Customer Journey: Why Your CRM Lies About What Customers Actually Do

Pull a customer record from your CRM. You'll see something like this: created from form submission on March 14, opened welcome email, clicked CTA, purchased product on March 19. Tidy. Linear. Useful for reporting.

Here's what actually happened. The same person Googled your product on March 4. Read a review on Reddit. Came back via direct visit on March 7, bounced after 8 seconds. Saw a Facebook ad on March 11. Clicked a competitor comparison on March 13. Filled the form on March 14 from a paid search ad — not the source the form attributed to. Opened the welcome email on a phone, then converted on March 19 from a desktop browser two cities away.

Your CRM saw step 4 onwards. Steps 1 through 3 happened in someone else's system, if they were captured at all. The result is that your CRM tells you a coherent story that is, in important ways, fiction.

Customer Journey is the feature that fixes this. It treats acquisition-to-conversion as one continuous story per identity, stitches every session and channel into a single timeline, and runs five attribution models on top so you can see who actually drove the conversion. Most SMB tools don't have it. The few that do (Bloomreach) are enterprise-priced — five-figure annual contracts and weeks of implementation.

The Identity Stitching Problem

The hard part of customer journey isn't the visualization. It's the stitching. A real customer leaves traces under a dozen different identifiers:

• Anonymous browser fingerprint on visit 1
• Different anonymous fingerprint on visit 2 (cleared cookies, different device)
• Email when they sign up
• Customer ID after first purchase
• Push token on mobile
• Phone number after they enter SMS opt-in

Stitch them right and you have one journey. Stitch them wrong and you have six fragmented profiles, each with a partial story, none of which matches the actual person.

The standard trick is the identify() call. When the SDK sees an anonymous visitor enter an email or sign in, it rewires all their prior anonymous events to the now-known customer ID. Done well, it preserves the earliest acquisition source — which matters enormously for first-touch attribution. Done poorly, the email becomes the "first touch" and you lose months of pre-conversion data.

The right architecture: older anonymous ID always wins in cross-device merges. A separate rootAnonymousId stays stable across identify() calls, so the journey row survives the anonymous-to-known transition without creating a duplicate.

Channel Resolution: Where Did They Actually Come From?

Once identity is stitched, the next question is: for each session in the journey, what channel drove it? Direct visit? Paid search? Organic social? Email click?

The right answer is a deterministic waterfall, evaluated at session insert (not later, because referrer data degrades over time):

1. EMAIL — if UTM source is your email tool, or special parameters like sm_eid/sm_cid are present
2. Paid search — if gclid/msclkid in URL, or paid-search UTM medium
3. Paid social — if fbclid/ttclid/li_fat_id, or paid-social UTM medium
4. EMAIL (again) — if UTM medium is "email" or referrer is a known webmail domain
5. Organic social — if referrer host matches Facebook, Twitter, LinkedIn, TikTok, Reddit, etc.
6. Organic search — if referrer host matches Google, Bing, DuckDuckGo, Yandex, Brave
7. INTERNAL — if referrer matches one of your own domains
8. REFERRAL — referrer present but none of the above
9. DIRECT — no referrer, no UTM

This resolution happens at session start, atomically. The reason is reliability. If you wait until aggregation time, the referrer is gone, the UTM has been clipped by a redirect, and you're guessing. Resolved upfront, the channel column is queryable forever.

The 5 Attribution Models

Once you have a clean journey with channel-tagged sessions, you can ask the question every marketer cares about: which channel deserves credit for the conversion?

There is no single correct answer. There are five reasonable answers, and the right one depends on what question you're asking.

The reason you want all five, not just one, is that they answer different strategic questions. First-touch tells you which channels generate new audience. Last-touch tells you which channels close. The two often disagree dramatically — and the disagreement is the insight.

Concrete example. Run First-touch and you might see organic search drives 60% of conversions. Run Last-touch on the same journeys and you see email drives 70%. Both are true. The honest reading: organic search builds the pipeline, email closes it. Cutting your SEO budget because email "drives the conversions" would be a disaster.

Stage and Status: Where Is This Customer Right Now?

Beyond credit-splitting, a journey gives you a position. Where in the funnel is this person, today, given everything they've done?

The classic stages, plus a status that's independent:

Stage (where in the funnel):

• AWARENESS — first or second session, browsing top-of-funnel pages
• CONSIDERATION — multiple sessions, deeper engagement
• DECISION — signals like form submit, add-to-cart, checkout-started; or any session URL hitting /checkout or /cart
• RETENTION — converted, plus activity after the conversion

Status (movement):

• ACTIVE — recently engaged, on track
• STUCK — in CONSIDERATION but no movement for 7+ days
• ABANDONED — no activity for 30+ days, no conversion
• CONVERTED — closed

The combination is what unlocks targeting that wasn't possible before. "Show this popup to anyone STUCK in CONSIDERATION for over 14 days" is a one-line segment definition. It would have required a custom data pipeline three years ago.

Journey-Based Triggers: What to Do With This

Capturing the journey is half the value. Acting on it is the other half. Three kinds of triggers fire on journey events:

Journey stage changed — fires the moment a customer moves from CONSIDERATION to DECISION, or AWARENESS to CONSIDERATION. Use it to send a stage-appropriate email: "we noticed you're comparing options, here's a head-to-head," or "thanks for showing interest, here's a getting-started guide."

Returning after N days — fires when a customer comes back after a configured gap. Useful for re-engagement campaigns or "welcome back" messaging that's actually conditional on the gap, not just calendar-driven.

Abandoned at journey stage — fires when a customer is classified ABANDONED while sitting in a particular stage. The treatment for abandoned-at-DECISION (lost a near-conversion) is very different from abandoned-at-AWARENESS (never engaged anyway). The trigger lets you send the right message based on how far they got.

Combine these with the channel adapter system and you can: trigger an SMS when a high-value customer enters DECISION, push a popup when a STUCK consideration-stage visitor returns, send a Slack notification to your sales team when a journey hits AWARENESS → CONSIDERATION on the enterprise pricing page. Any journey event → any channel.

Why This Is the Moat

Your CRM doesn't have this. Klaviyo doesn't have this beyond email-only attribution. Hotjar doesn't have it because Hotjar can't trigger campaigns. Bloomreach has it — for an enterprise contract and weeks of implementation.

The reason is engineering complexity. Customer Journey requires:

• A behavioral SDK to capture sessions across devices
• An identity stitching service that handles the anonymous-to-known transition
• A channel resolver that runs at session insert
• An aggregation engine that rebuilds journeys idempotently
• Five attribution models with proper edge-case handling (clock skew, single-touchpoint, half-life tuning)
• A stage classifier with sane thresholds
• A trigger system that fires on journey-scale events with cooldowns to prevent flow storms

That's months of engineering work for a team. It's why no SMB platform has shipped it. It's also why, when one finally does, it's a moat — not because customer journey is exotic, but because it's so much work that competitors will be years behind even if they start now.

What to Do With This

If you've been running marketing reporting on email opens and last-click attribution, you're working with a small slice of what your customers actually did. Switching to a tool with proper customer journey isn't a feature upgrade — it's the difference between guessing and knowing.

The first thing you'll discover is which of your channels is undercredited. The second is how long your real consideration window is (almost always longer than you think). The third is how many of your "first-touch" attributions are wrong because they were assigned at the email signup, not the first actual session.

None of those discoveries are theoretical. They reshape how you allocate budget, how you sequence campaigns, and which channels you double down on. Which is the actual point of attribution — not to win arguments about credit, but to make better resource decisions.