cohort analysis groups subscribers by signup month and tracks how many remain at month 1, month 2, month 3, and so on. The shape of the curve tells you what's happening.

A healthy curve drops sharply in the first 60-90 days (the subscribers who shouldn't have signed up self-select out), then flattens. A flat tail at 40-50% means the customers who survive the first three months are sticky — your product fits, your operations work, your cancel flow is doing its job. A curve that keeps dropping past month 6 means the product itself has a finite use-case — fine for some categories (gift subscriptions, seasonal), bad for consumables where you wanted indefinite recurring revenue.

• Month-1 retention — typically 70-85% for consumables. Below 70% means the widget over-promised or the product under-delivered.
• Month-3 retention — typically 50-65%. The 'is the product actually sticky' indicator.
• Month-6 retention — typically 40-55%. The 'is the cancel flow working' indicator (assuming you have one).
• Month-12 retention — typically 25-40%. The 'is this a real business or a one-year fling' indicator.
• Curve shape matters more than any single point — flattening means the cohort has stabilized; continued decline means structural decay.

Compare cohorts to each other. If the May cohort retains better than the January cohort at the same month-3 mark, something you shipped in spring helped. If it retains worse, something hurt. This is how you find out whether the new save flow you launched, the price increase you put through, or the supplier change you made actually moved retention — by comparing the cohorts that experienced each change.

ARPU, AOV, and the difference no one explains

Average Revenue Per User (ARPU) is monthly revenue divided by active subscribers. Average Order Value (AOV) is per-order revenue. They're different numbers and they answer different questions. ARPU answers 'how much is each subscriber worth per month, normalized'. AOV answers 'how much does each delivery cost'. A $90 quarterly subscription has $30 ARPU and $90 AOV.

AOV moves when you add cross-sells, upsells, or bundle products. ARPU moves when AOV moves and also when cadence shifts — a subscriber switching from monthly to weekly increases ARPU even if AOV stays the same. Watching them together tells you whether revenue growth is coming from each order getting bigger or from orders happening more often.

If you're running cross-sells in the customer portal or one-click add-on emails before renewals, AOV is the metric to watch. If you're running cadence-switch nudges ('subscribers who switch to monthly delivery save 10%'), ARPU is the metric to watch. Don't measure one and report the other — they'll diverge and you'll lose track of what worked.

Dunning metrics: where 5-10% of your MRR is hiding

Failed payments are the silent killer. Roughly 5-10% of renewal charges fail every month (expired cards, lost cards, insufficient funds, fraud blocks). Without smart retries and a recovery flow, every one of those is involuntary churn — a customer who didn't choose to leave but did anyway. Your dashboard should surface this as its own metric, separate from voluntary churn.

• Payment recovery rate — of failed renewals, what percentage you recover within 14 days. Good is 50-65%, excellent is 70%+. Below 40% means your retry schedule or your dunning emails aren't working.
• Average retries to recovery — if it takes 5 retries to recover, you're spamming your payment processor and risking gateway penalties. 2-3 retries should be enough with the right schedule.
• Time-to-recovery — most recoveries happen in the first 7 days. Track the curve so you know when to give up and write off the subscriber.
• Involuntary churn share — what percentage of total churn is involuntary. If it's above 30%, fix dunning before you touch the voluntary-churn save flow.

Cancel reasons: the only leading indicator that matters

Most metrics are lagging — they tell you something happened. Cancel-reason capture is one of the few leading indicators on the dashboard. When the share of cancels citing 'too expensive' starts to creep up, you'll see it in cancel-reasons three weeks before it shows up as a churn-rate increase. When 'product quality' becomes the dominant reason, you have a supply-chain problem brewing that won't hit headline metrics until two cohorts later.

Capture is the hard part. Optional open-text fields get filled by 15% of cancellers and the data is unstructured noise. Required dropdowns with 5-7 specific options get filled by 90%+ and produce data you can actually trend. Pick the dropdown.

• 'Too expensive' — usually a pricing problem if it's >30% of cancels; otherwise normal background noise
• 'Got too much / accumulating' — cadence mismatch; offer pause or interval switch
• 'Not using it' — habit/usage problem; better onboarding emails, recipe content, usage prompts
• 'Found a cheaper alternative' — competitive pressure; check pricing pages of competitors
• 'Quality issues' — operational; check supplier, batch, or fulfillment partner
• 'No longer needed' — natural lifecycle end; usually unrecoverable, accept it
• 'Other' — keep an open field, but treat any reason >15% as a signal to add a new option

Vanity metrics you can safely ignore

Most dashboards bury the important numbers under a pile of vanity metrics that exist because they're easy to compute, not because they help. Some examples worth defaulting to off:

• Total subscribers ever — a number that only goes up regardless of business health. Useless except for the company anniversary post.
• Total revenue ever — same problem. Always up, never tells you whether this month was good.
• Average subscription length — biased toward subscribers who already churned (you can't measure length on someone who's still active). Use month-N cohort retention instead.
• Most popular product — useful for the warehouse, not for the business model. Watch share-of-MRR per product instead.
• Subscriber count by day-of-week-they-signed-up — interesting trivia, never actionable.

From dashboard to decision: what actually changes what

A dashboard is only useful if it leads to a decision. Here's the rough mapping from a moved metric to the action that addresses it. Treat this as a starting heuristic, not a recipe — every store has its own quirks.

• Voluntary churn rising → improve cancel save flow, check cancel reasons, run a survey of recent cancellers
• Involuntary churn rising → review dunning retry schedule, enable account-updater service, audit gateway fraud rules
• Month-1 retention dropping → check widget claims vs product reality, audit onboarding emails, look for fulfillment delays on first orders
• Month-3 retention dropping → product-quality or cadence-mismatch issue; survey cohort directly
• AOV dropping → cross-sells underperforming, or product mix shifting to cheaper SKUs; re-test portal cross-sells (see our A/B testing guide)
• New MRR dropping → acquisition problem, not a retention problem; check ad spend, conversion rate on subscription PDPs, widget visibility
• Net new flat with both ends rising → leaky bucket; growth is hiding decay; fix retention before scaling acquisition

The biggest dashboard mistake is reacting to a single week. Subscription metrics have weekly seasonality (renewal-day concentration), monthly seasonality (the 1st spike), and quarterly seasonality (holiday gift subscriptions). Don't react until you've seen three data points in the same direction.