.png)
Frameworks, core principles and top case studies for SaaS pricing, learnt and refined over 28+ years of SaaS-monetization experience.
Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.
Using Segment for Effective SaaS Pricing Data Collection
Pricing teams rarely suffer from a lack of data. They suffer from data that answers the wrong question.
A typical SaaS company can count page views, logins, clicks, sessions, API calls and feature use. Yet ask a harder pricing question - whether a 40-seat customer receives twice the value of a 20-seat customer, whether an automated workflow is a better meter than a user, or whether a package boundary suppresses expansion - and the analytics stack often becomes surprisingly weak. Product events describe what happened inside the product. Pricing requires those events to be tied to the account, its package, its contract, its usage, and ultimately the money it pays.
Twilio Segment is unusually well placed to solve the collection side of that problem. Its Track, Identify and Group calls can standardise behaviour across web, mobile and server sources; its warehouse connections can preserve those events for account-level analysis; and Protocols can enforce a tracking plan before bad data spreads downstream. As of 13 August 2026, however, Segment itself still prices its Connections product primarily around Monthly Tracked Users, or MTUs, even though the infrastructure processes events and the value of the product often grows with data volume rather than the number of distinct people being tracked.
Monetizely's position is that Segment should be the governed collection layer for SaaS pricing evidence, not the billing ledger and not the final analytical source of truth. SaaS companies should organise Segment around account identity and a small set of economically meaningful events, reconcile those events with contracts and invoices in the warehouse, and only then use the evidence to change packaging, metrics or rates.
Pricing evidence becomes useful only when events are tied to the account
Monetizely's 5-Step Pricing Framework begins with Goals and Segmentation, where the company defines what pricing needs to achieve and which customers it intends to serve. Packaging decides which capabilities sit together. Pricing Metric determines what unit makes the bill grow, while Rate Setting determines how much to charge for that unit. Operationalization makes the design work in quoting, entitlements, metering, billing, revenue recognition and reporting. The sequence, developed in Monetizing Agentic AI, matters for Segment because telemetry gathered without a pricing objective can become a large collection of events with little commercial use.
The pricing-metric step is where Segment becomes especially important. A team considering per-seat pricing needs to know whether value grows with active users. A company considering usage pricing needs to know which product action best tracks customer value. An outcome model requires evidence linking product activity to an observable result. Segment can collect each signal, but only if the tracking plan was built to distinguish them.
For B2B SaaS, the account must sit above the user. Segment's current Track specification supports a groupId specifically for attributing activity to a company or account in B2B use cases, while its warehouse schema stores group_id, user_id, event names and timestamps that can be queried together. Segment also recommends received_at for most time-based warehouse queries because client clocks can make other timestamps less reliable.
A pricing tracking plan therefore needs fewer generic engagement events and more events that reveal how the economic relationship is changing.
| Pricing question | Segment record to capture | Properties that matter | What it can tell the pricing team |
|---|---|---|---|
| Is seat pricing scaling with value? | Identify + Group + User Activated |
groupId, role, plan, paid-seat status, activation date |
Whether paid seats are actually used and whether account value rises with active people |
| Would usage pricing work better? | Job Completed, API Request Completed, Report Generated |
account, quantity, workload type, compute class | Whether workload expands faster than headcount and whether one unit dominates customer value |
| Are package boundaries working? | Feature Used, Limit Reached, Upgrade Started |
plan, feature, limit, account, entitlement | Which gated capabilities create upgrade pressure rather than frustration |
| Is willingness to pay changing? | Pricing Viewed, Checkout Started, Quote Accepted |
package shown, price shown, billing term, cohort | How conversion changes when the offer or displayed price changes |
| Does greater use produce commercial value? | Product event joined to warehouse finance data | account, ARR, renewal, expansion, churn | Whether usage predicts revenue expansion, retention or contraction |
Segment's Track API already allows properties such as plan and account type to accompany an event, and the warehouse creates event-specific columns from those properties. As of August 2026, the platform also warns against unbounded event names such as creating a new event for every product number; stable events with properties are easier to govern and query.
The implication is simple: a pricing data model should describe accounts consuming value, not merely people interacting with screens.
Segment's collection architecture is stronger than its own pricing architecture
Segment deserves high marks for operationalising customer data. The harder question is whether the way Segment packages and prices that capability is equally strong.
The scorecard below grades the three steps of Monetizely's framework that matter most to this teardown.
| Framework step | Grade | One-line rationale |
|---|---|---|
| Packaging | B | Free, Team and Business create a clear adoption path, but Protocols - the governance capability most relevant to trustworthy pricing data - remains a Business-tier add-on as of 13 August 2026. |
| Pricing Metric | C+ | MTUs are understandable, but one lightly active visitor and one customer generating thousands of commercially important events can each count as one tracked user. |
| Operationalization | A- | Track, Group, warehouse schemas, filtering, validation, transformations and usage APIs provide unusually strong controls for turning product activity into structured data. |
Segment gets the difficult plumbing largely right. Its weaker point is the meter sitting above that plumbing.
The current pricing structure makes the tension visible. As of 13 August 2026, Segment Connections has a Free plan, a Team plan beginning at $120 a month and a custom-priced Business plan. Team includes 10,000 MTUs, then charges declining overage rates as tracked-user volume rises.
Source: Twilio Segment Connections pricing and Twilio Help Center, accessed 13 August 2026.
Segment's historical pricing pages are preserved in the Internet Archive, including the 2019 Wayback record cited in the source list. We use the live vendor page for the numerical comparison above rather than infer old commercial terms from incomplete archived renders.
The current model has a useful feature: one highly engaged customer does not create an ever-rising bill merely by generating more events. Yet the same property weakens the connection between Segment's revenue and the work its infrastructure performs. The official product limits make that mismatch concrete: as of August 2026, Segment specifies limits on event properties and inbound event throughput, including a standard threshold of 1,000 events per second before additional arrangements may be required.
An MTU is therefore a customer-facing simplification layered over infrastructure whose workload is fundamentally event-driven.
The warehouse must reconcile product behaviour before billing sees it
The biggest mistake in using Segment for pricing is turning a convenient event stream directly into a bill.
Product telemetry and commercial metering serve different standards. An analytics event can occasionally arrive late without creating a financial dispute. A billable event needs an agreed definition, deduplication, account attribution, entitlement logic, a billing period and a reconciliation path when finance or the customer challenges the count.
Segment's warehouse model helps bridge the gap. Each Track call carries an event ID, Segment uses IDs for server-level deduplication, and custom Track events can become event-specific warehouse tables with their own properties. Group calls add account identity, while Identify calls maintain user-level traits.
Governance becomes critical once telemetry influences money. As of August 2026, Protocols lets Business-tier customers define a Tracking Plan, flag events that violate it, block non-conforming events and route blocked activity into quarantine for review. Segment warns that blocked events can otherwise be permanently discarded.
The operating architecture should therefore divide responsibility deliberately.
| System | What it should own for pricing | What it should not own |
|---|---|---|
| Segment | User and account identity, product events, exposure to pricing tests, feature use, limit events, candidate usage measures | Contract interpretation, invoice calculation, recognised revenue |
| Data warehouse | Joining events to CRM accounts, contracts, invoices, costs and cohorts; metric evaluation; historical analysis | Customer-facing entitlement enforcement |
| Entitlement layer / product | Which package the account has and which features or limits are available | Retrospective pricing analytics |
| Billing system | Rated quantities, minimums, commitments, overages, credits, invoice generation | Deciding whether a product event is a good value metric |
| ERP / finance system | Invoices, collections, revenue records and financial reconciliation | Raw behavioural telemetry |
The warehouse is the point where behavioural evidence becomes commercial evidence. Segment should get the event there cleanly; billing should act only after the business has decided that the event deserves financial meaning.
A concrete example shows why. Suppose an API product receives both browser and server events for Report Completed. Twilio documented in 2026 that duplicate events can arise when the same activity is emitted through multiple connected libraries and recommends destination filtering where appropriate. A dashboard overstating reports by 2% is undesirable. An invoice overstating them by 2% is a customer dispute.
Four SaaS models show why pricing telemetry must follow the meter under consideration
No universal event list can support every SaaS pricing strategy. A useful Segment implementation starts from the pricing choices management may realistically make.
Four current B2B software businesses make the distinction concrete. As of 13 August 2026, Jira uses per-user cloud pricing; Datadog combines host and consumption measures across products; Snowflake charges compute through credits plus separate storage and transfer economics; and HubSpot combines platform editions, seats and marketing-contact capacity.
| Named SaaS example | Published pricing unit, 13 Aug 2026 | What a Segment-based pricing study would need to collect |
|---|---|---|
| Atlassian Jira | Standard listed at $7.91 per user/month on the current US pricing page; monthly cloud billing is tied to licensed users. | Paid seats, active seats, role, team size, project participation and automation use by account |
| Datadog | Infrastructure Pro lists $15 per infrastructure host/month billed annually; other products add units such as containers, custom events and monitored resources. | Hosts monitored, workload volume, telemetry produced, product modules activated and account-level infrastructure growth |
| Snowflake | Compute consumes credits; storage is charged by stored data and data transfer can add separate charges. | Compute-producing actions, workload class, query frequency, account usage growth and customer outcomes generated by that consumption |
| HubSpot | Marketing Hub combines an edition with marketing-contact capacity and Core Seats; HubSpot's current bundle page lists Professional at $890/month on a monthly commitment and additional marketing contacts from $50 per 1,000 in the configuration shown. | Marketing contacts, active users, campaign activity, CRM growth, feature adoption and account conversion |
The same Login Completed event would tell very little about all four. Jira's pricing question may turn on the relationship between licensed and active seats. Snowflake's turns on resource consumption. HubSpot requires contact growth. Datadog needs infrastructure and telemetry measures.
Segment's role is not to force those businesses into one data structure. Its role is to make the candidate measures comparable at the account level, so pricing leaders can ask which one grows most reliably with value, cost and willingness to pay.
A practical pricing study should therefore retain three families of signals:
Access signals: licensed users, active users, roles, admins and collaborators.
Consumption signals: jobs, reports, API calls, records, compute-heavy runs or other completed units of work.
Commercial signals: package, quoted price, ARR, expansion, downgrade, renewal and churn.
The first two can originate in Segment. The third should be joined from CRM, CPQ, billing and finance data before any pricing conclusion is drawn.
Segment's own history gives us a useful warning about confusing a convenient metric with a durable one.
Peter Reinhardt, then Segment's chief executive, explained on Twilio's February 2021 earnings call that Segment could let customer data be collected "in a consistent way" across disparate digital properties. Jeff Lawson made the business problem equally clear in May 2021, describing companies struggling to build "one picture of my customer" from data spread across systems.
Those statements describe Segment's enduring strength. Collection becomes more valuable as the customer's data estate gets richer and as more systems rely on the resulting record.
The commercial performance reported later was less compelling. Twilio disclosed Segment revenue of $75.7 million in Q1 2025, up 1% year over year, with Segment dollar-based net expansion of 94%. In Q2 2025, Segment revenue was $75.5 million, essentially flat, while Segment DBNE was 95%; Twilio attributed the pressure to churn, contraction and reduced use among existing customers. Segment nevertheless produced a 74.3% non-GAAP gross margin and $6 million of non-GAAP operating income in Q2.
Those disclosures do not establish that MTU pricing caused Segment's growth problem. They do show that the business had attractive software gross margins while expansion remained weak, which makes the quality of the expansion metric worth challenging.
Twilio's management has since integrated Segment more tightly into the wider platform. Khozema Shipchandler said on the August 2025 earnings call that the acquisition's original thesis was that "more intelligent communications were gonna be the future of this company." Thomas Wyatt, discussing a separate Twilio price increase on the same call, offered another useful pricing principle: "we think we have pricing power in that capability."
By Q2 2026, Twilio no longer reported Segment as a separate operating segment; company-wide revenue reached $1.50 billion, up 22% year over year, and company-wide DBNE reached 116%. The integration makes a clean Segment-only growth diagnosis impossible from public results after the 2025 reporting change.
Our view of Segment's next pricing reset is still firm: make committed accepted-event volume the primary meter for Connections, not MTUs. A paid plan should include an annual event commitment, predictable volume bands and overage pricing, while MTUs remain an operational guardrail rather than the revenue engine.
Three reasons support the reset:
An account sending one million accepted events asks Segment to do materially more data work than an account sending ten thousand, even when both represent the same number of people.
B2B SaaS activity can rise sharply while user count stays almost flat, especially when APIs, automated workflows and machine-to-machine activity create more of the product's value.
A governed accepted-event measure can exclude malformed or deliberately blocked traffic, something Segment's present controls already support. As of August 2026, events blocked through a Tracking Plan or Source Schema do not count towards MTU usage because they are stopped before the MTU calculation pipeline.
The annual commitment is not a second primary metric. It simply gives the buyer budget certainty around an event-volume model.
Data teams naturally want completeness. Pricing teams need discrimination.
An event belongs in a pricing data set when its presence can change a decision about segmentation, packaging, metric or rate. Recording 200 button clicks creates more warehouse columns. Recording
Workflow Completedwith account, plan, workflow type and quantity can tell management whether workflow volume is a better expansion metric than seats.Segment's own warehouse documentation reinforces the discipline. As of August 2026, Segment advises against unbounded event and property spaces, creates custom-event tables from incoming properties, and automatically evolves schemas as new values arrive. A poorly designed event can therefore become a persistent warehouse problem, not merely an untidy analytics label.
Protocols raises the standard further. Its tracking plan can define expected events, required properties, permitted values and data types, then surface violations when live data no longer conforms. As of August 2026, the feature is a premium Business add-on rather than a Free or Team capability.
For pricing work, Monetizely's position is that the tracking plan should be treated as a financial data dependency even before an event becomes billable. A company deciding whether to charge per workflow next year cannot reconstruct twelve months of clean workflow history after the decision has already been made.
The practical agenda for SaaS leadership follows from that point:
Choose the pricing question before funding new instrumentation. Management should decide which two or three alternative meters could plausibly replace the current model, then require enough historical evidence to compare them.
Make pricing telemetry a cross-functional asset. Product may emit the event, data engineering may maintain the pipeline and finance may reconcile the revenue, but the definition of a commercially meaningful event needs executive ownership across those functions.
Create a standing evidence window before major pricing resets. Collect enough account-level history to observe expansion, contraction and renewal behaviour before changing the metric. A pricing redesign based only on a fortnight of feature activity is weaker than one tied to a full customer lifecycle.
Separate the test of customer value from the test of the rate. First establish whether workflows, seats, contacts or another unit explain expansion and retention. Only after the metric earns that role should rate research determine what one unit is worth.
Require a finance-grade path before promoting any event to a billable unit. The final go/no-go decision should include reconciliation, dispute handling and invoice explainability, not merely a statistically attractive correlation in the warehouse.
Segment can make all five decisions better because it creates a consistent behavioural record across the product. Its highest value to a pricing team, however, comes from knowing where that role ends. The event stream supplies evidence; the warehouse establishes the commercial relationship; management chooses the pricing model; billing executes it.
That division of labour is what turns customer telemetry into pricing intelligence rather than another analytics dashboard.
Assumptions
Current prices and product terms are stated in US dollars and reflect official vendor pages available on 13 August 2026; regional pricing, negotiated enterprise terms and taxes can differ. The recommended Segment event-volume architecture is Monetizely's strategic recommendation, not a disclosed Twilio plan. No claim is made that Segment's MTU metric caused its 2024-25 contraction; the public financial evidence supports the expansion diagnosis but not a causal attribution.
Footnotes
Monetizing Agentic AI, Ajit Ghuman and Akhil Gupta, Monetizely, 2026. https://www.amazon.com/Monetizing-Agentic-AI-Handbook-Transformation/dp/B0H7Z13VKJ/
Twilio, “Connections pricing,” and Twilio Help Center, “How are MTU overages billed on the monthly Team plan in Twilio Segment?”, accessed 13 August 2026. https://www.twilio.com/en-us/products/connections/pricing https://help.twilio.com/articles/52687910022299
Internet Archive, historical captures of Segment pricing, 2019 archive index. https://web.archive.org/web/2019*/https://segment.com/pricing/
Twilio Segment, “Spec: Track,” accessed 13 August 2026. https://www.twilio.com/docs/segment/connections/spec/track
Twilio Segment, “Warehouse Schemas,” accessed 13 August 2026. https://www.twilio.com/docs/segment/connections/storage/warehouses/schema
Twilio Segment, “Protocols Overview,” “Customize Schema Controls,” and “Review and Resolve Event Violations,” accessed 13 August 2026. https://www.twilio.com/docs/segment/protocols https://www.twilio.com/docs/segment/protocols/enforce/schema-configuration https://www.twilio.com/docs/segment/protocols/validate/review-violations
Twilio Inc., Form 10-Q for the quarter ended 30 June 2025; Twilio Q2 2025 earnings-call transcript, 7 August 2025. https://www.sec.gov/Archives/edgar/data/1447669/000144766925000122/twlo-20250630.htm https://www.fool.com/earnings/call-transcripts/2025/08/07/twilio-twlo-q2-2025-earnings-call-transcript/
Twilio Inc., “Twilio Announces Second Quarter 2026 Results,” 6 August 2026. https://investors.twilio.com/news-releases/news-release-details/twilio-announces-second-quarter-2026-results
Atlassian, “Jira Pricing,” accessed 13 August 2026. https://www.atlassian.com/software/jira/jira/pricing
Datadog, “Datadog Pricing,” accessed 13 August 2026. https://www.datadoghq.com/pricing/list/
Snowflake, “Understanding overall cost,” accessed 13 August 2026. https://docs.snowflake.com/en/user-guide/cost-understanding-overall
HubSpot, “Create a Bundle,” pricing configurator, accessed 13 August 2026. https://www.hubspot.com/pricing/bundle
Twilio Q4 2020 earnings-call transcript, 17 February 2021, comments by Peter Reinhardt. https://www.fool.com/earnings/call-transcripts/2021/02/17/twilio-twlo-q4-2020-earnings-call-transcript/
Twilio Q1 2021 earnings-call transcript, 5 May 2021, comments by Jeff Lawson. https://www.fool.com/earnings/call-transcripts/2021/05/05/twilio-twlo-q1-2021-earnings-call-transcript/
Twilio Q2 2025 earnings-call transcript, 7 August 2025, comments by Khozema Shipchandler and Thomas Wyatt. https://www.fool.com/earnings/call-transcripts/2025/08/07/twilio-twlo-q2-2025-earnings-call-transcript/
Twilio Segment, “Product Limits,” accessed 13 August 2026. https://www.twilio.com/docs/segment/connections/rate-limits
Twilio Help Center, “Filter Duplicate Events from Browser and Server in Segment,” accessed 13 August 2026. https://help.twilio.com/articles/51746514133531
Twilio Help Center, “Do Blocked Events Count Toward My MTU Usage?” and Twilio Segment, “Source Schema,” accessed 13 August 2026. https://help.twilio.com/articles/43452158392731 https://www.twilio.com/docs/segment/connections/sources/schema