Salesforce customer data platform: reverse ETL or Data 360?

Six tools compared on one job: putting unified customer data into the Salesforce objects your reps use.

Is your customer data already modeled in Snowflake, BigQuery or Databricks? Then you don't need a second database: you need a reverse ETL tool that writes those rows into Salesforce. Hightouch is the pick for admins who want to run syncs without an engineer, with a free plan for 2 active syncs. Census, now Fivetran Activations, is the natural pick if Fivetran already loads your data. Is your truth inside Sales Cloud, Service Cloud and Marketing Cloud? Then Data 360, the CDP Salesforce builds into its platform, is the native answer. We compared six tools on how they match records, protect your API allocation, and what they really cost. Prices are in US dollars, read on each vendor's pricing page in 2026.

Romain CochardCEO of Hack'celeration
Updated 20266tools compared115,000API calls a day in a 15-license Enterprise org3,500free activation rows a month on Fivetran

The ranking is editorial: a sponsored spot is always labeled on its card, and no affiliate link changes a score.

At a glance

6 ways to sync customer data into Salesforce

The order follows one question: how well each tool writes unified customer data into Salesforce objects, for a team whose data already sits in a warehouse or an event stream. The score is each tool's overall score, the same on all our pages, such as the best reverse ETL tools and the best customer data platforms; that is why a tool can rank higher here with a lower score.

Best forFree planTeam sizeVisit
1HightouchBest for admins4.0/5Free (2 active syncs); paid plans usage-based✓Data team on Snowflake, BigQuery or DatabricksVisit →
2Census (Fivetran Activations)Best if you already pay Fivetran4.0/5Free up to 3,500 activation MAR/mo✓dbt-led data teamsVisit →
3PolytomicBest for two-way sync3.7/5From $500/mo; 14-day trial—RevOps teams syncing both waysVisit →
4RudderStackBest value for events + reverse ETL4.1/5Free $0; Growth $265/mo✓Product teams with event dataVisit →
5Twilio SegmentBest for web and app events4.2/5Free $0/mo; Team from $120/mo✓Teams tracking web and app visitorsVisit →
6Salesforce Data 360Native pick3.9/5$500 per 100k Flex Credits, or $240 per 1k profiles/year✓Orgs that live inside Salesforce cloudsVisit →

Data 360's free provisioning is for Salesforce customers, with a limited amount of storage and credits. Adobe Real-Time CDP, Tealium, Amperity and Treasure AI are covered in the guide below, not in the table.

Is your tool missing from this page? Email us: hello@hackceleration.com

Warehouse or platform

Reverse ETL or Data 360?

Pick a reverse ETL tool if…

  • Your data team already builds customer tables in Snowflake, BigQuery, Databricks or Redshift.
  • You want a few fields on Accounts, Contacts or Leads (a score, a plan, a last login), not a new platform.
  • Salesforce is one destination among others: your ad and email tools need the same audiences.
  • You want a pilot this month: a free plan and one sync prove the value.

Pick Data 360 if…

  • Most customer data already sits in Salesforce clouds and you have no warehouse team.
  • You plan to use Agentforce or Marketing Cloud on unified profiles inside the platform.
  • You prefer one vendor and one contract, and accept credit-based billing.
  • You need Zero Copy access to Snowflake, Databricks, BigQuery or Redshift without duplicating the data.
Before you compare

Reverse ETL, CDP and Salesforce: the basics

Four short cards: what the category is, how a sync works, what to demand, and what each billing unit means. Already know it? Jump to the six tools.

01

Reverse ETL and CDP: what the words mean

Basics

A customer data platform (CDP) collects data about the same person from several places (your site, your app, your CRM, your order system), matches it into one profile, and sends audiences to the tools that act on them. Salesforce sells its own CDP, Data 360.

Reverse ETL is narrower. ETL copies data from your apps into a data warehouse such as Snowflake, BigQuery or Databricks. Reverse ETL does the opposite: it takes a table your data team has already built in that warehouse (a lead score, a lifetime value, a last-login date) and writes it into Salesforce fields. A tool that does reverse ETL plus profiles and audiences on top of the warehouse is called a composable CDP.

So the real question behind "Salesforce customer data platform" is where your truth lives. If it lives in the warehouse, you want reverse ETL. If it lives in Salesforce and its clouds, you want Data 360.

02

How a sync into Salesforce really works

Basics

A sync usually runs in four steps. It runs a query on your warehouse (you pay that compute to your warehouse vendor, on top of the tool). It compares the result with the previous run to find rows that changed. It matches each row to a Salesforce record through an identifier: an External ID field, an email, or the Salesforce record ID. Then it writes the changes through the Salesforce APIs, usually the Bulk API, in batches.

The match step is where projects break. For an "update or create" sync, Hightouch and Fivetran Activations want an explicit External ID field in Salesforce, marked unique. The Salesforce record ID only works for update-only syncs, since nobody but Salesforce can set it. Ask your admin to create that field before the first sync.

03

What to demand before you sign

Checklist

Six questions sort the tools faster than any feature grid:

  • Which sync modes does it offer for Salesforce (insert, update, upsert, mirror, delete), and which objects (custom objects, person accounts, Opportunity Contact Roles)?
  • How does it protect your API allocation: Bulk API by default, a daily cap, adjustable batch size?
  • How often it can sync: Fivetran Activations runs 15-minute syncs on Standard; 1-minute syncs start on the Enterprise plan. A score your reps read once a day needs neither.
  • What happens on failure: are rejected rows listed with the Salesforce error, and retried?
  • What is the billing unit, and what counts in it: rows changed, rows activated, tracked users, events, credits?
  • Where your data is processed: region, private connection to the warehouse, and a SOC 2 report you can actually read.
04

The five billing units, explained

Pricing

Each tool counts something different, which is why list prices don't compare.

  • Active syncs (Hightouch free plan): the number of syncs switched on at the same time, whatever their size.
  • Monthly active rows, MAR (Fivetran Activations): distinct rows sent in a month, counted across all the syncs of one activation (one source to one destination); a row updated ten times counts once.
  • Events (RudderStack Core): every tracked action, and every record its reverse ETL sends.
  • Monthly tracked users, MTU (Segment): every unique person Segment saw in a month, anonymous visitors included.
  • Flex Credits and profiles (Data 360): credits are spent per action with a rate per action; profile pricing charges per 1,000 unified profiles a year.

To compare two quotes, translate both into one number: what a nightly sync of your Accounts costs over a year.

The tools, one by one

Six tools, how they really write to Salesforce

For each one: who it is for, how it connects to your org, an order of price, and its limit. Other names that come up in buyer guides: Adobe Real-Time CDP and Tealium suit enterprise marketing stacks priced on quote; Amperity and Treasure AI are enterprise CDPs; Hevo Activate and Integrate.io are pipeline tools with a Salesforce connector. Need someone to scope the project? A Salesforce agency can size it before you sign.

01

Hightouch: reverse ETL an admin can run without an engineer

Best for admins

Hightouch reads a table or a SQL model in your warehouse and writes it into Salesforce objects: a churn score on the Account, a product usage field on the Contact, a lead created from a free-trial signup. Its free tier is still plain reverse ETL, but it now sells itself as a composable CDP and an agentic marketing platform. Composable means it builds audiences and profiles on top of your warehouse instead of copying your data into its own database.

With Salesforce, it supports Sales, Service, Financial Services and Health Cloud, and five sync modes: Insert, Update, Upsert, Archive and All (the last one only for picklist metadata). Upsert is the mode you want most of the time, since it sends one API request per batch of rows. It uses the Bulk API by default, with batches of 1,000 records and ten batches in parallel, and you can switch to the REST API when your org has more REST headroom than Bulk headroom.

Price: the free Basic Reverse ETL plan covers up to 2 active syncs, unlimited destinations and unlimited seats. Paid plans (Composable CDP and the Agentic Marketing Platform) are usage-based with no MTU limit and no public rate: you get a number after a demo, and the free plan is the only public price. Those paid plans list SOC 2 and SOC 3, ISO 27001, HIPAA compliance and deployment in an EU region.

Its limit: the free plan is now small. Two active syncs is enough for one Account score and one Contact update, not for a RevOps program. And Archive is a destructive delete, not a recoverable archive: every Salesforce record your model returns is deleted. Want options around it? See the Hightouch alternatives.

Price
Free: 2 active syncs; paid plans usage-based, by demo
Salesforce modes
Insert, Update, Upsert, Archive, All
API
Bulk API by default, REST optional
Its limit
Free plan capped at 2 active syncs
See Hightouch →
02

Census (Fivetran Activations): reverse ETL for teams that already live in dbt

Best if you already pay Fivetran

Census is the reverse ETL tool Fivetran bought. It now sells it as Fivetran Activations, under one contract with Fivetran's connectors and transformations. If your Salesforce data already flows into Snowflake or BigQuery through Fivetran, this is the shortest path to send the modeled result back.

It writes to all standard and custom Salesforce objects, with five sync behaviors: Create Only, Update or Create, Update Only, Mirror and Delete. With Mirror, the Salesforce object follows your model, removals included. It primarily uses the Salesforce Bulk API to keep API consumption low, and it can sync Lead or Contact and Lead or Account in a single multi-destination sync, which saves you from writing two models when you don't know if a person is already a contact.

Price: the free plan includes 3,500 monthly active rows (MAR) for activations, next to 500,000 MAR for ingestion connectors. Above that, you pay per MAR, and each new activation gets a 14-day free trial. Fivetran's own example puts the median Snowflake-to-Salesforce activation at 12,007 MAR for $202.25 a month at list price (Standard plan, companies of 1 to 200 employees). Audience Hub, the builder that lets marketers make audiences without SQL, is an Enterprise feature.

Its limit: the bill. A reviewer wrote in 2026 that costs are hard to predict when you add tables or sync data that changes often, and another found inconsistencies in the Salesforce setup documentation for Activations during a proof of concept. Run that proof of concept on a sandbox while the trial clock runs. Leaving Fivetran altogether? Read how to switch from Fivetran.

Price
Free to 3,500 activation MAR/mo, then per MAR
Salesforce modes
Create Only, Update or Create, Update Only, Mirror, Delete
API
Primarily the Bulk API
Its limit
MAR billing is hard to forecast
See Census →
03

Polytomic: one tool for Salesforce in and out

Best for two-way sync

Polytomic does ETL (Salesforce into your warehouse), reverse ETL (your warehouse into Salesforce) and change data capture in the same product. That matters when the same person is edited in both places: a sales rep updates a phone number in Salesforce while your product writes the plan name. Instead of two vendors and two schedules, you keep one.

Its most useful feature for a Salesforce admin is quiet: when Polytomic can read your org's API limit, it caps its own daily calls at 75% of it. You can also set your own daily call limit. In an Enterprise org with 15 Salesforce licenses (115,000 calls a day), that leaves at least 28,750 calls for your other integrations. Sandboxes get their own connection, which keeps test syncs away from production.

Price: pricing begins at $500 a month; Standard and Enterprise are quoted after a call. Enterprise adds on-premises deployment and SSO. There is no free plan, only a 14-day trial.

Its limit: the entry ticket. You can't run a free pilot past 14 days, and everything above the $500 entry price is quoted after a call, so you can't size next year's bill before talking to sales.

Price
From $500/mo; Standard and Enterprise by quote
API guard
Caps its own calls at 75% of your org limit
Deployment
Cloud, or on-premises on Enterprise
Its limit
No free plan; quote above $500/mo
See Polytomic →
04

RudderStack: event collection and reverse ETL on one bill

Best value for events + reverse ETL

RudderStack is the warehouse-first answer to Segment: it collects web, app and server events, stores them in your warehouse, and runs reverse ETL from there. For a Salesforce team, that means product signals (a feature used, a seat added) can land on the Account without a second vendor. Its reverse ETL writes to Salesforce through a Bulk Upload destination on Bulk API 2.0, in Upsert mode only: no mirror, no delete.

Price: the Core Free plan is $0 for 250,000 events a month and includes 10 reverse ETL connections. Growth starts at $265 a month for 1 million events, with 25 reverse ETL connections, a 30-day trial and 15% off on annual billing. There is a catch in the unit: each record sent by reverse ETL counts as an event on your Core plan. If a nightly sync rewrites a score on all 20,000 of your Accounts, that is about 600,000 events in a 30-day month, more than double the free quota on its own.

RudderStack also sells Lookout, a separate audience product billed on activated records (a person's first sync to a destination from an audience, counted once per audience and per destination). Its Growth plan starts at $1,000 a month, billed annually, so $12,000 a year, with 40,000 activated records a year included. Updates, re-syncs and removals are not billed.

Its limit: on the Free plan, going over your event quota is not a soft overage. Data processing stops after your second consecutive overage, which can freeze your Salesforce syncs at month end. Identity resolution (Profiles) is an Enterprise feature. Watch the meter, or size the plan on events plus reverse ETL rows.

Price
Free $0 (250K events/mo); Growth from $265/mo
Reverse ETL
10 connections on Free, 25 on Growth
Unit
Each reverse ETL record counts as an event
Its limit
Free plan: processing stops after a second overage
See RudderStack →
05

Twilio Segment: when the data starts on your website and app

Best for web and app events

Segment is a widely used event collector: you add its tracking to your site and app, it builds a stream of who did what, and sends it to hundreds of tools, Salesforce included. Its reverse ETL module also reads your warehouse and writes rows to Salesforce, which lets a team start with events and add warehouse models later without changing vendors.

Price: the Free plan is $0 a month for 1,000 visitors and 500,000 reverse ETL records a month. Team starts at $120 a month for 10,000 visitors and 1,000,000 reverse ETL records, then adds $12 per 1,000 extra monthly tracked users between 10,000 and 25,000, $11 between 25,000 and 100,000, and $10 above. At 25,000 tracked users, Team lists at $300 a month. Team comes with a 14-day trial, and annual billing saves up to 20%. Identity resolution and unified profiles sit on the CDP plans, which are quoted.

Know the unit: a monthly tracked user is any unique user Segment collects data on in a billing month, identified or anonymous. Anonymous blog readers count, even if they never become a Salesforce lead. If most of your traffic is content, filter what you track before you sign.

Its limit: you buy Segment for its events. If your goal is only to push warehouse tables into Salesforce, a dedicated reverse ETL tool does it with fewer moving parts. Leaving it? Read how to switch from Segment.

Price
Free $0/mo; Team from $120/mo
Reverse ETL
500,000 records/mo on Free, 1,000,000 on Team
Unit
Monthly tracked users, anonymous included
Its limit
Profiles and identity only on quoted CDP plans
See Twilio Segment →
06

Salesforce Data 360: the native CDP, for orgs that stay inside Salesforce

Native pick

Data 360 (formerly Data Cloud) is the CDP Salesforce builds into its own platform. It works differently from a reverse ETL tool: it ingests data, unifies it into profiles, and makes those profiles and segments readable by Sales Cloud, Marketing Cloud and Agentforce. With Zero Copy, it can also read Snowflake or Databricks tables in place instead of copying them.

Price: Flex Credits cost $500 per 100,000 credits, spent on any Data 360 action. Profiles cost $240 per 1,000 profiles a year, with one Flex Credit per profile, 25 calculated insights and 100 segments included. Enterprise Profiles cost $420 per 1,000 profiles a year, with two credits per profile. For 50,000 profiles, that lists at $12,000 a year on Profiles. Bringing data in is free, and Salesforce customers can provision Data 360 for free with a limited amount of storage and credits.

Why it ranks last here: this page is about getting customer data into Salesforce objects your reps already use. Data 360 is the better choice when you want unified profiles inside the Salesforce platform; it is a heavier project when you only want a score on the Account. For the full Data 360 comparison, see our Salesforce CDP guide.

Its limit: forecasting. Every action has its own credit rate (data prep, for one, is counted per million rows processed), and a calculated insight spends credits again each time it refreshes. Salesforce's own example: a batch calculated insight on 20 million rows costs 800 Flex Credits per run at the base rate. Refreshed daily, that is about 24,000 credits a month, or $120 at list, for one insight. Open the Digital Wallet, Salesforce's free usage tracker, before your first segment.

Price
$500 per 100k Flex Credits; Profiles $240 per 1k/year
Ingestion
Free
Warehouse
Zero Copy reads Snowflake and Databricks in place
Its limit
Credit spend is hard to forecast
See Salesforce Data 360 →
Traps and tips

What admins learn after the first sync

The numbers that change your budget and the setup details that break syncs. If Salesforce itself is the problem, see how to switch from Salesforce; if your warehouse is still to pick, start with the best data warehouses.

01

Count your Salesforce API calls before you count rows

Trap

An Enterprise or Professional-with-API org gets 100,000 API calls per 24 hours, plus 1,000 per Salesforce license. With 15 licenses, that is 115,000 calls a day, shared by every integration, and calls made by installed managed packages count against the same total. The Bulk APIs count too. Unlimited and Performance orgs get 5,000 calls per Salesforce license instead of 1,000.

In REST mode, Hightouch sends 200 records per request by default, so 100,000 changed rows burn 500 calls. That sounds small until your enrichment tool, your marketing automation and your new reverse ETL all run at 2 a.m. Set API Usage Notifications in Setup at 70%, and prefer tools that use the Bulk API or cap themselves.

Base allocation
100,000 calls/24h + 1,000 per license
Example
15 licenses = 115,000 calls
02

Free plans shrank: read the unit, not the $0

Budget

Hightouch's free plan is capped at 2 active syncs; a reviewer wrote in 2026 that it used to allow up to 25 syncs. Fivetran Activations gives 3,500 monthly active rows, enough for a pilot on one segment of Accounts, not for your whole base. RudderStack's free 250,000 events include reverse ETL records, so one nightly Account sync can eat the quota. Segment's free plan gives 500,000 reverse ETL records a month but caps you at 1,000 visitors and one sync to your warehouse a day. Polytomic has no free plan at all, only a 14-day trial.

Use a free plan for one sync that proves value (a product-usage field on the Account is the classic), then negotiate with that number in hand.

Hightouch
2 active syncs
Fivetran Activations
3,500 MAR/mo
03

Know what happens at the quota line

Trap

Overage rules differ more than prices. On RudderStack's Free plan, data processing stops after your second consecutive overage. On Segment Team, each extra 1,000 tracked users costs $12, then $11 and $10 at higher bands. On Lookout, RudderStack's audience product, records above the 40,000 included in Growth are billed on usage. On Data 360, each Flex Credit usage type is billed at its base rate for the first 300,000 credits of the month, then discounted automatically.

Ask each vendor, in writing, three numbers: your unit cost today, the cost if your volume doubles, and the price at renewal in year two.

04

The real first-year floor

Budget

Annual minimums matter more than monthly prices:

  • RudderStack Lookout Growth: $1,000 a month billed annually, so $12,000 up front.
  • Data 360 Profiles: $12,000 a year for 50,000 profiles, before any discount.
  • Polytomic: from $500 a month, so $6,000 a year at the entry price.
  • Segment Team: $300 a month at 25,000 tracked users, $3,600 a year.
  • Fivetran annual contracts: the "up to 22%" discount is for big contracts. On the Standard plan, an annual list price up to $37,900 gets 5%, and the 22.6% rate starts at $145,340.

Then add your warehouse compute: each sync runs a query. Materialize synced models as tables, not views, and sync no more often than your reps actually look at the field.

Lookout Growth
$12,000/year
Polytomic entry
$6,000/year
05

Features locked to a higher plan

Trap

The entry plan rarely carries the feature that made you shortlist the tool. Check these before you sign:

  • Fivetran Activations: Audience Hub and 1-minute syncs start on Enterprise; Standard syncs every 15 minutes.
  • RudderStack Core: identity resolution (Profiles) and HIPAA are on Enterprise; Growth gives 25 reverse ETL connections, Free 10.
  • Twilio Segment: Regional Segment (EU or US data processing) and HIPAA eligibility, an add-on, are Business features.
  • Polytomic: on-premises deployment and SSO are on Enterprise.
  • Data 360: Profiles include 25 calculated insights and 100 segments; Enterprise Profiles raise that to 100 and 500.
Fivetran
Audience Hub on Enterprise
Segment
EU or US region on Business
06

Four Salesforce gotchas that break syncs

Setup
  • Empty strings don't clear fields: Salesforce ignores empty strings on update. Send null to blank a field.
  • Enhanced domains change your instance URL: reauthorize the connection or syncs break.
  • Row locks (UNABLE_TO_LOCK_ROW) come from too many parallel batches on related records: lower parallelization to five or fewer.
  • Person accounts need LastName mapped, not FullName.
07

Zapier or a script instead?

Choice

For a few hundred records a day, a Zap on Salesforce or a scheduled script works. Past that, per-record API calls eat your daily allocation, and nothing tells you which rows failed when someone renames a column. Buy a reverse ETL tool when you sync more than one object, more than once a day, or when a sales team will act on the field.

FAQ

Salesforce customer data: your questions

  • Is Salesforce Data 360 the same as Salesforce CDP?

    Yes. Data 360 is the official new name of Data Cloud, the product still searched as "Salesforce CDP". If you signed a Data Cloud contract, its agreement and pricing are honored through the end of your current term, and renewal under the new packaging is discussed with your account executive. Customers on the older Data Services credits are not forced to migrate, but moving to Flex Credits or Profile pricing means swapping all of their existing Data Services credits.

  • What happens when a sync hits the Salesforce daily API limit?

    Usually nothing breaks at once. When a paid org reaches its daily limit, Salesforce still lets operations run past it by a certain amount, up to a hard cap, and only as an occasional buffer. That grace does not apply to trial orgs, Developer Edition or sandboxes. Past the cap, calls return an error, so a large backfill can block your other integrations until the 24-hour window moves on: run backfills at night and keep API Usage Notifications on.

  • Why does a sync report success when Salesforce records didn't change?

    Because Salesforce can return a successful response for a batch even when some of its rows failed. Read the row-level errors, not the run status. Hightouch has split retries to resend only the rejected rows, which helps with temporary errors such as UNABLE_TO_LOCK_ROW. The other usual cause is a target field that can't be edited, such as a formula field: Salesforce rejects updates to it.

  • Can I sync warehouse data to Salesforce custom objects?

    Yes. Fivetran Activations supports all standard and custom objects, plus Opportunity Contact Roles. Hightouch writes to standard and custom objects and can also publish platform events, which let a Salesforce flow react to warehouse data. The integration user needs read and write access to every object and field you map, and Delete permission on any object a sync deletes from.

  • How long does a first sync take to set up?

    Connecting the tool is the quick part. The time goes to the Salesforce side: creating External ID fields marked unique, granting the integration user its permissions, and agreeing on which system wins when a rep edits the same field. On Fivetran, each new activation gets a 14-day free trial, so have that prep done before you create it. Building customer models from scratch in the warehouse is a separate project.

  • Can reverse ETL keep customer data in the EU?

    Yes, on the right plan. Fivetran runs its data connectors in the US, Canada, the EU, the UK, Australia, India and Singapore, and Activations keeps a region setting for each source. Hightouch lists deployment in an EU region among its paid-plan features. On Segment, Regional Segment (EU or US) is a Business feature.

  • Can I sync Salesforce sandboxes and production with the same tool?

    Yes, with two connections. Polytomic recommends one connection for the sandbox and one for production. In Hightouch, older sandbox versions can cause processing delays or errors, so refresh the sandbox before you test a large sync.