Customer Data Platform

6 tools, 6 different IDs.
All the same human.

A CDP — Customer Data Platform — stitches together the fragmented identities flowing from your website, mobile app, CRM, commerce engine and contact centre into a single real customer profile. It stops one human existing as five different rows in five different systems; it creates one source of truth for marketing, sales and service.

Segment, RudderStack, Tealium, mParticle or a composable CDP — we pick the right foundation with you and implement it. The identity-resolution problem that supposedly takes years gets cleanly solved in 8-12 disciplined weeks.

Request CDP audit 8-week plan

Email [email protected]

Phone +90 532 ••• 47 12

Device ID aa-bb-cc-9f1

CRM User #38421

CDP
ID GRAPH

Ali Yılmaz

cdp_id: 0x7f4a…b8

LTV₺ 18.420

Orders14

SegmentVIP-Loyal

Churn12%

Last seen2h ago

ChanneliOS app

events.live· 247/s

12:04:11page_view/products/348

12:04:12add_to_cartsku:JK-104

12:04:14identify[email protected]

12:04:15purchase₺ 1.249,00

12:04:17email_opencamp:VIP-44

12:04:19app_openiOS 17.4

12:04:21searchkış botu

12:04:23consent_grantmarketing

12:04:11page_view/products/348

12:04:12add_to_cartsku:JK-104

12:04:14identify[email protected]

12:04:15purchase₺ 1.249,00

12:04:17email_opencamp:VIP-44

12:04:19app_openiOS 17.4

12:04:21searchkış botu

12:04:23consent_grantmarketing

5 IDs → 1 profile

Why fragmented data is fatal

6 tools, 6 different customer IDs. Blind to the same human.

Your customer visits your website, buys in-store, uses the mobile app, opens your email, scans the loyalty card, calls support. These 6 interactions are saved in 6 different systems under 6 different IDs. Marketing hits the same person with 3 conflicting campaigns; sales doesn't know the history; support asks "is this your first call?". Without a CDP, customer 360 is only a slide title — it does not exist in real life.

A different ID per channel

Web cookie, mobile device_id, email hash, CRM ID, POS number — the same human lives as five rows in five systems. Without a stitching rule, nobody recognises the same person.

Duplicate campaigns

A coupon from Mailchimp, another from Klaviyo, a third from Meta. The same customer receives three conflicting offers in one week; brand trust erodes, discount margin melts.

Wrong attribution

Each ad panel claims the same conversion. Summed up, revenue appears tripled. Budget shifts to the wrong channel; true contribution disappears.

Hollow recommendations

"Items for you" still pushes what they just bought, because the order system isn't wired to the recommender. Personalisation becomes annoying spam.

Service quality

The customer says "I'm calling about last month's order"; the agent asks everything again. NPS drops, CAC becomes waste, churn balloons.

KVKK / GDPR risk

When a customer asks "delete all my data", five systems get searched manually. Miss one — legal exposure. Single-point deletion is mandatory.

The stitching flow

Sources unify at one point. Destinations receive clean data.

Your sources (Shopify, Salesforce, Web Pixel, Mobile SDK, contact centre, loyalty system) flow into the CDP; the CDP performs identity resolution + enrichment; a single clean profile is piped continuously to destinations (Mailchimp, Klaviyo, Meta Ads, Google Ads, Slack, Webhooks, BigQuery). In the modern "reverse ETL" stage, the CDP also pushes warehouse data back into operational systems — for recommendations, personalisation and service.

Sources

Shopify

Salesforce

Web Pixel

Mobile SDK

POS / Loyalty

Call Center

CDP Identity Graph

Destinations

Mailchimp

Klaviyo

Meta Ads

Google Ads

Slack alert

Webhook / BigQuery

Without CDP — chaos

6 tools, 30 point-to-point integrations — 30 maintenance jobs.

With CDP — clean star

6 tools, 6 integrations — controlled from one point.

Identity match rate

A well-built identity graph yields 78% deterministic + 18% probabilistic match.

Deterministic = email, phone, customer no — exact. Probabilistic = same IP + same device + similar behaviour. Most businesses start at 38% deterministic; with the right schema and event design they reach 78% in 90 days.

Deterministic match 78%

Probabilistic match 18%

Unmatched (anon) 4%

Cross-device consolidation 71%

Anon → known conversion 54%

Who it's for

Every team that sees the same customer across systems — a CDP relieves them.

E-commerce · web + app + offline

One customer buys across 3 channels; without a CDP no behavioural segmentation, no lookalikes.

SaaS · multi-product

Core + add-on + mobile — same user; in-product events + billing + support split across systems.

Marketplace

Buyer + seller + supplier — three personas with shared payments and messaging; one profile is mandatory.

Financial services

KYC + application + call + branch — regulated channels; unified profile + consent tracking.

Telco · GSM + fibre + IPTV

4 lines, 4 bills, 1 household — household-level identity graph enables cross-sell and churn modelling.

Omnichannel agency

Agencies serving web + email + paid + loyalty get one clean data plane; campaigns ship faster.

Healthcare

Patient + appointment + tests + prescription; EHR + CRM + web — GDPR/KVKK-compliant Customer 360.

B2B · long cycles

Lead + opportunity + product use + support; one account profile for account-based marketing.

10 capabilities, one team

From identity to consent, schema to activation — the 10 real layers of a CDP.

A CDP is not just a tool installation. Schema design, identity rules, consent layer, real-time event architecture, destination management and auditable governance — all of it works together. Partnerfy runs the 10 layers as one team.

Identity resolution

Deterministic + probabilistic identity graph; cross-device + cross-channel stitching; household + account roll-up.

Real-time event ingestion

Web, mobile, server, batch — tens of thousands of events per second with schema validation.

Audience builder

No-code UI for behaviour + attribute + event-based dynamic segments; live counts.

GDPR / KVKK consent layer

Central consent management; each destination filtered by consent; "delete all my data" with one click.

Destination piping

Server-side + client-side delivery to 200+ marketing, sales, BI and operations tools.

Reverse ETL · warehouse → CRM

Enriched data from Snowflake / BigQuery pushed back to Salesforce, HubSpot, Klaviyo.

Predictive scoring

Churn score, LTV prediction, purchase probability — from warehouse models into live segments.

Source connectors

Shopify, Magento, Stripe, WooCommerce, Salesforce, HubSpot, Zendesk + custom webhooks.

Tracking plan & schema

Event dictionary, naming convention, parameter types, validation — the foundation for clean data.

Governance & audit

Who saw which data, which destination it flowed to; full audit log, PII masking, role-based access.

Process

From audit to activation: an 8-12 week disciplined roadmap.

01

Week 1-2 · Source audit

Web, mobile, CRM, commerce, POS, loyalty, contact centre — every system's events, fields and frequencies are documented; missing / duplicate / wrong fields are listed.
02

Week 2-3 · Identity graph design

Which fields are deterministic (email, phone, customer_id), which probabilistic (device, IP, fingerprint); priority and tie-breaking rules written.
03

Week 3-4 · CDP selection

Segment, RudderStack, Tealium, mParticle, or composable on Snowflake/BigQuery — decided alongside event volume, budget, team level and regulatory needs.
04

Week 4-8 · Wiring the pipes

Source connectors + tracking plan + schema + identity resolver + consent layer — taken to production; staging validation rounds run.
05

Week 8-10 · Building audiences

In the no-code audience builder pilot segments (e.g. 14-day cart abandoners, 30-day dormant VIP, high-churn risk) are designed and live counts validated.
06

Week 10-12 · Activating destinations

Mailchimp, Klaviyo, Meta, Google, Slack, webhook, BigQuery — opened in sequence; each with consent + dedup + frequency cap; first real campaign sent.

CDP platforms we work with

There is no single right CDP. We build the one right for you.

Segment RudderStack Tealium mParticle Treasure Data Hightouch Census BlueShift Lytics Iterable CDP Snowflake-based CDP BigQuery n8n

Warehouse

Snowflake / BigQuery / Redshift

Modelling

dbt + SQL

Activation

Hightouch / Census

Event ingestion

Segment / RudderStack

Composable CDP: warehouse + dbt + activation instead of a sealed box = zero licence, full control.

Client stories

Same CDP spine. Different sectors. Same outcome: a unified customer.

E-commerce CAC −31%

Lifestyle brand

Lookalike + suppression segments built on the CDP cut Meta + Google CAC by 31%; ROAS 2.4× → 3.8×.

B2B SaaS 2× trial→paid

Vertical SaaS

In-trial behaviour + email engagement + CRM stage unified in the CDP; trial-to-paid 4.2% → 8.6%.

Telco 4 brands 1 profile

4-brand group

GSM, fibre, IPTV, content — household identity graph built; cross-sell conversion 2.1% → 6.4%.

Marketplace GMV +18%/mo

Online marketplace

Buyer + seller in one profile; dormant seller reactivation automated, GMV up 18%/month.

Finance 0 audit findings

Insurance carrier

Web + call + agent + app; centralised consent + deletion — 0 findings in audit.

Healthcare NPS 42→71

Private hospital group

Patient + appointment + tests + prescription unified; service NPS 42 → 71.

FAQ

The 8 questions decision-makers ask before buying a CDP

These are two different layers and they don't replace each other. A CRM holds the relationship records your sales team enters manually: who called, who said what, which deal sits in which stage. A CDP collects all the behavioural data flowing passively: web pages, in-app events, email opens, product usage, purchases. CRM manages "people you've met"; the CDP unifies every touchpoint from the anonymous visitor you haven't met yet to the customer you know well. The most efficient architecture is the two synced: the CDP enriches behavioural signal, writes segments back to the CRM, and the sales rep sees the full story on one screen.

Segment is the most mature enterprise CDP in the world; vast destination ecosystem, strong governance, Twilio Engage as a messaging layer on top. Licence cost scales fast with event volume — high-volume e-commerce reaches USD 10-50k/month. RudderStack is open-source-based with a self-hosted option; your data stays in your own cloud, ideal for KVKK / data residency; licence is dramatically lower but the team needs to be more technical. Switching from one to the other is doable but takes 4-6 weeks. In the decision matrix event volume, regulatory needs, in-house technical capacity and budget all weigh in; we run the numbers with you.

A classic CDP (Segment, Tealium, mParticle) is sold as a single box: event collection, identity stitching, segmentation, destination delivery — all hosted by the vendor with your data inside their stack. A composable CDP is the approach of "my data already lives in Snowflake / BigQuery / Redshift; all modelling is done with dbt; if I add an activation layer (Hightouch, Census) on top I assemble the CDP from parts". The upside: licence cost drops dramatically, data stays in one source of truth, engineering works with their existing tools. The downside: the marketer's self-service UX isn't as polished as a classic CDP. Which architecture fits depends on your warehouse maturity, BI capability and marketing team's technical level.

For KVKK and GDPR, PII covers name, email, phone, national ID, IP, device ID, precise location. Inside the CDP these fields sit in a separate vault, hashed and salted; outbound flows to destinations are filtered by consent and masked where required. When a customer asks "delete all my data", a single API call triggers deletion in both the CDP and every connected destination, and the action is logged for audit. Sensitive categories (health, religion, finance) get additional access controls and bespoke retention. A correctly built CDP answers "who saw what, where did it flow" in seconds during an audit — that is the difference between a clean inspection and a fine.

A CDP's real value becomes obvious above 50-100k users; the typical B2C threshold is 100-500k active customers and 50M+ events/month. For smaller brands the licence cost of a classic CDP doesn't balance the benefit — at that point the composable approach (Snowflake + Hightouch) or a simple ETL + CRM strategy makes more sense. In B2B the numerical threshold can be lower; 5-10k active accounts + long sales cycle + multiple products often surfaces CDP ROI early. On data maturity, a minimum: event tracking must have been running at least 3 months; brands starting from zero should invest in a tracking plan first, then the CDP.

In classic CDPs the monthly licence is tied to event volume: USD 1.5-5k/month for 5-10M events, USD 10-50k/month or more for 50M+. Self-hosted RudderStack or open-source variants only pay the cloud cost (USD 200-2,000/month). With composable CDP, if the warehouse (Snowflake/BigQuery) already exists, the marginal extra is just the Hightouch / Census licence (USD 200-2,000/month). One-off setup: small-to-mid brands 8-12 weeks + USD 20-50k; enterprise projects 12-20 weeks + USD 60-150k. ROI typically arrives in 4-9 months through ad CAC reduction, email conversion lift and operational savings.

Yes — but in a different architecture: composable CDP. If the warehouse already exists, data shouldn't be re-copied into a CDP box; that is wrong both for cost and for governance. Instead the warehouse stays as the single source of truth; on top of it you place event ingestion (RudderStack, Segment) + an activation layer (Hightouch, Census); modelling is done with dbt. In this architecture marketing builds audiences in the UI while engineering works on a single data plane. If your warehouse isn't yet mature (no event table, no dbt, broken production reports), a classic CDP gives faster ROI; we evaluate the fork with you.

For a typical mid-size brand 8-12 weeks is realistic: first 2 weeks for source audit and identity graph design; 2-3 weeks for CDP selection and contracting; 4 weeks for source wiring, tracking plan + schema, consent layer; 2-3 weeks for the audience builder and first pilot segments; final 2 weeks for activating destinations and the first real campaigns. Enterprise projects (telco, bank, multi-brand groups) stretch to 16-24 weeks because compliance, legal and IT-security approvals run in parallel. Pressure to "finish in 3 months" usually skips the tracking plan or the consent layer — short-term speed, long-term debt. A correctly executed 12-week project runs maintenance-free after month 12.

Finally see your customer as one profile.

A free 30-minute call to review your data sources, identity strategy and CDP requirements; we share the best-fit architecture (classic vs composable) and the 3 segments to win in the first 90 days.

Book a call Contact us

6 tools, 6 different IDs. All the same human.