CDP, what is it good for? – What do Customer Data Platforms do? Do you really need one – and if so, which type?

The CDP (Customer Data Platform) market is now so broad with such a range of scopes and capabilities that it’s increasingly hard to make sense of it as one vendor category.

Asking “Do I need to invest in a CDP?” is becoming akin to asking “Do I need to invest in a Transport Enabling Device?” – if that ‘TED’ category included wheels, engines, ropes and batteries alongside full-blown automobiles and yachts.

Set against that scope for confusion, the potential value of first party customer data is far clearer; the question then becomes which technology is best to bring that value to fruition?

In this article we’ll explore the practicalities of what different flavours of CDP do and don’t do, which use cases they support, and whether you in fact need any of them at all.

What (actually) is a CDP?

Often the simplest question is the most important! And in the case of CDPs, the answer can span a broad assortment of responses depending on who – or rather which vendor – you ask.

Functionally, CDPs typically purport to do all or some of the following:

1. Storage – act as a repository for data points associated with a prospect or customer (or related identifier).
2. Identify Resolution – merge the facts and behaviours that have been gathered about users when it becomes possible to identify them as being related to the same person.
3. Enrichment – act as a data clean room to extend first party data with third party data attributes via anonymised sharing.
4. Segmentation – build segments of customers or prospects to be targeted with (or excluded from) specific propositions, campaigns, communications or offers.
5. Activation – turn those segments into channel outcomes such as personalised website experiences, triggered emails and targeted advertising.

Some CDPs cover everything above and aim to become master data repositories for large portions of your customer data at their core. At the other end of the spectrum, some CDPs only do activation and lean on other existing platforms to do everything else – sometimes these are referred to as “composable CDPs”.

And then CDP-like functionality is increasingly baked into other CRM, CMS and analytics platforms.

Google Analytics 4 stores behavioural data, can resolve identity across mobile apps and websites, segment and then activate audiences within Google Ads: it’s basically a CDP, albeit one very much focused on activating a subset of data within the Google advertising ecosystem.

Your email marketing platform can upload customer lists to Meta for advertising – it’s becoming a CDP.

Your content management system tracks and segments behaviours and uses it to drive personalisation – it’s joined the CDP party too.

Architecturally, in a modern tech stack there are so many platforms that are incident on customer data that conceptually pretty much anything can call itself a CDP.

Which brings us on to…

Use Cases

If use cases for a CDP are not clearly defined and prioritised, then it’s impossible to evaluate what a smart technology investment might be in this space.

Good use cases define how specific data drives specific outcomes, using clear language of how the customer is impacted. For example:

• “Exclude existing customers from our Meta acquisition advertising”
• “Trigger reactivation emails when a customer has not been active on the website or mobile app for a month”
• “Instigate customer service follow-up calls when high value customers abandon an online support process”

Even better use cases put a potential business value on the outcomes, and hence can help with prioritisation.

Less good use cases focus on generic capabilities as opposed to specific customer outcomes, and might deploy industry buzzwords that are conveniently open to interpretation:

• “Drive real-time omni-channel personalisation”
• “Provide a single view of the customer”

Gap Analysis

Armed with use cases, any evaluation of CDP capabilities can then be focused on what you’re specifically looking to enable, and what the functional gap is in terms of storage, identity resolution, enrichment, segmentation and/or activation.

For example:

• If your CRM already has a segment of customers you want to exclude from advertising, and that CRM also has a connector to Meta to upload it, then you don’t need a CDP at all (for that use case).
• If you already have a data warehouse of behavioural data from web and mobile app usage (e.g. Google BigQuery), then you perhaps only need some activation CDP capability to connect that to your email marketing platform for triggered emails.

There is also a strategic gap analysis in relation to other technology plans. For example, if the organisation is already building out a data warehouse with segmentation capabilities in Databricks, then you may not want to be building out what will essentially become a duplicate data warehouse with segmentation capabilities in a Oracle CDP.

Finally, one attribute to probe clearly is the one of “real time”. Ultimately nothing is real time – there’s always an acceptance of whether it is tolerable to wait milliseconds, seconds, hours or indeed a day for a changing data point in one place to manifest into a change in treatment for a customer elsewhere. But some data points don’t change that quickly (e.g. customer age), or can be potentially better localised in the channel in which they occur (e.g. personalising a page based solely on the previous page viewed).

Segmenting The Market

Just like understanding the diversity of your own customer base, segmenting the CDP market is a good way to break down their capability focus.

To aid clarity (or controversy), the quadrant map below adds some keywords to the Customer Data Platform labels to emphasise the practical focus of these systems, with some examples (non-comprehensive) of vendors operating in these segments.

The Classics (Put all your Customer Data in our Platform)

Sometimes referred to as “Legacy CDPs” by their composable colleagues, a more polite emphasis is to set these out as platforms that aim to do the full spectrum of storage, identity resolution, segmentation and activation. Critically the starting point tends to be onboarding substantial amounts of data into the platform, which can result in a duplication of storage versus those source platforms (especially if one of them is an existing data warehouse).

The Enrichers (Match your Customer Data with our data Platform)

Often not dissimilar to the Classics in functionality, but with added data clean rooms and proprietary third-party data, so a big focus on enrichment (and a desire to accumulate as much client data as possible to enable that). Often have grown out of previous Data Management Platforms that did similar enrichment based on cookie data for advertisers.

The Orchestrators (Activate your Customer Data in our Platform)

An overlap with marketing automation means some platforms can major on the orchestration of triggered marketing, especially through direct response channels such as email or SMS, as part of the CDP proposition.

The Composables (Get your Customer Data out of an existing Platform)

Whereas the Composable CDPs are typically more focused on activating data from an existing data warehouse via external channels and tools, operating essentially as connectors to synchronise data from A to B. The theory being you can make use of your existing technology platforms for storage and plug a far more specific capability gap with these; some will offer segmentation and identity resolution as optional modules.

Pricing

If you are using the full gamut of CDP capabilities, rate cards can be complex, and anticipating what your “data tax” might end up as can be especially challenging, even with a calculator.

Even for composable (activation only) CDPs, the pricing models can vary quite significantly. Some charge by the connector, some by the number of times those connectors are “synchronised”, some by the number of underlying customer records irrespective of connector usage, some at a completely flat fee irrespective of anything else.

This all makes the ROI equation… complicated. But far less complicated if you have mapped out your use cases and can therefore cost some very specific scenarios and see how the costs scale in practice. In fact, without doing that, you could be signing a blank cheque.

Other Gotchas

“We integrate with X hundred connectors” – but what kind of connector? Dig into the documentation for some CDPs and that “connector” is a batch SFTP upload rather than anything more real time into an API. That might be all you need but definitely probe the capabilities and limitations of the connectors you know you will be leaning on to enable your use cases.

Some CDPs (here’s looking at you, Adobe) require you to map your data into a proprietary schema before you can start working with it. If their schema fits your business model and data sources, great, if not, it could be a substantial challenge and overhead to shoe-horn into.

The Non-CDP

Extract Transform and Load (ETL) is the process of getting data out of an API and into a data warehouse. And Reverse ETL is the process of getting it out of a data warehouse and into an API. What’s the difference between a Reverse ETL Solution and a Composable CDP? Potentially not a lot, apart from the name. And increasingly, cloud data warehouses may integrate their own reverse ETL connectors and reduce the market for composable CDPs as a result.

To Sum Up

Should the CDP vendor category even exist anymore?

Perhaps not: cloud data warehouses are getting increasingly functional across areas that CDPs traditionally covered as independent products. And the direction of travel is both towards far more point solutions for specific capabilities (e.g. activation) alongside CDP-like functionality being integrated into a myriad of other experience and marketing automation platforms.

Stepping back, we all need to join the dots with customer data and focus on what makes it valuable – to us and to them. Nail your specific use cases and it becomes much easier to identify what capabilities across that storage, identity resolution, enrichment, segmentation and activation are required, and what is missing. Let that gap analysis firmly set the scene for how you look at solutions in this space and you’re far less likely to end up mis-sold or lost.

As an independent consultancy, we’re always here to help navigate the market, help you make the most of what you’ve got already and make smart choices for now and the future.