Lead Attribution: Models, Setup Fields and a Lead Value Example
Lead attribution is the process of recording which marketing channels, campaigns and pages a prospect interacted with before becoming a lead, then giving those touchpoints credit for the lead and for the revenue it later produces. Done well, it tells you which sources bring leads that actually close, not just which ones fill your form inbox.
This guide covers the models, the exact fields to store on every lead, a seven-step setup that works with any CRM, a worked example that turns leads into dollars per source, and the setup mistakes that quietly corrupt the data.
What Is Lead Attribution?
“Attribution” simply means assigning credit for an outcome to its causes. In marketing and sales, the outcome is a lead, an opportunity or a closed deal, and the causes are the ads, searches, posts, emails, referrals and conversations that came before it.
Lead attribution narrows that to one question asked about every new lead: where did this person come from, and what got them to raise their hand? In practice that means three things:
- Capture: record the source, medium, campaign, landing page and referrer of the visits that led to the form fill, call or booking.
- Store: save those values on the lead record in your CRM, so they stay attached as the lead moves through sales.
- Credit: apply a model (first touch, last touch or multi-touch) to decide which source gets the lead, the pipeline and the revenue.
| Term | What it credits | Typical question |
|---|---|---|
| Lead attribution | Lead creation (form, call, booking, trial) | Which channels bring us leads, and which bring good ones? |
| Revenue attribution | Closed revenue | Which channels produce the money? |
| Marketing attribution | Any conversion, across the funnel | How do our channels work together? |
| Conversion tracking | Nothing: it records that the action happened | Did the form submit fire, and how many times? |
Lead attribution sits between conversion tracking and revenue attribution. For a business without a checkout, the lead record is where the marketing source and the sales outcome meet, which is why getting it right matters more than any model debate.
Why Lead Attribution Matters
- Ad platforms grade their own homework. Each platform counts conversions with its own rules and windows, and several can claim the same lead. Add their reported conversions together and the total is often higher than the number of new leads in your CRM. Lead attribution gives each lead one source of record.
- Cheap leads are not the same as good leads. Cost per lead rewards the channel with the loosest form fills. Attribution that follows the lead to qualification and revenue shows cost per qualified lead and revenue per lead, which often rank channels in the opposite order.
- Budget follows evidence. Once every lead carries a source, you can move money from channels that generate volume to channels that generate deals.
- Sales gets context. A rep who sees that a lead read three pricing pages after a comparison search starts a different conversation than one who sees “source: website.”
Lead Attribution Models
A model is the rule that decides how credit is split when a lead had more than one touchpoint. Single-touch models give all of it to one interaction. Multi-touch models share it.
| Model | How credit is split | Best for | Blind spot |
|---|---|---|---|
| First touch | 100% to the first recorded interaction | Finding the channels that introduce new prospects | Ignores what convinced them later |
| Last touch | 100% to the final interaction before the lead converted | Short cycles, judging closing tactics | Ignores what created the demand |
| Linear | Equal share to every touch | Seeing every channel that played a part | Treats a fifth retargeting view like the first discovery |
| Time decay | More credit the closer a touch is to the conversion | Long cycles where recent contact matters | Undervalues discovery |
| U-shaped (position-based) | Typically 40% first, 40% second anchor, 20% shared by the middle | Valuing both discovery and conversion | The second anchor is defined differently by different sources (see below) |
| W-shaped | 30% each to first touch, lead creation and a later milestone (opportunity or close, depending on the source), 10% to the rest | B2B funnels with clear stages | Needs reliable stage dates in the CRM |
| Full path | Adds a fourth anchor at the closed deal, rest shared | Account-based and long B2B cycles | Heavy setup, many touches unseen |
| Data-driven | Algorithm compares converting and non-converting paths | High lead volume with good tracking | Opaque, needs volume, only sees tracked touches |
When people ask for “the four types of attribution,” they usually mean the first four rows: first touch, last touch, linear and time decay. Position-based, W-shaped and data-driven are later additions to that list.
Where the Top Guides Disagree: What “U-Shaped” Means
Read a few lead attribution guides and you will find two definitions of the U-shaped model. Some say it gives 40% to the first touch and 40% to the last touch. Others say 40% goes to the first touch and 40% to the touch where the person became a lead.
Both are in use, and they only differ when you measure past the lead. If the conversion you are attributing is the lead itself, the last touch before conversion and the lead-creation touch are the same interaction, so the two definitions give identical results. If you attribute opportunities or revenue, they split: one rewards the touch that created the lead, the other rewards the touch just before the deal closed. Pick one, write it into your reporting notes, and do not compare numbers from tools that use the other.
First Touch vs Last Touch for Leads
These are the two views every lead record should carry, because together they answer most budget questions. First touch tells you which channel found the person. Last touch tells you which channel brought them back to convert.
| First touch | Last touch | |
|---|---|---|
| Credits | The interaction that introduced the prospect | The interaction right before the form, call or booking |
| Rewards | Organic search, content, social, PR, referrals | Branded search, email, retargeting, direct visits |
| Use it to | Decide where to find new prospects | Tune the pages and campaigns that convert |
| Main drawback | Ignores what persuaded them | Ignores what created the demand in the first place |
The main drawback of last touch is worth repeating for lead generation: a prospect who found you through a blog post, came back a week later by typing your name into Google and then filled the form gets credited to branded search. Cut the content because it “doesn’t generate leads” and the branded searches dry up a few months later. Deeper dives: first touch attribution and last click attribution.
How to Choose a Lead Attribution Model
| Your situation | Start with |
|---|---|
| Low lead volume, short cycle | First touch and last touch side by side. Anything fancier is noise at low volume. |
| Leads convert in days, one or two channels | Last touch, with first touch as a check before cutting any upstream channel |
| Sales cycle of weeks to months, several channels | First and last touch on the lead, plus W-shaped or position-based once stage dates are reliable |
| High volume and clean tracking across channels | Data-driven as a third view, never the only one |
| B2B deals involving several people | Any of the above at contact level, plus account-level rollups (see the B2B section) |
Whatever you choose, the more useful habit is comparing two models on the same leads. When first touch and last touch agree, the decision is easy. When they disagree, you have found a channel that opens journeys and another that closes them, and you should fund both.
How to Capture Lead Attribution on Every Lead: 7 Steps
Most attribution problems are capture problems. The model cannot credit a source that never reached the lead record. This setup works with HubSpot, Salesforce, Pipedrive, a booking tool or a spreadsheet.
Step 1: Standardize your UTM tags
Tag every link you control: ads, emails, social posts, partner listings, QR codes. Google’s documentation is explicit that UTM parameter values are case sensitive, so “Facebook,” “facebook” and “fb” become three sources. Write a short naming sheet (lowercase, underscores, fixed values for medium such as cpc, email, paid_social) and make it the only way links get built. Do not add UTMs to internal links on your own site: they restart the session and overwrite the real source.
Step 2: Define every lead conversion
List every way someone becomes a lead: contact form, demo or quote request, phone call, chat, calendar booking, free trial, gated download. Track all of them. If you only track the main form, every lead who called instead looks like it came from nowhere. For help choosing and counting them, see conversion events.
Step 3: Save the first and latest touch on the visitor
UTM parameters only exist in the URL of the landing page. Two clicks later they are gone, and so is your attribution, unless something saved them. Add a small script that runs on every page and, on each new visit, records the UTM values, the referrer and the landing page into first-party storage. Keep two sets: first touch (written once, never overwritten) and latest touch (updated on every new campaign or referral visit, but not on direct visits).
Step 4: Pass the values through hidden form fields
Add hidden fields to each lead form (for example ft_source, ft_medium, ft_landing_page, lt_source, and so on) and have the script fill them from storage when the form loads. The values then arrive in the CRM with the submission, no matter which page the form sits on. For phone leads, use call tracking numbers per source or ask on the call and log it.
Step 5: Ask how they heard about you
Add an optional question to the form: “How did you hear about us?” Offer a short list (search engine, AI assistant such as ChatGPT, podcast, social, friend or colleague, event, other) plus a free-text box. This catches word of mouth, podcasts, communities and AI answers that software sees only as a direct visit.
Step 6: Store the fields on the CRM record and protect first touch
Create a CRM property for every hidden field and map them. Then set one rule: first-touch fields are written when the contact is created and never overwritten. Latest-touch fields update on each later conversion. Many CRMs have their own “original source” field; keep it, but keep yours too, because a CRM’s automatic source can only see what its own tracking code saw.
Step 7: Report qualified leads, revenue and value per lead by source
Join lead stages (qualified, opportunity, won) and deal values to the source fields. Report by first-touch source and by latest-touch source. The worked example below shows what that report looks like and why it changes decisions.
The Lead Source Fields to Store (Template)
Copy this field list into your CRM. It is enough for every model in this guide.
| Field | Holds | Set when | Overwrite? |
|---|---|---|---|
| ft_source / ft_medium / ft_campaign | UTM values or referrer-derived source of the first visit | First visit, saved by script | Never |
| ft_landing_page | URL path of the first page seen | First visit | Never |
| ft_referrer | Referring domain of the first visit (e.g. google.com, chatgpt.com) | First visit | Never |
| ft_date | Date of the first visit | First visit | Never |
| lt_source / lt_medium / lt_campaign | Source of the latest non-direct visit | Each new campaign or referral visit | Yes |
| lt_landing_page | Landing page of the latest non-direct visit | Each new campaign or referral visit | Yes |
| conversion_page | Page where the form was submitted | Each conversion | Yes (keep history) |
| self_reported_source | Answer to "How did you hear about us?" | First form fill | Never |
| lead_created_date / qualified_date / won_date | Stage dates | By sales process | No, append |
| deal_value | Closed revenue for the lead or account | On close | Yes |
Self-Reported vs Software Attribution: Which to Trust
Once you collect both, they will disagree often. That is the point: they measure different things. Software sees the clicks that led to your site. People remember what made them care. Use this rule to reconcile them:
| Software says | Lead says | What to record and why |
|---|---|---|
| Direct | Podcast, friend, AI assistant, event | Use the self-reported answer for discovery. Direct means the click carried no source, not that nothing caused it. |
| Organic search (branded query) | Podcast or colleague | Self-reported for discovery, organic search for the conversion path. Branded searches are usually the second step. |
| Paid social | Search engine | Keep both. The ad may have started it and the person later searched. Compare across many leads, not one. |
| Organic search | Organic search | Agreement. High confidence. |
| Any | Blank or "other" | Software value only. Do not guess. |
Report self-reported attribution as its own column rather than merging it into the software source. Over a quarter, the gap between the two columns tells you how much of your demand comes from channels no tracking code can see, which is exactly the demand most at risk of being cut.
Worked Example: From Leads to Lead Value by Source
The numbers below are illustrative, for a B2B services firm over one quarter, with leads grouped by first-touch source. They show why cost per lead alone points budget in the wrong direction.
| First-touch source | Leads | Spend | Cost per lead | Qualified | Won | Avg deal | Revenue |
|---|---|---|---|---|---|---|---|
| Organic search | 120 | n/a | n/a | 30 | 6 | $9,000 | $54,000 |
| Paid search | 200 | $16,000 | $80 | 40 | 5 | $8,000 | $40,000 |
| Paid social | 260 | $10,400 | $40 | 26 | 2 | $7,000 | $14,000 |
| Referral | 40 | n/a | n/a | 20 | 5 | $10,000 | $50,000 |
- Organic search: 6 x $9,000 / 120 = $450 per lead.
- Paid search: 5 x $8,000 / 200 = $200 per lead. Cost per qualified lead: $16,000 / 40 = $400. Revenue per ad dollar: $40,000 / $16,000 = 2.5.
- Paid social: 2 x $7,000 / 260 = $53.85 per lead. Cost per qualified lead: $10,400 / 26 = $400. Revenue per ad dollar: $14,000 / $10,400 = 1.35.
- Referral: 5 x $10,000 / 40 = $1,250 per lead.
Judged on cost per lead, paid social wins. Judged on revenue, it is the weakest channel by far, and both paid channels cost the same $400 per qualified lead. Now run the same 18 won deals by last touch instead:
| Source | Won deals, first touch | Won deals, last touch |
|---|---|---|
| Organic search | 6 | 3 |
| Paid search | 5 | 7 |
| Paid social | 2 | 1 |
| Referral | 5 | 4 |
| Email / direct | 0 | 3 |
| Total | 18 | 18 |
Under last touch, organic search loses half its deals to paid search and email, mostly prospects who found the firm through an article and came back through a branded ad or a nurture email. Neither view is “true.” Together they say: organic search finds the buyers, paid search and email bring them back. Put a value on every conversion type with how to calculate conversion value and these numbers become available long before deals close.
How Many Touches Before a Lead or a Sale?
You will see confident numbers quoted for this. Treat them with caution: the count depends on your price point, your sales cycle, how many people are involved, and above all on what your tools can see. Podcasts, conversations, AI answers and ad impressions rarely appear as touches at all. Count your own instead:
- Pull the leads that became won deals in the last two quarters.
- For each, count recorded sessions or campaign touches before lead creation, and again between lead creation and close.
- Report the median and the range, not the average, which a few very long journeys will distort.
- If most leads show one or two touches, your tracking is probably losing the earlier ones, not your buyers deciding instantly.
Lead Attribution for B2B Buying Groups
Contact-level lead attribution has a known limit in B2B: deals are made by groups. Several people from the same company may fill in forms, and others browse without ever converting. Credit the deal to the one contact who booked the demo and you miss the colleague who found you through a search three months earlier.
- Roll leads up to accounts. Associate every contact with its company in the CRM and report the earliest first touch across all contacts on the account, as well as each contact’s own.
- Attribute opportunities, not just leads. Credit pipeline and closed revenue to the sources present on the account before the opportunity was created.
- Accept what stays invisible. Anonymous visitors who never fill a form cannot carry lead attribution. Self-reported answers and sales conversations are the honest supplement.
Lead Attribution for Organic Search and AI Assistants
Organic search is the channel most often undercounted in lead attribution. Search engines do not pass the query, so “google / organic” is as specific as the source gets. The useful unit is the landing page: which articles, service pages and comparison pages produced leads, and what those leads were worth. The SEO conversion tracking guide covers that report in detail.
AI assistants add a second problem. ChatGPT, Perplexity and similar tools send visits with a referrer or a utm_source some of the time, and with nothing at other times, in which case the lead looks like Direct. Record the referrer domain in your first-touch fields so the identifiable AI leads are separated out, and use the self-reported question for the rest. SEOConversion reports conversions and their value by landing page for organic search and AI assistants, and when an assistant sends no referrer, the visit stays Direct rather than being guessed. More on this in AI conversion tracking.
Lead Attribution in GA4
Google Analytics 4 can tell you which channels drive lead events, but it is not a lead attribution system on its own: it does not know which leads qualified or closed. Three facts matter when you use it:
- Only three models remain. GA4 offers data-driven, paid and organic last click, and Google paid channels last click. Google removed first click, linear, time decay and position-based in November 2023, which is one more reason to store first touch on the lead yourself.
- Direct gets no credit unless the whole path was Direct. That is stated on the same attribution page, and it is why GA4 and your CRM’s source fields will not match.
- Lookback windows are limited. The key event lookback window defaults to 90 days (30 or 60 are the alternatives). Touches older than the window do not count, which matters for long B2B cycles.
Mark your lead events as key events, read them in the conversion paths and channel reports, and treat GA4 as the view of the web journey while the CRM remains the record of lead quality and revenue.
When Lead Attribution Is Wrong: Failure Modes and Fixes
| Symptom | Likely cause | Fix |
|---|---|---|
| Most leads show source "direct" or blank | UTMs lost after the landing page; form on a different page | Save UTMs and referrer in first-party storage on landing and fill hidden fields from storage |
| Hidden fields empty on some forms | Form loaded in an iframe or a third-party embed the script cannot reach | Use the form tool's URL parameter prefill, or pass values into the iframe URL |
| First-touch source changes over time | A later form fill or CRM import overwrote it | Make first-touch fields write-once; block imports from updating them |
| Leads credited to your booking or payment tool | Visitor returns from a third-party domain, which becomes a new referral | Add those domains to your referral exclusions and ignore them in the script |
| Same source split into several rows | Case and spelling differences in UTM values | Lowercase everything on capture and enforce the naming sheet |
| Paid platform shows more leads than the CRM | Overlapping claims, view-through credit, duplicate submissions | Use the CRM as the count of record and dedupe by email |
| Phone leads missing from reports | Calls not tracked as conversions | Track tel: clicks and use per-source call numbers or log the source on the call |
| Internal links carry UTMs | Banners or menus tagged with UTMs | Remove UTMs from internal links; they restart attribution mid-visit |
Open your site in a private window with ?utm_source=test&utm_medium=test&utm_campaign=lead_attribution_check, click to a second page, submit a form, and open the new lead in your CRM. If every field you expect is not there, fix capture before you argue about models.
Limits of Lead Attribution
- Correlation is not causation. A source present before a lead did not necessarily cause it. Attribution describes paths; only experiments such as holdouts or geo tests prove lift.
- In-market buyers inflate whoever they touched last. Someone ready to buy will click whatever is in front of them.
- Offline and clickless influence is invisible. Events, word of mouth, TV and ad impressions shape decisions without producing a tracked click.
- Privacy limits shorten journeys. Cookie limits, consent choices and device switching break one person into several visitors, so first touch is often the first touch you saw, not the first that happened.
Lead Attribution Tools
| Option | Good for | Watch out for |
|---|---|---|
| Spreadsheet + self-reported question | Very small lead volumes, service businesses | Manual work, no web journey |
| CRM source fields + hidden form fields | Most lead gen businesses | Needs the capture script and write-once rules |
| GA4 | Channel and page view of lead events | No lead quality or revenue, limited models |
| Dedicated attribution software | Many paid channels, sending qualified-lead events back to ad platforms | Cost, and it still only sees tracked touches |
For a structured way to compare paid options, see marketing attribution tools.
FAQ
What does attribution mean in marketing and sales?
Attribution means assigning credit for an outcome to the things that caused it. In marketing and sales, it is the process of deciding which channels, campaigns and touchpoints get credit for a lead, an opportunity or a closed deal. Lead attribution is the part that answers where each lead came from.
What are the four types of attribution?
In marketing, the four rule-based models most guides list are first touch, last touch, linear and time decay. Position-based (U-shaped), W-shaped and data-driven models are usually counted on top of those. In psychology, "attribution" means something else entirely: how people explain the causes of behavior.
What is the difference between first-touch and last-touch attribution?
First touch gives 100% of the credit for a lead to the interaction that introduced the person to your brand. Last touch gives 100% to the final interaction before the lead converted. First touch shows which channels find new prospects; last touch shows which ones close. Store both on every lead so you can compare them.
What is a drawback to using the last-touch attribution model?
It gives zero credit to everything before the final interaction. Channels that create demand, such as content, organic search, social and podcasts, look like they produce no leads, while branded search, email and retargeting look better than they are. Budgets that follow it drift toward channels that harvest demand instead of creating it.
How many touches before a sale?
There is no reliable universal number. Counts depend on the business, the price, the length of the sales cycle and what your tools can see, and most tools miss offline and clickless touches. Count your own: take recently won deals, list the recorded touches before the lead was created and before the deal closed, and look at the median.
Can I do lead attribution without expensive software?
Yes. Consistent UTM tags, a small script that saves the first and latest source in first-party storage, hidden fields on your forms and matching fields in your CRM cover most needs. Add a "how did you hear about us" question for word of mouth and podcasts. Paid attribution tools mainly save setup time and stitch more touches together.
Know which pages turn visitors into valuable leads.
SEOConversion autocaptures form submits and tel/mailto clicks, lets you put a value on each conversion, and reports conversions and value by landing page for organic search and AI assistants.
Start free