Quick Summary
Email attribution means assigning conversion credit to specific emails in a sequence or broadcast. To set up a model, pick a rule such as last-touch, set a conversion window that matches your sales cycle, and tag every link with the same UTM parameters every time.
- Last-touch is the easiest place to start, mostly because you can verify it by hand.
- Match the conversion window to your sales cycle. A trial sign-up usually wants something in the range of 3 to 7 days.
- Pick one UTM naming convention, keep the values lowercase, and then leave them alone.
- Clicks are a touch. Opens are not, because open tracking does not report reliably.
- Click your own tracked link and finish the conversion yourself before the first real send.
What an Email Attribution Model Measures
An email attribution model is a rule that decides which email gets credit when a contact converts.
Email attribution is the practice of assigning conversion credit to specific emails in a sequence or broadcast, so you know which messages actually move people toward a purchase or a reply.
Without a model, the numbers drift toward whatever story you already believe. The newsletter that happened to land on the Tuesday someone bought looks heroic, and the four-email sequence that did the persuading two weeks earlier gets nothing.
Start by listing the conversions you care about. A booked call. A trial sign-up. A paid upgrade. Then decide which email touches count toward each one.
First-Touch vs Last-Touch Attribution: Choosing the Starting Point
First-touch attribution gives all credit to the first email a contact engaged with. Last-touch gives all credit to the final email before conversion.
Use first-touch when the question you are asking is which emails bring new people in, rather than which ones close them.
| Model | Credit Goes To | Best For | Main Risk |
|---|---|---|---|
| First-touch | First email clicked | Judging acquisition emails | Ignores nurturing |
| Last-touch | Final email before conversion | Judging closing emails | Ignores early work |
| Linear | Equal share to every touch | Long sequences | Overvalues weak touches |
For most small teams, start with last-touch. It is not the most accurate model, but you can open two records, look at the dates, and see whether the answer is right. That matters more than theoretical accuracy at the stage where nobody trusts the dashboard yet. Whatever window you settle on, write it on your email KPI dashboard so every reader works from the same definition.
Switch to a multi-touch view once you have a few months of data and want to see the full path.
Defining Conversion Windows and Touchpoint Rules
A conversion window is the number of days after an email touch during which a conversion still counts toward that touch. A 7-day window means a click on day 1 earns credit only if the purchase happens by day 8.
Set the window to match your sales cycle:
- Short cycles, such as a free-trial sign-up: 3 to 7 days (a paid upgrade at the end of a trial needs the trial length plus a few days)
- Mid-length cycles, such as demos: 14 to 30 days
- Long cycles, such as enterprise deals: 60 days or more
Then define what counts as a touchpoint. Common rules include:
- A click counts as a touch
- An open alone does not count, since open tracking is unreliable
- A reply counts as a touch (many tools can stop the sequence on a reply)
- Unsubscribes never count as touches
Write these rules down before you build any reports. If you change them later, you can no longer compare this quarter to last quarter, and someone will eventually ask you to.
Setting Up Email Conversion Tracking
Conversion tracking connects an email click to a later action on your site, so the model has something to count. Adding UTM parameters is the easy half. What decides whether the attribution ends up usable is the naming system and the event list you commit to before the first send.
Build a UTM convention you will not have to rewrite
Every link in every email needs four parameters, spelled the same way every time. Case matters: Email and email land in separate rows in most analytics tools.
| Parameter | What it identifies | Example value | Rule |
|---|---|---|---|
utm_source | The send type | sequence or broadcast | Splits 1:1 from bulk in reports |
utm_medium | The channel | email | Always lowercase, always the same word |
utm_campaign | The sequence or broadcast name | welcome_seq | Use the tool's name for it, lowercased, with underscores for spaces |
utm_content | The step or variant | step2 or variant_b | One value per link, never combined |
A finished link looks like ?utm_source=sequence&utm_medium=email&utm_campaign=welcome_seq&utm_content=step2. Keep a one-page sheet listing every campaign name and step label you have ever used.
One failure mode worth guarding against: if you reuse a campaign name for a relaunch, the new sends merge with the old ones in reporting, and there is no way to pull the two periods apart afterward.
Define the conversion events before you tag anything
List every action that counts as a conversion and give each one a stable name. A workable starter set for a B2B team:
trial_signupdemo_bookedpricing_page_view(a soft signal, not a conversion on its own)paid_upgradereply_received(logged manually or by the sending tool)
Keep event names lowercase with underscores, and never rename an event after it has collected data. Renaming breaks the historical series and your month-over-month comparison.
Connect the click to the event
There are three common ways to join an email click to a later conversion, and they differ in how much they survive privacy restrictions:
| Method | How it works | What it survives | Main limitation |
|---|---|---|---|
| UTM on the landing page | The parameter rides the URL into the session | Cookie blocking, since it is first-party | Lost on a later visit unless your analytics tool keeps it in its own cookie (which Safari caps at 7 days) |
| Hidden form field | The UTM value is copied into a field on your form and stored with the lead | The gap between the form submit and the deal closing | Empty if the form is sent on a later visit, unless the value is also kept in a cookie |
| First-party cookie or local storage | The click is stored on your own domain | Third-party cookie blocking | Cleared by the user, and cut short by some browsers (Safari expires storage set by script after 7 days) |
A hidden form field can write the campaign value onto the contact record at the moment of conversion. Ask your web developer to save the UTM values in a first-party cookie when a tagged link lands, overwriting them on each new tagged visit, and copy them into a hidden field when the form is sent. That field gives you last-touch. For first-touch, add a second cookie that is written once and never overwritten, copied into its own field. In Safari a cookie set by script lasts at most 7 days, so first touches older than that are lost there unless your own web server sets the cookie. With both fields on the record you can run first-touch and last-touch side by side.
Set the tracking you can actually trust
Open tracking is unreliable and should never be the only touch you count. Apple Mail Privacy Protection loads the tracking pixel for the reader, so contacts who never looked at the email can still show as opens, and some inboxes block images so real readers never register. Clicks are not perfectly clean either: some company mail filters open every link to scan it, often within a minute of delivery. A click with no matching visit in your analytics is often a scanner (or a reader whose browser blocks your analytics), so check a few before you trust a campaign that wins on clicks alone. Read opens next to clicks and replies rather than on their own. For how this plays out when you send from a personal Gmail account, see how open tracking behaves in personal Gmail.
Once clicks and conversions are joined, you can turn conversions into a return figure. For the formula, see the calculation for email marketing ROI.
Before you send the first tracked campaign, run one test: click your own link, complete the conversion, and confirm the campaign name appears on the resulting record. Once that works, the attribution will start showing you where leads stall, which is a different problem from measuring them. This rundown of lead conversion fixes covers that side of it.
Building a Multi-Touch Model Across Email Sequences
A multi-touch model spreads credit across several emails. The linear version gives each touch the same share. It is crude, and it is also where almost everyone starts, because it is easy to explain to someone who did not build it.
Weighted versions shift the shares. A position-based model gives more of the credit to the first and last touches and splits the rest across the middle; a time-decay model gives more to the touches closest to the conversion. Pick the weights before you look at the results, or you will end up choosing the weights that flatter the email you already liked.
Sequence logic changes which touches a contact sees in the first place, which is easy to forget when you are reading the report rather than the sequence. A contact on a branch that skipped step 3 can never give step 3 credit. For a deeper look, see how branching logic works inside a sequence. Engagement scores answer a different question, which contacts to follow up with first, and how to use engagement scores to prioritize leads covers that.
Keep the branching shallow. Every extra level adds paths, and with each one it gets harder to trace which emails a contact actually received.
Resolving Conflicts Between Email and Other Channels
Email rarely works alone. A contact may read a sequence, then see a paid ad, then convert from a direct visit. Each channel wants the credit.
Pick one of the first two rules for overlap and apply it everywhere, and do the third regardless:
- Give email the credit only when the email click came within the conversion window
- Credit the channel that sent the most recent touch
- Report channel overlap as its own line, so no channel quietly takes the win
Write the rule into your reporting notes. Teams that change the rule mid-quarter spend the next review arguing about the numbers rather than doing anything with them.
Tracking Attribution Without Cookies or a CRM
Safari and Firefox block third-party cookies by default, ad blockers strip more, and plenty of small teams run email without a CRM at all. Neither problem makes attribution impossible, but both change which signals you can trust. Lean on first-party data you already own, and be honest about what it can and cannot prove.
Know which signals survive
| Signal | Where it lives | Survives cookie blocking | Survives a long gap before conversion |
|---|---|---|---|
| UTM on the landing URL | The session | Yes, first-party | Only if your analytics tool or your own cookie keeps it |
| Hidden form field | The lead record | Yes | Yes, if the form is the conversion |
| First-party cookie or local storage | Your own domain | Yes | Until cleared |
| Email platform click log | The sending tool | Yes | Yes, stored server-side |
| Self-reported source | A form question | Yes | Yes |
The email platform's own click log is stored on the sending side and does not depend on the browser at all. Pull the click date and the conversion date from two separate systems and join them on the contact's email address.
Use self-reported source as a cross-check
Add one question to your sign-up form: "How did you first hear about us?" Store the answer on the contact. It is imperfect, since people misremember, but it gives you a second reading to compare against the click data. When the two disagree badly for one campaign, the likelier explanation is broken tracking, not a failed campaign.
Reconcile against sales records
Keep a simple sheet with one row per closed deal: contact email, close date, the campaign name from the hidden form field, and a source note from the rep. Join it to the click log by email address.
If the rep's note and the hidden form field disagree on more than a handful of deals, stop and fix the form field. Every model you build on top of that field inherits the error.
Set expectations for precision
Cookieless data is partial. A contact who clicks on a phone, then converts on a laptop two weeks later, may never be joined at all. Treat the output as directional: use it to rank which emails sit closest to revenue, not to claim a precise revenue figure per send. Put that caveat in the report itself, so nobody downstream reads more precision into the number than the data supports.
Where the sending tool fits
SEMAOS sends the broadcasts and 1:1 sequences and records the clicks. Platform sends always track clicks; sends from a connected Gmail or Outlook mailbox track them only once you switch mailbox click tracking on and verify a custom tracking domain, which needs a paid plan. It does not see purchases, trial starts or deals on your site and has no attribution report, so the join to your sign-ups and deals happens in your analytics or your sheet. The exception is its own landing pages: a signup there saves the UTM tags of the link that visit arrived on, on the contact. Pages set no cookies, so it works like a hidden field read straight from the URL, not the cookie-backed one above. There is no export of individual clicks; a contact's page lists each email with its clicks, and each send's page shows when it was first clicked. SEMAOS also computes a nightly 0 to 100 engagement score per contact on paid plans and the trial. The score is built from opens, clicks and sequence activity, not conversions, and it does not include replies, so it is not a weight for attribution.
Your First Attribution Check This Week
Run one check this week:
- Pick one conversion event, such as a trial sign-up.
- Use the window you picked for that cycle, such as 7 days for trial sign-ups.
- List the last 30 days of sign-ups with their email addresses and dates.
- For each sign-up, pull the email clicks inside the window before it from your sending tool's click log, joined on the email address. Give last-touch credit to the latest click and first-touch credit to the earliest (first-touch inside a short window; the first-touch field gives the longer view). Sign-ups with no click in the window get no email credit. If your sending tool has no export of individual clicks, read them off each contact's record by hand; for a month of sign-ups that is slow but doable.
- Compare which email gets credit under each model.
- Note any email that wins under both models, since it probably matters most.
Repeat the check each month. Once you trust the numbers, you can decide which model to keep. For the sending and click tracking side, see what SEMAOS includes.
FAQ
Should I use first-touch or last-touch email attribution?
Start with last-touch. It is not the most accurate model, but you can open two records, check the dates, and confirm the answer by hand, which matters more than theoretical accuracy before anyone trusts the dashboard. Use first-touch when the question is which emails bring new people in, and move to a multi-touch view once you have a few months of data.
How long should an email conversion window be?
Match it to your sales cycle: 3 to 7 days for short cycles such as a free-trial sign-up (a paid upgrade at the end of a trial needs the trial length plus a few days), 14 to 30 days for demos, and 60 days or more for long enterprise deals. Write the window down before building reports, because changing it later makes quarters uncomparable.
Do email opens count as an attribution touchpoint?
No. Open tracking is unreliable and should never be the only touch you count. Count clicks and replies as touches, treat unsubscribes as never counting, and read open data only alongside clicks and replies.
Can you track email attribution without cookies or a CRM?
Yes, using first-party signals: UTM parameters on the landing URL, a hidden form field that stores the campaign value on the lead record, your email platform's server-side click log, and a self-reported source question on the sign-up form. Treat the output as directional ranking rather than a precise revenue figure per send.