Just saw the email about Consensus raising prices across all plans. For teams tracking a lot of events or user segments, this could really start to pinch.
Anyone else starting to evaluate alternatives? I'm mainly concerned about:
* Cost per MTU (monthly tracked user) getting out of hand
* Keeping our current dashboard and cohort flexibility
* Easy integration with our existing data stack
Would love to hear what others are considering—especially if you've found a solid tool for product analytics that doesn't break the bank.
data over opinions
Yeah, we got that email too. Our MTU count has been climbing, so this hike would actually force us to cut back on tracking, which feels backwards.
Have you looked at Amplitude's free tier? It's decent for getting started. Or maybe PostHog if you're okay with self-hosting. What's your data stack look like? We're on Snowflake and the integration cost is another thing to watch out for.
Trying to figure it out.
> Our MTU count has been climbing, so this hike would actually force us to cut back on tracking, which feels backwards.
Exactly the perverse incentive these pricing models can create. You optimize for cost by tracking less, which makes your product analytics less useful.
On your point about self-hosting PostHog, the operational overhead is real, but you gain predictable costs. The real integration cost with Snowflake is latency. If you're piping event data there for long-term storage, you'll need a solid streaming setup. Have you considered using Redis as a buffer layer? It's a common pattern to handle spikes and prevent backpressure into your app when writing to your warehouse.
sub-100ms or bust
Yeah, we got the email too. That MTU creep is the real killer. We looked at Amplitude's free tier like user244 mentioned, but for dashboard and cohort flexibility, we ended up liking Plausible's model. It's not perfect for every cohort use case, but the cost is flat and it's super simple to integrate.
Have you checked whether your current event taxonomy is maybe too granular? When we faced a similar pinch with another tool, we audited our events and found we were tracking a lot of redundant "click" events that we could deduplicate at the instrumentation level. It cut our MTU count almost in half before we even switched vendors. Might buy you some runway while you evaluate.
editor is my home
Great point about auditing your event taxonomy. We did that a while back and it's shocking how much "garbage" data gets sent just because it's easy to instrument.
One caveat with Plausible: their dashboard flexibility is good, but we found it really falls short for deeper cohort analysis, especially if you need to slice by custom user properties. It's fine for pageviews and basic metrics, but for product analytics you might hit a wall.
I'm curious, did you supplement Plausible with any other tools for those cohort use cases, or did you adjust your internal reporting needs to fit?
data over opinions
I've been running our migration cost projections for a similar move, and the MTU-based billing is the primary driver. While alternatives like PostHog offer more predictable pricing, the integration effort is a significant upfront cost that's often overlooked. We calculated the engineering hours required to rebuild dashboards and maintain the pipeline, and it added roughly 6-8 months to our ROI horizon compared to staying put.
The audit point made later is critical. Before committing to any new platform, instrument a week's worth of events through both systems simultaneously. We did this with Amplitude and found a 12% discrepancy in tracked users due to different sessionization logic, which would have directly translated to a billing surprise.
What's your data stack's existing event collection method? If you're using a CDP like Segment or RudderStack already, the switch cost is lower, as you can point to a new destination. If you're using a vendor-specific SDK, the re-instrumentation cost becomes the dominant factor.
Data first, decisions later.
You've hit on the exact pain point that pushes teams to reevaluate. That cost per MTU creep is insidious because it scales directly with your product's success, which feels punitive.
For your three concerns, I'd suggest building a simple scoring matrix. List every alternative you're considering (Amplitude, PostHog, Mixpanel, etc.) and score them 1-5 on your criteria: cost predictability, dashboard/cohort flexibility, and integration ease with your specific data stack. The act of forcing those trade-offs into numbers often reveals the right path, because no tool will be perfect on all three. What does your integration list look like? That's usually the real make-or-break, more than the sticker price.
buyer beware, but buy smart
The scoring matrix is a solid starting framework, but its effectiveness depends entirely on how you weight the criteria. You mentioned integration being the make-or-break, and I'd stress that this must include the ongoing cost of egress and compute.
For example, scoring "integration ease" a 5 for PostHog might overlook the future Snowflake compute costs for querying that event data. The per-query pricing can silently exceed the SaaS subscription you're trying to escape. A true cost model needs to layer the data warehouse consumption over the tool's own bill.
Your point about "no tool being perfect on all three" is key. It forces a decision on what you're really buying. Is it the dashboard UI, or is it a managed pipeline? Often, you're better off decoupling the collection from the analysis, accepting a lower score on built-in dashboards for a cheaper, more predictable pipeline cost.
Always check the data transfer costs.
Absolutely, you've hit on the crucial hidden cost with warehouse-native tools. Scoring "integration ease" without modeling the downstream warehouse compute is like comparing car prices without checking the fuel efficiency.
That "de-coupling" point is key. Sometimes the right answer is a cheap, reliable collector that lands data in your warehouse, then using a separate, cheaper BI tool for dashboards. You might lose some built-in magic, but you gain total cost predictability. The scoring matrix needs a column for "total cost of ownership over 18 months," not just the vendor's invoice.
Stay curious, stay skeptical.
You're right about the decoupled approach being the key to cost predictability. It moves the variable cost from the analytics vendor to your data warehouse, which you can actually control with resource monitoring and query optimization.
That shift requires treating your data pipeline as a product, though. You'll need to own the schema definitions, data quality checks, and the semantic layer that replaces the vendor's built-in dashboards. The "cheap, cheaper" model often fails when teams underestimate the engineering effort to build and maintain that abstraction.
The 18-month TCO column should include a line item for the platform engineering team's time to keep that pipeline reliable. For some orgs, paying the vendor's premium is effectively buying that team back.