Amplitude

Most teams looking at Amplitude vs Heap are not actually choosing between two analytics platforms — they are choosing between two postures toward instrumentation. Amplitude assumes you will name your events; Heap assumes you would rather not. Pick the posture your team can actually maintain, not the one the demo made you envy.

Typical fit: who each tool is built for

Typical Amplitude customer

Series C+ PLG or multi-product B2B SaaS with a named analytics or RevOps owner, an experimentation program with at least one running test per squad per quarter, weekly cohort syncs into Salesforce or HubSpot, and procurement that requires SSO, SCIM, and audit logs. Budget band is the $30K–$200K+/yr range — analytics is a line item that survived a budget review.

Typical Heap customer

Series A–B PLG with one to three PMs sharing instrumentation duty and no analytics team. Greenfield product surface where event taxonomy was never written down. CS using replay and retroactive funnels for QBR prep, not yet syncing cohorts into CRM at scale. Growth running ad-hoc "what changed yesterday" investigations weekly. Budget is low five-figures climbing toward low six-figures, with Contentsquare bundling on the horizon at renewal.

Neither if you're…

• Buying analytics to "fix the CRM" — wrong category; see HubSpot or Salesforce.
• An indie or AI-native team wanting analytics + replay + flags + LLM obs bundled — PostHog is cheaper.
• A team that needs in-app guides and a feedback portal alongside analytics — see Pendo.

When Amplitude wins

Amplitude wins when governance is the binding constraint. Three concrete patterns:

• Multi-product experimentation. When three squads each run feature flags and web experiments against the same user, Amplitude's experimentation layer keeps the math honest. Heap's A/B tooling is lighter and harder to govern at multi-squad scale.
• Audience syncs into CRM and engagement tools with a clean writeback contract. RevOps defines a PQL cohort, syncs it to Salesforce or Customer.io, and the AE who picks up the lead can trust the definition because the taxonomy lives in one place. See the RevOps lead scoring playbook for the system view: input = product events, AI step = Global Agent drafts the cohort, human review = RevOps approves before sync, writeback = Salesforce custom field, metric = PQL→Opp rate.
• AI agents on governed data. Global Agent and the specialized dashboard/replay/feedback agents become useful once event names are stable. On clean data they cut the analyst queue meaningfully; on messy data they produce confident-wrong cohorts faster than humans would.

When Heap wins

Heap wins when instrumentation discipline is the binding constraint — usually because it doesn't exist yet.

• Retroactive event definition. A CSM asks "how many trial accounts hit the billing page twice last week" on a Tuesday; in Amplitude that question costs a sprint of instrumentation. In Heap, autocapture has already recorded the clicks, and the event can be defined Tuesday afternoon. For Series A–B teams shipping weekly, that gap is the product.
• Sense AI on autocaptured streams. Anomaly detection without prompt engineering — useful for CS standups asking "what changed yesterday?" The human review step still matters (flag → cause validation before any CRM alert), but the input axis is essentially free.
• Replay tied to autocapture in one workflow. Click → replay deep link → Zendesk ticket without a separate replay ingestion path. CS teams use this for adoption-risk detection — see customer success risk detection.

When you need both

Rare but real. The pattern: Heap on a new product surface for fast iteration, Amplitude on the mature product where experimentation and governance matter. Both feed the same warehouse, and a reverse-ETL layer (Hightouch) handles the cohort sync to CRM so the writeback contract is consistent regardless of which analytics tool owned the input. Most teams don't need this — pick one, ship, revisit at the next stage. If you do run both, make one team own each tool's taxonomy; shared ownership rots both.

Pricing and per-account math

Amplitude Starter is free up to 10K MTUs and 2M events/mo; Plus starts at $49/mo annual and caps at 300K MTUs or 25M events. Growth and Enterprise are quote-based — market reports typically land $30K–$200K+/yr at scale.^[1] Amplitude also runs a Startup Scholarship: one year of Growth free if you're under $10M funding and under 20 employees.^[1]

Heap's free tier carries session-volume and feature limits; Growth and Pro are custom per session volume.^[2] Mid-market Heap contracts typically sit below Amplitude Enterprise but above PostHog pay-as-you-go at the same volume.^[2] The 2024 Contentsquare acquisition is reshaping packaging through 2026 — confirm what's in your tier at renewal.^[3]

Per-account math sanity check (illustrative, not invented dollars): if you have 50K MAU and run weekly cohort syncs into Salesforce, Amplitude Plus likely caps out and pushes you to Growth — model the Growth quote against a Heap mid-market quote at the same MAU before committing. Below 10K MTUs, Amplitude Starter is genuinely free; Heap's free tier hits feature limits faster.

Feature overlap and gaps

Both cover product analytics, behavioral cohorts, funnels, retention, and session replay. The wedge is governance vs. autocapture.

The buying mistakes we see most

1. Buying Amplitude Enterprise while still arguing over what "Signed up" means. Cost: $60K+/yr for mutual exclusion and SCIM the team can't use. Fix: stay on Plus until taxonomy ownership is named in a doc.
2. Picking Heap because the demo answered a question fast, then syncing autocaptured noise into Salesforce. Cost: AEs lose trust in product-qualified leads within one quarter; the cohort gets muted. Fix: schedule a taxonomy + identity audit before any CRM writeback, and gate the sync on a 10-account manual review.
3. Choosing on AI demos rather than data readiness. Both Global Agent and Sense AI degrade on duplicate users, orphan events, and weak taxonomy. Cost: confident-wrong cohorts shipped to CS, wrong customers contacted. Fix: run the week-1 test below before automating any agent output.

What to test in week 1

Amplitude one-week test: pick one revenue-tied metric (e.g., "activated within 7 days"). Document event definitions in a shared doc. Build the cohort in Amplitude. Manually review 10 accounts in the cohort against CRM records. Sync a test audience to CRM or post a weekly Slack summary — human approved. Measure: % of cohort accounts where CRM activity matched product truth, and time saved vs. prior manual pull.

Heap one-week test: pick one PQL or expansion signal ("opened pricing page twice + viewed billing settings in 14 days"). Use autocapture to retroactively build the cohort from the last 30 days. Manually review 10 accounts against CRM records — does the click pattern reflect real interest, or noise? If noise exceeds 30%, do not sync to CRM; fix definitions and identity resolution first.

If either week-1 test fails the manual review step, the AI agents are not the bottleneck — data readiness is.

Migration and coexistence

Heap → Amplitude: the painful migration. Heap's autocaptured event names rarely map cleanly to Amplitude's named-event model; expect a 60–90 day instrumentation project, not an export. Common pattern: run both in parallel for one quarter, instrument the top 20 events natively in Amplitude, then deprecate Heap event-by-event. Cohort definitions need to be re-authored, not copied.

Amplitude → Heap: rarer (downstream simplification), usually driven by a Contentsquare bundle deal at renewal. The autocapture history starts fresh — you cannot retroactively recover Amplitude's historical events into Heap.

Coexistence: Heap on new product surface, Amplitude on mature product, both feeding a warehouse (Snowflake/BigQuery), with Hightouch handling cohort syncs to CRM. Works when one team owns each tool's taxonomy; rots when shared.

FAQ

Does autocapture eliminate the need for an event taxonomy? No. It defers the cost — you pay in cleanup by month six instead of in instrumentation on day one. Plan for the cleanup pass.

Can Amplitude do autocapture? Amplitude has autotrack capabilities, but the platform is instrumentation-first by design. If retroactive event definition is the core need, Heap is built around it; Amplitude is built around named events you commit to.

Does Sense AI replace Amplitude's Global Agent? Different jobs. Sense AI flags anomalies on autocaptured streams (read-mostly). Global Agent answers natural-language questions, drafts charts, and connects to MCP for cross-tool workflows. Both still require human review before any writeback to CRM.

How does Heap's Contentsquare integration affect the comparison? Heap's packaging and feature parity are settling through 2026 post-acquisition.^[3] If you're evaluating Heap on a multi-year contract, confirm what's bundled in your tier and what migrates to Contentsquare's broader experience analytics surface.

What if we already use PostHog? Different decision tree — see PostHog vs Amplitude. PostHog and Heap both target Series A–B but optimize differently: PostHog bundles flags + replay + LLM obs; Heap optimizes autocapture and retroactive analysis.