Amplitude vs Heap: A Practical SaaS Buyer Guide 2026

What you are actually choosing

Product analytics is the record of what people do inside a software product, used to see who activates, who returns, and where a journey stops. A product lead who typed amplitude vs heap is close to a decision, and the decision is a meter, a history window, and a set of controls that appear only on some plans. It is not a logo preference.

TL;DR: choose Amplitude when you need a published event allowance, session replay with a stated monthly cap, and native experiment limits. Choose Heap when you need clicks stored before anyone names them, and you can accept a sales quote once sessions pass the entry tier. Heap is not a standalone company. Acquisition close: December 7, 2023 according to Contentsquare (2023). That notice also says Contentsquare acquired Hotjar in 2021 and serves more than 1.3 million websites. The Heap name and the Heap plan page are still public, so this comparison is Heap as sold now, inside Contentsquare, against Amplitude, which still publishes plans under its own name.

A 2023 forecast is worth reading only as context, not as a vote. PM analytics forecast: 70% by 2028 according to Contentsquare (2023), quoting an April 11, 2023 Gartner note that put the 2023 baseline at 19%. That note does not say which product those managers will buy, and it does not replace a count of your own events and sessions.

The category decision comes first. If the team needs governed product behavior, replay, and experiments in one contract, stay on this page. If the team needs a warehouse metric store and will not open either charting UI, neither product is the system of record, and the rest of this guide will not fix that.

Key Takeaways

• Heap sits inside Contentsquare. The product name remains, and the plan page is still a Heap page, so older articles that treat the two as unrelated rivals are out of date.

• The meters do not match. Amplitude bills and caps on events. Heap bills and caps on sessions. A plan that looks lighter can be the costly one after you translate your own traffic.

• Amplitude publishes a zero-dollar start on its self-serve paid tier and keeps the larger tiers on a sales quote. Heap's paid tiers do not publish a dollar rate.

• Session replay is counted on every Amplitude plan and is listed as an add-on on Heap's upper tiers. Do not assume the entry cards include the same qualitative tools.

• Autocapture on either product does not appoint an owner. Someone still has to decide which action means activated, invited, or paid.

• Public recommendation rates sit close together. Use them as a sanity check, then decide on the meter, the history window, SSO, and replay.

Who this is for

This guide is for a product lead at a B2B SaaS company who will sign, or strongly shape, the analytics choice in this quarter. You have a web app, likely a mobile app as well, and a handful of journeys that matter: invite sent, seat activated, key feature used, renewal risk visible to a success manager. You do not need a lab report. You need to know which public limits will break first.

Amplitude recommend rate: 90% according to SelectHub (2026), from 2,735 reviews, beside Heap at 87% from 1,528 reviews, on a page last updated September 28, 2026. Those rates are too close to crown a winner. They do say both products have a long review trail, so a thin feature list is not the risk. The risk is buying the wrong unit.

Use this checklist before anyone requests a quote.

• Write the billing unit in one sentence a finance partner can read.

• Count last month's events and last month's sessions from the product database, not from a slide.

• Mark whether single sign-on is required on day one.

• Mark whether replay must sit in the base plan.

• Name the person who may approve a new event name or a new label.

• Decide what the team will do as usage nears the published cap, before a vendor email arrives.

Red flags: nobody owns event names or session definitions, you require a public dollar price on every paid Heap tier before any sales conversation, or the warehouse is the only system of record and nobody will use either charting UI.

How we evaluated these tools

We weighted six buying criteria. The weights are ours, for a B2B SaaS product lead, not a score either vendor published. They sum to 100%. A tie on a marketing page does not survive a weight that is about the contract you will actually sign.

Capture model is 25% because the wrong unit makes every later chart a translation problem. If finance thinks in sessions and the contract counts events, the renewal conversation starts from a misunderstanding. Data history is 15% because retention and seasonality die when the tier keeps a short window. Session replay is 15% because a funnel drop without a way to watch the session sends the team back to tickets and guesses. Experimentation is 15% because a second testing contract is a second security review. Identity and permissions are 10% because single sign-on that arrives only on a higher tier can block a rollout even when the charts are fine. Price you can put in a budget is 20% because a zero-dollar start and a pure quote are different risks, and this reader is close enough to choosing that an unmodelable bill is a real defect.

Our read, kept separate from the published grids, is this. Heap fits the capture weight if the next quarter's questions are not in the tracking plan yet. Amplitude fits it if a named owner will keep event meanings stable and the team still wants the autocapture the pricing page includes on every plan. History favors the Amplitude self-serve paid tier on the numbers below, unless you buy Heap's add-on years. Replay favors Amplitude unless you have already priced Heap's add-on. Experimentation is a published count on Amplitude and a gap on Heap's public matrix, which is not the same thing as proof that Heap cannot test. Identity favors Heap's entry card if security will not wait for a higher Amplitude tier. Price favors Amplitude only up to the point the public page stops printing dollars. Above that, both are quote-based, and Heap is quote-based for every paid tier.

Published limits, side by side

Free event cap: 2,000,000/month according to Amplitude (2026). The same page says the Free plan needs no credit card and has no time limit, includes 10,000 monthly session replays, 1 year of data access, 10 saved charts per organization, 1 active experiment, 1 guide or survey, unlimited feature flags, and unlimited seats. Autocapture and retroactive data transformation are marked included on every plan. The Plus plan on Amplitude starts at $0, requires a credit card, includes the first 2,000,000 events, scales to 70,000,000 events and up to 700,000 monthly tracked users, keeps 10,000 monthly session replays, extends data access to 2 years, and includes 20 behavioral cohorts. Growth and Enterprise use custom event-based pricing, which we treat as quote-based. Growth lists 20,000 monthly session replays and Enterprise lists 50,000, with active experiments at 2 on both and an add-on for more. SSO is absent on Free and Plus and included on Growth and Enterprise. Webhooks are absent on Free and included from Plus upward. Group behavior analysis is not included on Free and is marked available from Plus upward, with an Accounts add-on described for Growth and Enterprise. A startup with fewer than 20 employees and under $10 million in funding can apply for one free year of Growth. The page lists a response target of 2 business days on Growth and 1 business day on Enterprise.

An MTU, on that same page, is a unique user who triggers one or more events in a calendar month. One account can therefore be one MTU and hundreds of events. That is why a seat count is a poor stand-in for the bill.

The Heap column is sourced in the pricing section below, where its citation sits once. Read the two columns as different units. A row that says 2,000,000 events next to 10,000 sessions is not a claim that one side is 200 times larger. It is a claim that you must bring your own ratio.

Pricing checked against the plan pages

Pricing checked October 10, 2026.

Entry session cap: 10,000/month according to Heap (2026). The entry tier lists 6 months of data history, SSO, and 1 project. The next tier lists 12 months of history, unlimited users and reports, and CSV export, and it tells you to install the snippet for an estimate rather than a dollar price. The Pro plan on Heap and the Premier plan on Heap are marked Custom Session Pricing, which we record as quote-based. The grid lists 3 projects on Pro and unlimited projects on Premier, 1 year of history on those tiers with add-on years, and an extension out to 3 years as an add-on. Session replay, heatmaps, and error analysis are add-ons on Pro and Premier. Warehouse sync is an add-on on Pro and included on Premier. Account analytics is listed on the Pro card. A web session ends after 30 minutes of pageview inactivity, and a mobile session ends after 5 minutes of inactivity. If you pass the monthly limit, Heap says it will not drop or throttle the data, and you must upgrade before you can view it again. The same page summarizes a Forrester study as a 40% reduction in sprint cycles for product enhancements, nearly 3,000 hours of manual tagging a year, and $1.7 million of incremental revenue over three years. That summary is the vendor's, not a result we measured.

A full cost of ownership cannot be computed from these pages once you leave the published inclusion. Amplitude does not print the dollar rate between the 2,000,000 event inclusion and the 70,000,000 event ceiling. Heap does not print a dollar rate for Growth, Pro, or Premier. Anyone who fills those cells with a remembered price is guessing. Ask for the quote in writing, attach your own volume, and keep the quote next to this table.

Two commercial edges sit outside the cells. Amplitude's startup path is one free year of Growth for companies with fewer than 20 employees and under $10 million in funding, which is a real discount only if you qualify and only for that year. Heap's commercial edge on the entry card is SSO without waiting for a higher tier, plus the promise that overage is held for viewing rather than thrown away. Neither edge is a reason to skip the volume math.

Amplitude for teams who will own the event names

Best fit: a B2B product team that will name an owner for the event list, wants replay in the base plan, and expects to run experiments without a second vendor. The Free plan is a real on-ramp if you are under 2,000,000 events a month and you do not need SSO or webhooks yet. The Plus plan on Amplitude is the self-serve step when you will pass that inclusion and you can tolerate a bill that scales without a printed per-event rate. Growth is the tier where SSO, a longer response target, more replay, and more experiments show up, and it is quote-based. Enterprise adds the tighter response target, role controls, and the higher replay allowance, and it is quote-based too.

Limitations are specific. Twenty behavioral cohorts on Plus will feel small the moment customer success, growth, and product each save their own lists. One active experiment on Free and Plus means a testing program is a Growth conversation, not a toggle. Guides and surveys start at 1 on the Free plan, so a team that also wants in-app guidance should read Pendo versus Amplitude for product-led SaaS and the shorter Pendo and Amplitude comparison before folding guides into this contract. Autocapture being included does not mean revenue events will be correct. A button rename can still split a trend if nobody maps old names to new ones. Group and account views are not a Free-plan promise.

Implementation, from the public docs rather than from a deployment we ran: send events with a required event name, and with a user or device identifier long enough for the API to keep. Deduplicate retries. Keep a human review on any rename, because a retroactive transform can repair a property and can also hide a real product change. Security review should include the EU data-center option listed on every plan, and should not assume SSO exists below Growth. Primary evidence is the pricing page and the HTTP API page cited in this guide.

Our analysis: pick Amplitude when the event list is a product asset you intend to govern, and when the included replay cap is enough for the debugging you actually do. Walk away from the Free plan if SSO is a launch blocker. Walk away from a handshake on Growth or Enterprise until the quote states the event volume, the replay add-on, and the experiment add-on as separate lines.

Heap for teams who need the clicks they did not name

Best fit: a team that will discover the question after the release, not before it. Autocapture plus later labeling is the point of the product. The entry tier is enough only while you stay inside 10,000 sessions a month and 6 months of history. The moment a B2B product has a real trial motion, that session cap is the first wall, and the paid price is quote-based. SSO on the entry card matters if your security partner will not wait. Account analytics on the Pro card matters if you sell to accounts rather than to anonymous visitors, and that card is still a quote.

Limitations are equally specific. Session replay is not in the entry feature list. It is an add-on on Pro and Premier, so a "we will just watch the session" plan is a second line item. History on the entry tier is 6 months, which cuts off annual renewal behavior. The view lock is the sharp edge: the page says data is not dropped when you pass the cap, and it also says you need to upgrade before you can look at it. A launch week that spikes sessions can freeze the charts on the week you most want them. Web and mobile do not share a clock. Thirty minutes of pageview inactivity ends a web session. Five minutes of inactivity ends a mobile session. Mixing those into one "session" KPI will lie.

Event name limit: 255 characters according to Heap (2025). On that page, heap.track sends a custom event the autocapture did not collect, property keys must be fewer than 512 characters, and values must be fewer than 1,024 characters. heap.identify attaches an identity that must be fewer than 255 characters, and two users in one environment cannot share it. Implementation is the snippet first, then a short list of server-truth events such as invoice paid or seat provisioned, sent through heap.track because the browser will not know them. A person should approve labels. A renamed button that gets a new label without a note will look like a new feature. Warehouse export waits for the Pro add-on or the Premier inclusion, so a data team that wants the raw stream on day one is not buying the entry tier. Primary evidence is the pricing page and this API reference.

Our analysis: pick Heap when retroactive labeling is the reason you are here, and when you have budget process for a session quote plus a possible replay add-on. Do not pick it from the entry tier if 10,000 sessions is already a normal month. Do not treat the Forrester summary on the pricing page as your savings. It is their published study summary, and your tagging hours are your own.

An illustrative month in each meter

The figures in the assumption rows are a teaching example, not a customer, a survey, or a test. The caps are the published ones cited above. The point is to show that both walls can arrive in the same month for a modest B2B account base.

Take 8,000 accounts in a month, each sending 400 events, which is 3,200,000 events, each carrying an event_type, with a repeated insert_id on the same device_id ignored for 7 days so a retry does not double-count a signup. The Free allowance is 2,000,000 events, so this illustration is 1,200,000 events over that cap, which is 60% over the inclusion. The Plus plan on Amplitude starts at $0 and lists a ceiling of 70,000,000 events, so 3,200,000 events sits under that ceiling and outside the Free cap, with no public per-event dollar rate for the gap. HTTP V2 ceiling: 50,000 events/second according to Amplitude (2026), and the same page tells Free-plan senders to stay within 1,000 events per second and 100 batches per second, with a suggestion of no more than 10 events per batch. The same 8,000 accounts at 3 sessions each is 24,000 sessions, which is 14,000 over the 10,000 session entry cap, or 2.4 times that cap, so the Heap paid tier for this illustration is quote-based.

Read the result this way. The illustration does not fit Amplitude's Free plan and does fit inside the published Plus ceiling, at an unknown dollar amount above $0. The same illustration does not fit Heap's entry session cap, so the next step is a quote, not a self-serve checkout. If your real ratio is 1 session to 20 events, or 1 session to 200 events, redo the arithmetic with your ratio. Do not reuse 400 and 3 as if they were industry averages. They are not.

Handoffs above the chart

The pain after the purchase is rarely the chart. It is the moment volume nears a cap, a webhook or a CSV lands, and someone has to decide whether to rename an event, pause a sync, or ask for a new tier. That decision should not be an accident inside a charting tool.

A proposed US Tech Automations workflow is a configurable design, not a live deployment and not a measured outcome. The trigger is a scheduled export of event counts, or a webhook where the plan includes webhooks, which the Amplitude pricing page includes from Plus upward and not on Free. The action compares the count with the 2,000,000 event Free cap and the 70,000,000 event Plus ceiling stored as thresholds, and it lists the event_type values that grew fastest. The output is a review note. Prerequisites are export or API access and a place to store the note. A person approves any rename and any plan change. Nothing in that path sends mail to a customer.

US Tech Automations can be configured for the Heap side in the same spirit. The trigger is the vendor email that the monthly session limit was passed, or a CSV export, which the Heap pricing page lists from the Growth tier upward. The action sets the session count next to the 10,000 session entry cap and pauses a warehouse sync. The output is an approval record a finance partner can read. Prerequisites are that email or a plan that includes CSV export. A person approves the tier change and any Heap Connect sync before it runs. This is a proposed design. It is not a report of a current customer.

Zapier, Make, and n8n can support run histories, retries, error branches, and audit evidence when a team configures them. That is a fair alternative, and for a single low-volume alert it is often the better one. The buyer still owns observability, idempotency, escalation, access control, and maintenance. A retry that ignores Amplitude's insert_id can write the same signup into the review note twice. A scenario that nobody watches will fail quietly on the week the cap matters. A proposed US Tech Automations design can keep the threshold check, the pause on the sync, and the approval record in one configured path, with the same prerequisites and the same human approval before a tier change or a rename. The difference is who maintains the path. It is not a claim of a better chart, and it is not a promise that the vendor bill goes down.

When NOT to use US Tech Automations

Skip this layer when the only job is a chart inside Amplitude or Heap and no file, ticket, or billing decision has to leave that tool. Skip it when one existing Zapier, Make, or n8n scenario already tells a named owner that a CSV arrived, and the volume is low enough that a person sees failures without a second system. Skip it when the current plan has no export and no webhook, because there is nothing for any orchestrator to read until the plan includes one. A simpler tool wins in those three cases, and adding a path on top would only create another thing to maintain.

Writing the decision note for finance is a separate chore from the analytics contract. A roundup of SaaS writing tools covers that chore. It does not change the meter you buy.

Common mistakes that skew this choice

Editor scores: 8.8 and 8.7 according to iTechGuides (2026), which discloses that it may earn a commission on some links, with Amplitude at 8.8 out of 10 and Heap at 8.7 out of 10 in an October 2026 update. A tenth of a point is not a buying reason. Use the score only to see that a public rubric also calls the race close.

The mistakes that actually move money are plainer.

• Comparing a session to an event as if they were the same object. They are not, and Heap's own page uses a 30 minute web clock and a 5 minute mobile clock.

• Assuming Amplitude is manual-only. The pricing page marks autocapture included on every plan. You still need an owner.

• Assuming Heap's entry tier includes session replay. The plan cards put replay on the Pro and Premier add-on line.

• Typing a dollar figure for the Growth plan on Amplitude, the Enterprise plan on Amplitude, or any paid Heap tier. Those prices are not on the pages checked for this guide. They are quote-based.

• Ignoring the view lock. Passing Heap's session cap is not only a billing event. The page says you upgrade before you can view the data again.

• Treating SSO as universal. Heap lists it on the entry card. Amplitude lists it on Growth and Enterprise.

• Taking the Forrester summary, or the 70% forecast from 2023, as your business case. Both are someone else's numbers.

Glossary

• Event. One tracked action, such as a click or a signup. Amplitude's Free plan is capped at 2,000,000 of these per month.

• Session. A stretch of activity from one user. On Heap's page, web ends after 30 minutes without a pageview, and mobile ends after 5 minutes without activity.

• MTU. A monthly tracked user: one person who fires at least one event in a calendar month. Amplitude's Plus row lists up to 700,000 of them beside the 70,000,000 event ceiling.

• Autocapture. Collection of interactions without a hand-written tag for each one. Both pricing pages claim a form of it. It does not decide which interactions matter.

• Cohort. A saved set of users defined by behavior. The Plus plan on Amplitude includes 20 of them per organization.

• Session replay. A recording of a visit, used to see what the chart cannot. Amplitude counts a monthly allowance. Heap lists it as an add-on on Pro and Premier.

• insert_id. Amplitude's dedupe key. The HTTP V2 page says a repeat on the same device is ignored for 7 days.

• heap.track. Heap's call for a custom event autocapture did not collect. The event name is limited to 255 characters.

Questions before you commit

Is Heap still an independent product?

No. Heap is part of Contentsquare, and the plan page is still published under the Heap name. Treat older comparisons that ignore the 2023 close as stale, and confirm in the quote whether you are buying Heap product analytics, a wider Contentsquare bundle, or both.

Does Amplitude publish a dollar price for every plan?

No. The Plus plan on Amplitude starts at $0, with the first 2,000,000 events included, and the Growth and Enterprise plans are quote-based. There is no public per-event rate on the pricing page for the span up to 70,000,000 events, so a budget above the inclusion needs a written estimate from the vendor.

What happens when Heap sessions pass the plan limit?

You do not lose the captured data, and you do lose the ability to view it until you upgrade. The pricing page says Heap will not drop or throttle, and it will email you to move to the right tier before the charts open again. Plan for that freeze before a launch week, not after.

Which entry tier includes session replay?

Amplitude includes session replay on every plan, with 10,000 monthly replays on Free and Plus. Heap does not list session replay in the entry feature set. It lists session replay as an add-on on Pro and Premier, so budget it as a separate line if watching sessions is a requirement.

Can Zapier, Make, or n8n cover the handoff?

Yes, when you configure run history, retries, error branches, and an audit trail, and when your team accepts ownership of idempotency, access, escalation, and upkeep. That stack is enough for a single low-volume notice. It gets fragile when the same path must pause a sync, compare a cap, and wait for a person before a tier change.

Who should walk away from both?

Walk away if you will not name an owner for events or sessions, or if a warehouse is the only place metrics will live. Also walk away from Heap's paid tiers if you cannot accept a quote, and from Amplitude's Free plan if you need SSO, webhooks, or more than 2,000,000 events in a normal month. In those cases the missing piece is a different buying process, not another chart.

Do review scores break the tie?

No. SelectHub's compiled recommendation rates are 90% for Amplitude from 2,735 reviews and 87% for Heap from 1,528 reviews, and an editor rubric that discloses commissions scores them 8.8 and 8.7. Use the gap only as a sign that both have users. Then return to your session count, your event count, and the tier that actually includes SSO and replay.

The fork, in one pass

Choose Amplitude when a named owner will keep event_type meanings stable, you want replay inside the base allowance, and you can live with a $0 start that becomes a quote once you need SSO, more than 20 cohorts, or more than the self-serve experiment count. The Free plan fits under 2,000,000 events a month with no credit card. The Plus plan is the published bridge up to 70,000,000 events, without a printed rate for the steps in between. Growth and Enterprise are quote-based, and the startup year of Growth applies only if you are under the published employee and funding lines.

Choose Heap when the value is labeling behavior after it was captured, you can run a sales conversation for anything past 10,000 sessions, and you have checked the 6 month history, the 30 minute and 5 minute session clocks, and the replay add-on. Do not buy the entry tier as if it were a smaller copy of Premier. Warehouse sync, account analytics, and replay sit higher up, on quote-based plans.

Choose neither as your only system if the real argument is a warehouse model plus a support inbox. Bring last month's events and last month's sessions to whichever quote you request. Pricing checked October 10, 2026, and any dollar figure that was not on those plan pages is not in this guide.

If a person still has to approve the export, the cap review, or the tier change after the chart exists, see how US Tech Automations configures this as a proposed path with an export prerequisite and a human approval, not as a replacement for Amplitude or Heap.