Most explanations of a customer data platform stop at what it does. The buyers shopping for one have usually moved past that: what they are deciding is whether the job needs another platform, and if so, whether now is the time. Their answers are the useful half of this page — what they weigh against buying, how often, and what they do first.
What is a CDP?
A customer data platform (CDP) is software that collects customer data from across your systems, resolves the records belonging to the same person into one profile, and pushes those profiles to the tools that act on them — email, advertising, service, analytics.
The three jobs are ingest, resolve and activate. In these interviews, identity resolution is one of the clearest jobs buyers associate with dedicated CDP capability: a warehouse, a CRM and an email platform all store customer data, but a CDP is expected to decide that several records are one customer, and keep that true as new data arrives.
What the interviews add is the part a definition leaves out:
Among buyers in market for a CDP, more than a third consider delaying the purchase or solving the problem another way. Delaying is the most common response — buying later, usually once the data or another system is ready. Others consider stretching a platform they already own, using the data warehouse, or building some of the capability in-house.
So the practical question is less which CDP than which job is unsolved, and whether something already in the stack can do it.
What is the difference between a CDP and a data warehouse?
A data warehouse is the central foundation for storing, transforming and modelling data. A CDP is built to make customer data usable across systems, by combining identity resolution, customer profiles and activation into the tools that act on them. The line has blurred because warehouse-native, or composable, CDPs now run those functions directly on warehouse data rather than inside a separate store.
A data warehouse — Snowflake, Databricks, BigQuery — is where data lands and gets modelled. A CDP adds identity resolution and activation. Warehouse-native CDPs such as Hightouch read from the warehouse rather than copying the data into a separate system.
That is why buyers now ask whether they need a separate CDP at all. Some treat the warehouse itself as the customer-data layer; others consider a warehouse-native CDP on top of it.
More than one in ten buyers in market for a CDP raise the warehouse as their CDP or as an alternative to one, and among buyers replacing a CDP it is nearly three in ten. A small app company puts it in one line: “We treat our data warehouse like a CDP.” An enterprise software company sets duplication as a requirement: “we do not want to duplicate the data and maintain two data sets as a source.”
Against a CRM
A CRM is the system of record for known relationships — accounts, contacts, deals, service history. A CDP assembles customer data from many systems, the CRM among them. Buyers do not always separate the two: an industrial distributor says “the line’s so blurred there that it’s hard to even differentiate between the two.” The full comparison is CDP vs CRM.
Against a DMP and an email platform
A DMP built anonymous advertising audiences, and buyers now describe it in the past tense (CDP vs DMP). An email platform holds rich data on the people on its list, and some buyers run it as a CDP light — where that stops working is covered in does your ESP replace a CDP.
Who is buying a CDP?
Marketing, CRM and lifecycle titles are the most represented among the in-market buyers in these interviews, mostly at mid-sized to large companies. More than half hold marketing or CRM and lifecycle titles, while data and IT titles account for about one in eight — no more common than across all the buyers we interview.
The title mix resembles the buyers shopping for email platforms more than a data-team-heavy cohort. Companies with $1B–$10B in revenue are over-represented against the wider corpus, and the very largest companies are under-represented. By industry the cohort looks like the corpus, led by retail and apparel.
What they want it for is consistent. A single view of the customer across channels and sources is the reason given most often, by more than a third. Personalization and getting audiences into advertising and other channels each come up in about one in six. Identifying customers who never identify themselves comes up in about one in twenty.
Do you need a CDP?
Most companies asking do not run a dedicated CDP, and many get the job done without one. In these interviews dedicated CDPs cluster in larger companies — about one in twelve companies under $50M in revenue run one, against nearly one in four above $1B — and where something does the CDP job, about half the time it is a tool bought for something else, most often the email platform or the CRM.
The buyers who go ahead name two jobs their stack could not do: resolving one customer across several systems, and getting those profiles into channels the existing tools cannot reach.
Among buyers in market, most are shopping for their first CDP — they outnumber buyers replacing one about three to one — and most do not weigh an alternative. But a large minority weigh a different answer first, and the order is the most useful finding on this page:
- Buy later, after the data or another system is ready
- Stretch a platform already in the stack
- Use the data warehouse
- Build it in-house
More than a third of buyers in market weigh at least one of these. A handful conclude they do not need a CDP at all.
Buy later, once something else is ready
About one in seven buyers in market put the CDP behind other work. The reasons split roughly three ways: cleaning up the data, finishing another system first — a CRM, an ERP, a PIM, a commerce replatform — and budget or an unprovable return.
A home-furnishings retailer ran a full evaluation, with an RFI and a score sheet, and concluded that “a CDP only works if you have clean data” — so it cleaned up first. An industrial manufacturer is stabilizing a CRM upgrade: “So after that comes the CDP program.”
Stretch a platform already in the stack
More than one in ten weigh making a system they already run do the job — most often an email platform, sometimes Shopify or the CRM. It covers segmentation within that channel, and buyers describe the boundary where it stops.
A DTC apparel brand says “Klaviyo is kind of our CDP light.” A fintech has “been using Braze as our CDP light” — and is buying a CDP for exactly what that does not cover: sending high-value segments to Meta and Google.
Use the data warehouse
Just as many weigh the warehouse, through a composable or warehouse-native setup or by treating the warehouse itself as the CDP. It is a replacement-buyer’s move: buyers replacing a CDP raise it about three times as often as first-time buyers.
A national insurer replacing its CDP: “Snowflake is already our data warehouse, so we would like to make it a Customer Data Platform as well.” A subscription commerce brand says composable vendors “are finding their way to the top of our list, more so than more traditional CDPs.”
Build it in-house
About one in sixteen consider building, and usually as a blend of built and bought rather than an outright alternative. The examples here involve teams prepared to own at least some of the capability internally.
A consumer-goods manufacturer concluded “a true CDP was not the solution at that time and neither was Marketing Cloud, so we ended up building it ourselves.” A global apparel company replacing its CDP argues that “some of it should also build in house, to make it more scalable.”
The alternatives differ sharply between the two buyer groups. First-time buyers who weigh an alternative more often wait or stretch what they own; replacement buyers raise the warehouse much more often. And a handful decide against the category outright — a DTC retailer weighing one asks “So do I need it? No,” and a global healthcare company that built its own data ecosystem says “We really don’t need a comprehensive Customer Data Platform.”
Does company size decide whether you need a CDP?
More than it decides most software purchases. Dedicated CDP ownership rises substantially with company size in these interviews: from about one in twelve companies under $50M in revenue to roughly one in four at $1B–$10B, before easing to about one in five above $10B. And at the small end, where anything does the CDP job, it is usually a tool the company bought for something else.
Company revenue
Run a dedicated CDP
Where something does the CDP job, how often it is a tool bought for something else
Under $50M
About 1 in 12
About 2 in 3
$50M–$250M
About 1 in 7
About 3 in 5
$250M–$1B
About 1 in 6
More than half
$1B–$10B
About 1 in 4
About 2 in 5
Over $10B
About 1 in 5
About 1 in 3
Dedicated CDPs here are platforms bought for the job — Segment, Salesforce Data Cloud, Adobe Real-Time CDP, Amperity, Tealium, BlueConic, Hightouch, mParticle and similar. The tools bought for something else are led by Klaviyo and Salesforce, then Bloomreach and Salesforce Marketing Cloud.
That strength of relationship with size is particular to CDPs, even though ownership does not rise in every band. Across the same range of company sizes, CRM ownership rises by about half, from roughly one in four companies to nearly two in five, and the data warehouse barely moves. A dedicated customer-data layer is one of the clearest size-driven purchases in these interviews.
The right-hand column is the other half of the story. Below a billion in revenue, the CDP job is more often done by a tool bought for something else — most often the email platform or the CRM — than by a CDP: the CDP light buyers describe above, at scale. The implication, ours rather than a buyer’s, is that a smaller company’s first question is usually not which CDP, but whether the tool already doing the job has hit its ceiling. Where that ceiling sits is the subject of does your ESP replace a CDP.
Which CDPs are buyers using?
Hightouch, the warehouse-native option, has the highest observed average rating here at 7.8/10; Segment is the most-rated, at 7.0/10. Everything else sits between 6.2 and 7.2.
| Platform | Avg rating | What buyers here emphasize |
|---|---|---|
| 7.8 | Getting audiences into other tools, and working from the warehouse the data already sits in. | |
| 7.3 | Identity resolution, raised by nearly half of the buyers who rate it highly. | |
| 7.0 | The most-rated platform here. Data collection and getting audiences into other tools; integrations come up among satisfied and dissatisfied buyers alike. | |
| 6.9 | Collecting data across sources — tags and events — raised by buyers at both ends of the rating scale. | |
| 6.8 | Integration and activation among buyers who rate it highly; cost is the most common complaint among those who do not. | |
| 6.6 | Integration among buyers who rate it highly; implementation effort among those who do not. | |
| 6.4 | — | |
| 6.2 | — |
The ratings narrow a shortlist more than they pick a winner. Treasure Data, ActionIQ and Simon Data are also named, by fewer buyers. For what the platforms are like to live with, what CDP buyers wish they’d known is the companion read.
How long does a CDP take, and what does it cost?
Often longer than expected, with buyers describing usage-based pricing rather than a simple flat licence. Buyers describe implementations that run materially longer than planned, including enterprise projects that reach roughly a year. Much of the work they describe is data validation and integration rather than switching the software on — the same data work the buyers in route one choose to do first. Some warehouse-native buyers describe a shorter path, because the underlying data already sits in the warehouse.
Buyers describe volume-based pricing on measures such as events, profiles or rows, with implementation services sometimes charged separately. The detail is on how long CDP implementations take and what CDP buyers actually pay.
Want this checked against your own stack?
Start my checkup →Should you buy a CDP?
Start with the unsolved job, not the category. In these interviews, the clearest jobs associated with dedicated CDP demand are resolving customer identity across systems and activating those profiles somewhere the existing stack cannot reach. The questions below are our recommendation rather than a framework buyers describe, built from where the buyers above landed.
Three questions get you further than a vendor comparison. First, is the data ready? The most common alternative in these interviews is buying later for exactly this reason, and a CDP resolves identity across sources without fixing records that are wrong at source. Second, is the system of record settled? Buyers finishing a CRM or a replatform put the CDP after it, and the CRM’s capabilities change what the CDP has to do. Third, what is the honest alternative? Price a CDP against the email platform you would stretch, the warehouse you already pay for, or a build — not against doing nothing.
If the answer is yes, the questions worth asking CDP vendors and how to write a CDP RFP are published separately. For buyers already on one, switching off the CDP and why companies replace their CDP cover the second decision, and Segment alternatives buyers actually switch to shows where one platform’s leavers went.
Common questions
What is a CDP?
A customer data platform is software that collects customer data from across your systems, resolves the records that belong to the same person into one profile, and pushes those profiles and segments to the tools that act on them. The three jobs are ingest, resolve and activate. In these interviews, identity resolution is one of the clearest jobs buyers associate with dedicated CDP capability: a warehouse, a CRM and an email platform all store customer data, but a CDP is expected to decide that several records are one customer and keep that true as new data arrives.
Do I need a CDP?
Not necessarily, and most companies asking do not run a dedicated one. In these interviews about one in twelve companies under $50M in revenue run a dedicated CDP, against nearly one in four above $1B, and where something does the CDP job, about half the time it is a tool bought for something else, most often the email platform or the CRM. Among buyers in market, more than a third weigh buying later, stretching a platform they own, using the data warehouse or building it. The relevant question is which customer-data job remains unsolved after accounting for the systems already in your stack.
What is the difference between a CDP and a data warehouse?
A data warehouse is the central foundation for storing, transforming and modelling data. A CDP is built to make customer data usable across systems, combining identity resolution, customer profiles and activation into the tools that act on them. Warehouse-native or composable CDPs, such as Hightouch, run that second job on top of the warehouse instead of copying the data into a separate system. In these interviews more than one in ten buyers in market for a CDP raise the warehouse as their CDP or as an alternative to one, rising to nearly three in ten among buyers replacing a CDP.
Can an email platform like Klaviyo or Braze replace a CDP?
For some buyers it already does part of the job. More than one in ten buyers in market for a CDP weigh stretching a platform they already own, most often an email platform. The boundary they describe is activation beyond email: one buyer using Braze this way is buying a CDP specifically to send high-value customer segments to Meta and Google, which the email platform was not doing.
Should you build a CDP in-house?
Some buyers do, and more weigh it: about one in sixteen buyers in market for a CDP consider building it, usually as a blend of built and bought rather than an outright alternative. The examples involve teams prepared to own at least some of the capability internally, and the buyers weighing it describe an ongoing build-or-buy debate rather than a settled choice.
Which CDP do buyers rate highest?
Hightouch, a warehouse-native CDP, has the highest observed average rating here at 7.8/10, followed by Amperity at 7.3/10 and Segment, the most-rated, at 7.0/10. Tealium rates 6.9/10, Salesforce Data Cloud 6.8/10, Adobe Real-Time CDP 6.6/10, mParticle 6.4/10 and BlueConic 6.2/10. Most of the field sits between 6 and 7.3, so the ratings narrow a shortlist more than they pick a winner.
What should be in place before you buy a CDP?
Clean enough source data, and clarity about the systems the CDP will connect to. About one in seven buyers in market put the CDP behind other work, including data cleanup, CRM or ERP projects, commerce replatforms, and budget or ROI questions. A CDP resolves identity across sources; it does not fix source records that are wrong, which is also where implementations tend to lose time.
Get a checkup on your tech stack
A checkup for your marketing stack, built from hundreds of verified buyer interviews: what matters most to you, a shortlist for your stack, and your next steps.
Start my checkupFree · no password · your answers are anonymized before they inform research like this.