Pravix, by Datalyze

Pravix: data quality monitoring for Mixpanel and PostHog

Pravix checks the events in your Mixpanel or PostHog project on a schedule and alerts you when a number breaks. It runs nine checks, from an event that stops flowing to a property that suddenly goes null, and says where the fix is needed. Your first run is free, with no card, and so is one data audit a month.

By Ansh AgrawalUpdated

signup_completed per day, last 14 days
Expected rangeToday: 10
  • Volumesignup_completed is at 10 today, against about 200 a day
  • Null %referral_code is null on 40% of signup_completed, up from under 5%
7 other checks passed

Teams Datalyze has worked with

  • CRED logo
  • AWeber logo
  • Silverfort logo
  • Final Round AI logo
  • Pixis logo
  • Coursebox logo
  • Uplers logo
  • Delightree logo
  • AlgoTest logo
  • Magma logo
  • TermPlus logo
  • Kryptos logo
  • Superhote logo
  • Wellness Coach logo
  • Inferless logo
  • Skip logo
  • Copyfy logo
  • Foriio logo
  • Fincart logo
  • Sarvam AI logo
  • Kruzee logo
  • TrueFoundry logo
  • Buildern logo
  • Yearn logo
What Pravix checks

What does Pravix alert you about?

Pravix runs nine checks on your events and user profiles. Each one checks against thresholds you set, so a drop, a gap or a shift shows up on the next run, not weeks later when a dashboard looks wrong.

CheckWhat it meansExample
Event count has spiked or droppedAn event's volume changes sharply in a short period.Signup Completed usually fires about 200 times a day and suddenly drops to 10: a backend issue or broken tracking code.
Event has stopped flowingA key event is no longer reaching your analytics tool.Add to Cart stops recording overnight. Without an alert, you'd find out when conversion looks wrong.
Event property has stopped flowingA property on an event is missing from new data.Purchase Completed used to send payment_method. New events don't have it.
Event property null % has gone upThe property is there, but too many events have it empty.Under 5% of Signup Completed events normally have referral_code null. At 40%, you've lost attribution data.
Event property distribution has shiftedThe mix of values for a property changes sharply.70% of purchases used credit_card; this week it's 10% and "other" has surged. The payment integration may be broken.
User property has stopped flowingA property on user profiles is no longer updated.country stops being set, so you can't segment users by geography.
User property null % has gone upMore profiles are missing a value for a property.2% of users normally have plan_type null. Now 25% do, so subscription analytics are wrong.
User property distribution has shiftedThe spread of a property across users changes sharply.80% of users had device = Android; now it's 40% with no change in traffic sources. That's a collection problem.
User profile count has spiked or droppedThe number of new profiles rises or falls unexpectedly.About 500 new profiles a day, then 5,000 today: bot signups, or duplicate profiles being created.
How it works

How do you set up data alerts in Pravix?

Setup takes minutes. After that, Pravix runs on its own schedule and you only hear from it when something breaks.

Free data audit

Is your analytics data telling the truth?

The Pravix data audit is a one-off check of your Mixpanel or PostHog project for the problems that quietly break dashboards: double-firing events, missing properties, one person split across many IDs, and drift from your tracking plan. It runs 13 checks, and every account gets one free audit a month.

Run your free audit

Free, read-only access, no credit card.

How does the data audit work?

  1. Create a free account

    Sign up with your work email. No credit card.

  2. Connect your project

    A Mixpanel service account or a PostHog personal API key. Read-only.

  3. Get your report

    Every issue with the affected events, a sample of the bad rows, and a CSV to hand to engineering.

What does the data audit check?

13 checks across four areas. Each reports the affected events and properties, how widespread the problem is, and example rows.

AreaCheckWhat it findsNeeds a tracking plan?
Event qualityDuplicate or double-firing eventsThe same event recorded more than once, which inflates every count.No
Event qualityPageviews not firingSessions that never record a pageview.No
Event qualitySingle-pageview sessionsSessions stuck at one pageview, a sign of a single-page app navigation gap.No
Property qualityNull or missing property valuesProperties absent, null or empty against the tracking plan.Yes
Property qualitySingle-value propertiesA property hardcoded to one value, or dominated by one.No
Property qualityProperty type inconsistencyThe same property arriving as more than one type.No
IdentityMissing pre-login activityNew users whose anonymous events never merged into the identified user.No
IdentityID explosionOne real person split across many distinct_ids.No
IdentityMalformed ID valuesEvents carrying a malformed or placeholder distinct_id.No
IdentityShared or default IDsOne ID standing in for many users, so different people collapse into one.No
Tracking planMissing planned eventsEvents in the tracking plan that never fire.Yes
Tracking planRogue or untracked eventsEvents in your data that aren't in the tracking plan.Optional
Tracking planType mismatch against the planA property arriving as a different type than the plan declares.Yes

Do you need a tracking plan for the audit?

No, but bring one if you have it. A plan is how the audit finds what's missing, not just what's broken. You choose before the audit runs.

With a tracking plan

Upload a CSV or spreadsheet in any layout and Pravix converts it. All 13 checks run, including:

  • Planned events that never fire
  • Required properties that are missing or empty
  • Properties arriving as the wrong type
  • Events in your data that were never planned

Without a plan

Start straight away. The 10 checks that only need your data still run:

  • Duplicate and double-firing events
  • Hardcoded or single-value properties
  • Identity and ID problems
  • Pageview gaps and inconsistent property types

What does the audit report look like?

A plain-language list of issues with the evidence attached. This is sample data; yours shows your own events.

Sample data
  • Duplicatescheckout_completed is double-firing
  • Propertysignup_completed is missing plan_tier on 100% of events
  • IdentityOne person is split across 15 IDs

After the audit, fix the issues with the CSVs in the report. To hear the moment tracking breaks again, turn on the scheduled checks above.

Why Pravix

Why do teams miss broken analytics data?

Because nobody can check every event by hand. Ansh Agrawal, who founded Datalyze and built Pravix, is a Mixpanel Certified Partner who sets up analytics for startups. In that work, the same problems kept turning up: events that stopped firing, properties that went missing, and distributions that shifted without anyone noticing.

None of them is hard to fix, but each is slow to find by hand. Pravix exists so a team hears about a broken event when it breaks, instead of spending days chasing hidden data problems.

If the problem is deeper than one broken event, a Datalyze analytics audit goes through the whole setup. For why teams stop trusting their data in the first place, read why nobody trusts your data, and how to fix it. Pravix is one of three Datalyze products, next to the Tracking Agent and the Datalyze MCP.

Pricing

How much does Pravix cost?

Your first run is free, with every check and no card. After that, Pro is $49 a month, billed by usage: 100k rows a month are included and each row after that is $0.001.

Free trial

Freeyour first run
  • One full run, no restrictions
  • Every detector
  • Share the results with your team
  • No credit card
Start for free

Pro

$49/ month
  • Unlimited runs
  • 100k rows included each month
  • $0.001 per additional row
  • Automated alerts for anomalies
  • All nine detectors
  • Unlimited data audits
Start with a free run

Without Pro, every account still gets one free data audit a month.

FAQ

Questions about Pravix

Which analytics tools does Pravix work with?

Mixpanel and PostHog. Connect Mixpanel with a service account or PostHog with a personal API key, and Pravix checks data quality in either.

Does Pravix store my analytics data?

No. Scheduled checks delete their data as soon as each run finishes. A data audit exports events into a temporary workspace that deletes itself after two days, and the report keeps only the few example rows needed to show each issue.

How much does Pravix cost?

Your first run is free with no restrictions, and every account gets one free data audit a month. Pro is $49 a month with unlimited runs and 100k rows a month included, then $0.001 per additional row. No credit card is needed to start.

How long does setup take?

Minutes. Connect your data source, pick the checks and thresholds you care about, and set a schedule: daily, weekly or custom. A data audit finishes in a couple of minutes on a small project and can take up to half an hour on a high-volume one.

Can I run the data audit without a tracking plan?

Yes. Without a plan, the 10 checks that need only your data run: duplicates, hardcoded values, identity problems, pageview gaps and inconsistent types. With a plan in any CSV or spreadsheet layout, the audit also finds planned events that never fire, missing required properties and type mismatches.

Pravix

Find out what's broken before your next board deck does

Your first run is free, read-only, and needs no credit card.

Get your first run free