Customer data platforms: do you need one, or better integration?

The CDP Institute, which coined the term, defines it as packaged software that builds a persistent, unified customer database accessible to other systems. In practice a product earns the label if it does four things:

1. Ingests first-party data from everywhere: website events, mobile app events, email engagement, purchases, support tickets, CRM fields.
2. Resolves identity, stitching anonymous web visits to the known person once they log in or buy, and merging duplicates across systems.
3. Builds segments that marketers can define without engineering help, such as "bought twice, no visit in 60 days, opened the last email."
4. Activates those segments by pushing them to advertising platforms, email tools, personalization engines and analytics.

That is the whole product. Everything else vendors add, from predictive scoring to journey orchestration, is a feature from an adjacent category bolted on. Treat those as nice-to-haves and evaluate the four core jobs.

The category is crowded and getting louder. Segment, mParticle, Tealium, BlueConic, Lytics and Treasure Data are the names that come up most, and the large suites are moving in: Salesforce announced a Customer 360 initiative at Dreamforce this month, and Adobe has been talking about a unified profile layer since its spring conference. Expect every vendor you already pay to claim they are a CDP by next year.

Three pressures are driving the interest, and only one of them is about marketing results.

Data is scattered. A typical mid-sized business now runs a website, an email platform, a CRM, an advertising stack, a support desk and a commerce or billing system, each with its own idea of who the customer is. Nobody can answer a simple question like "how many customers who bought in Q1 have we emailed since."

Privacy rules arrived. GDPR took effect in May and California's Consumer Privacy Act was signed in June, taking effect in 2020. Both assume you can find every record about a person and delete it. If customer data lives in nine places, that is a project every time someone asks.

Walled gardens tightened. Facebook and Google want advertisers to bring their own first-party audiences, and the match rates reward clean, complete customer lists. Teams with fragmented data are building audiences from whatever export happens to be handy.

All three are real. None of them requires a six-figure platform if the underlying plumbing is fixed first.

Before buying, see what you can get from three building blocks that most businesses already have or can add cheaply.

A cloud data warehouse. BigQuery, Redshift or Snowflake can hold every customer record and event you generate for a few hundred dollars a month at typical mid-market volumes. This is the foundation whether or not you add a CDP later.

Pipelines into it. Managed connectors such as Stitch or Fivetran copy data from your CRM, email tool, ad platforms and commerce system into the warehouse on a schedule, with no code to maintain. Add a lightweight event collector on the website and app so behavior lands there too.

A reverse path out. The piece that is still awkward in 2018. Getting a warehouse segment into Facebook Custom Audiences or an email list usually means a scheduled script or a small integration service. It is not hard, but it is custom work, and this is the gap CDPs sell into.

With these three you have unified storage, SQL-defined segments and a privacy lookup that takes one query. What you lack is marketer self-service and sophisticated identity resolution. Decide whether those two are worth the difference in price.

Score yourself on the questions below. Three or more yes answers and a CDP is probably justified. Fewer than three, invest in integration and revisit in a year.

Note what is not on the list: the size of your customer base. A business with 20,000 customers and six data sources has a stronger case than one with two million customers and a single commerce platform.

Pricing is opaque and volume-based, but ranges are consistent enough to plan with.

• Developer-oriented tools such as Segment start around $120 a month for modest event volumes and climb into the thousands as traffic grows. They do collection and routing well and identity resolution modestly.
• Enterprise CDPs typically quote $50,000 to $300,000 a year depending on profiles, events and connected destinations. Implementation partners charge a similar amount again in year one.
• Suite add-ons from the large marketing clouds are priced as modules, often attractive on paper if you already own the suite and expensive if you do not.

Beyond the invoice, budget for the work no vendor includes: deciding what a customer is, agreeing a canonical identifier, cleaning the CRM before it is ingested, and documenting consent so the privacy promise is real. These usually take longer than the technical setup, and skipping them is how a CDP becomes an expensive duplicate of the mess you already had.

• Buying the platform before agreeing the customer definition across teams.
• Connecting every source on day one instead of the three that drive a known use case.
• Treating identity resolution as automatic; it needs rules, thresholds and review.
• Letting the vendor's demo use case become the roadmap.
• Forgetting the reverse path: a unified profile nobody activates is a reporting tool.
• Ignoring consent fields, then discovering the segments cannot legally be used for email.

We approach customer data as a technology and growth problem together, because a profile is only valuable when a campaign, a page or a conversation uses it. Discover: we inventory every system that holds customer data, how each identifies people, and which questions the business cannot currently answer. Plan: we write the customer definition, choose the warehouse-first or CDP route with the decision rule above, and sequence sources by the use case that pays first. Build: we set up pipelines, identity rules, the segments that matter and the activation path into advertising and email, with consent handled from the start. Improve: we measure whether unified data changed results, such as match rates, retention campaigns and reporting time, and extend from there. Brand, growth, web and technology stay connected so the data work shows up in what customers see.

A representative engagement is described in our digital transformation work, where a business with customer records spread across commerce, support and marketing systems moved to a single warehouse-based profile with scheduled activation into its advertising and email tools, before deciding whether a packaged CDP was warranted. The lasting result was a shared definition of a customer and a reporting layer every team trusted.