Attribution
How to read source-to-outcome reporting without collapsing source-specific signals
Attribution in SGEN is the reporting body that connects origin signals — UTM tags, referrers, channel labels, source records — to business outcomes like form submissions, phone taps, and tracked conversions. Where Analytics tells you what visitors did and Ads reporting tells you what your paid spend produced, Attribution answers the question that sits underneath both: where did the result come from?
This page covers what Attribution reads, what it deliberately does not read, and how to interpret the source-to-outcome view without confusing it with the source-specific reporting surfaces it consumes. Attribution is built to sit on top of those surfaces — not to replace them. The reporting law is that source measurement, source visibility, and source-attributed outcomes stay distinct.
What is this for?
Attribution exists for one question: when an outcome happens on your site (a form submit, a phone tap, a tracked conversion), where did the person come from? The answer pulls together UTM tags, referrer data, channel labels, and source records from the visitor's session and ties them to the outcome record.
Read this page when you need to:
- Answer "what source produced this lead?" for a specific form submission or phone tap.
- See which sources are driving outcomes versus driving traffic.
- Compare source quality, not source volume, across organic, paid, and direct.
- Build a stakeholder report that maps revenue or leads to the source that produced them.
- Diagnose why a high-traffic source produces few outcomes — or why a low-traffic source punches above its weight.
Good use cases
Lead-source debrief on a single form submission. A new lead lands. Open the form submission record and read the attribution panel — UTM source, UTM medium, referrer, landing page, and session timeline. You learn whether this lead came from the Friday social post, the Google Ads campaign, or a direct visit. The answer guides which channel gets credit and which spend you double down on.
Channel-quality comparison. Open the attribution summary view for the last 30 days. Each channel row shows visits, outcomes, and outcome rate. Organic search drove 800 visits and 6 form submits — 0.75% conversion. Paid social drove 1,400 visits and 4 form submits — 0.29% conversion. The volume picture says paid social wins; the attribution picture says organic produces better leads per visit. The decision is now informed instead of guessed.
Spike investigation. Form submissions doubled this week. Open Attribution filtered to Form Submit outcomes for the affected period. The source breakdown tells you whether the spike is one channel (a viral post, a paid push, a partner referral) or spread across channels (a brand event, a press mention, a seasonal pattern). The right next action depends on the answer.
Quarterly source review. Pull a 90-day attribution view broken out by channel. The output is a clean source-to-outcome table for stakeholder review — no manual cross-referencing of Analytics with Forms with Phone Taps, because Attribution already did that math.
Tracking-quality audit. Run an attribution view filtered to a known campaign window. If the campaign UTMs do not appear as a source, the tagging is broken on the campaign side — links going out without UTM parameters, or with parameters that get stripped at click time. Attribution surfaces the gap before the campaign post-mortem turns into a guessing match.
What NOT to use this for
As a single source of truth for ad-platform reporting. When the ad platform's own dashboard and SGEN's attribution view disagree, that is not a bug — it is the predictable result of two systems counting on different rules. Use Attribution for site-side source-to-outcome reading. Use the ad platform's dashboard for ad-platform-side cost-and-click reading. Reconcile carefully, do not assume the two are identical.
For cross-device or cross-session identity stitching. Attribution reads what was knowable at outcome time from the visitor's current session. If someone clicks your Google Ads link on mobile, leaves, and submits a form from desktop a day later, the desktop session does not know about the mobile click — the form gets attributed to whatever the desktop session shows. Cross-device stitching is the job of a marketing-attribution tool that lives outside SGEN.
As a substitute for source-specific reporting. Attribution consumes signals from Analytics, Forms, Phone Taps, and Ads reporting. It does not replace any of them. When you want the raw event-by-event audit of what hit your server, go to Analytics. When you want the form submission record itself, go to Forms. Attribution is the layer that connects them — not the layer that owns them.
For marketing-campaign field grouping. Attribution is reporting, not a campaign-field schema for marketing operations. UTM tags are how the data arrives, but Attribution does not provide a CRM-style campaign object you can edit, group, or assign owners to. That type of campaign management lives in your CRM or marketing platform.
As a privacy bypass. Attribution reads the source data the visitor's browser sends and the tags your team appends to outbound links. It honours Tracking Consent for any signal that requires consent. When a visitor declines tracking, the attribution view shows what was available without the consent-gated signals — not the full picture.
How this connects to other features
- Analytics — the source data Attribution reads. Page views, referrers, and form-submit events flow through Analytics first; Attribution reads them in aggregate to produce the source-to-outcome view.
- Ads reporting — the paid-cost-and-result reading. Attribution tells you what the paid source produced in outcomes; Ads reporting tells you what it cost. Read together for the full paid picture.
- Forms — the outcome side of attribution for form submissions. Each submission record carries the attribution context from the visitor's session.
- Phone Taps — the outcome side of attribution for phone-tap events. Phone-tap records carry the same attribution context.
- Tracking Consent — governs which signals Attribution is allowed to read. Consent declines reduce what Attribution can show, by design.
- Google Analytics 4 integration — the parallel marketing-attribution surface. SGEN's Attribution covers site-native outcomes; GA4 covers the marketing layer above.
Before you start
- The site is collecting source signals — at minimum UTM parameters on inbound links and referrers from the visitor's browser. Without source signals, Attribution falls back to a (direct) bucket for everything.
- Outcome-side modules are wired. Forms is the primary one for most sites; Phone Taps where mobile-tap conversions matter; Ads reporting where paid is part of the mix.
- You are signed in to SGEN with admin access.
- Tracking Consent posture is decided. Sites that operate under strict consent rules will see a larger (consent-not-granted) bucket in Attribution. That is expected, not broken.
- Your team is tagging outbound links with consistent UTM parameters. Inconsistent or missing tagging is the single biggest cause of "attribution looks weird" complaints — and the only fix is upstream link discipline.
Where to go
- Open the left navigation in the SGEN admin.
- Open the Attributions area within SG-Modules. The default view opens to source-to-outcome summary for the default period.
- Each individual outcome record (a form submission, a phone tap) also carries an inline attribution panel — open the record to read the per-outcome source detail.
Steps
1. Pick the outcome type
Attribution can read different outcome types — form submissions, phone taps, tracked conversions, and (where wired) ecommerce events. The outcome selector at the top of the view scopes the source breakdown to one outcome type at a time.
Reading all outcome types stacked together produces a blurred view. Pick one outcome type for the question you are answering — form submits for lead-quality questions, phone taps for mobile-engagement questions, ecommerce events for revenue-attribution questions.
2. Pick the period
Open the date-range picker and choose the window. The view defaults to last 30 days, which is a reasonable starting point. For campaign-specific debriefs, pick the campaign window plus a few days on either side to catch delayed conversions.
The URL captures the selected period so the view is bookmarkable and shareable.
3. Read the source breakdown
The source table shows one row per attributed source — organic / google, cpc / google-ads, social / instagram, email / newsletter, (direct), (consent-not-granted), and so on. Columns show visits, outcomes for the selected outcome type, and the outcome rate (outcomes divided by visits).
The outcome rate is the column to read first. A source with high visits and low rate is a traffic source; a source with low visits and high rate is a conversion source. Treat them differently in your marketing decisions.
4. Drill into a single source
Click a source row to see the underlying outcome records for that source over the period. The list shows each form submission or phone tap with timestamp, page, and (where available) the visitor's session identifier. Use this view for spot-checks — when a source row looks wrong, the underlying records tell you whether the issue is volume, quality, or attribution mis-tagging.
5. Read the per-outcome attribution panel
For any single form submission or phone tap, the record carries an attribution panel that names the source, medium, campaign, landing page, and referrer at the moment of the outcome. Open this panel when you need the full story behind one specific lead.
What success looks like
- The attribution summary view renders within a couple of seconds once the outcome type and period are picked.
- Source rows account for the majority of outcomes — a healthy attribution setup has under 30% of outcomes falling into the
(direct)bucket. - Campaign-tagged inbound links surface as their own rows (
cpc / google-ads,email / newsletter-may, etc.) rather than disappearing into(direct)or(unknown). - The per-outcome attribution panel on a form submission record shows UTM source, UTM medium, landing page, and referrer fields populated for the majority of records.
- Tracking Consent declines show up as a labelled
(consent-not-granted)bucket, not as silent zeros. - Bookmarked attribution views return the same outcome type, period, and source breakout on revisit.
What to do if it does not work
The (direct) bucket is huge. That usually means inbound links are missing UTM parameters. Audit the campaign side — newsletters, social posts, ad creatives — and add consistent UTM tagging. Direct traffic in the 10-25% range is normal. Direct over 50% is a tagging problem upstream, not an Attribution problem in SGEN.
A specific UTM tag does not appear as a source. The most common cause is a URL-shortener or redirect that strips the query string. Test the actual outbound link in incognito and confirm the UTM parameters survive the click. If they get stripped, the source side is the fix — use a shortener that preserves parameters, or stop using one.
Form submissions show no attribution data. Check that the form is on a page that participates in the analytics tracking, and that the visitor session captured the source signal at landing time. A form on a page that bypasses analytics will produce submissions with empty attribution.
Attribution disagrees with the ad platform's reported conversions. This is normal and expected. The two systems count under different rules — attribution windows, last-click vs. first-click, view-through vs. click-through. For paid spend decisions, treat the ad platform's reported numbers as the platform-side source of truth and treat Attribution as the site-side source of truth.
The view is empty for a known active period. Confirm the outcome type is set to one that has events in the period (Form Submit, Phone Tap, etc.). All Outcomes can mislead when one type dominates. Also confirm Tracking Consent is not silently blocking the signals — if a site recently turned on strict consent rules, the immediate effect is fewer attribution-eligible signals.
Numbers shift between visits to the same view. Some shift is normal as late-arriving session data and attribution windows close out. Significant shifts after a few hours indicate either an integration problem or a date-range that includes the current incomplete day. For stable reporting, end the date range on the prior day.
Examples
Example A — Lead-source debrief on one submission. A new form submission lands at 14:22 on Friday. Open the record. The attribution panel readsutm_source: google, utm_medium: cpc, utm_campaign: spring-launch, landing page /spring-launch, referrer google.com. This lead came from your Google Ads spring-launch campaign. Credit the channel and the campaign accordingly.Example B — Channel-quality comparison for the marketing review. Pull last 30 days, outcome type Form Submit. Organic search drove 812 visits and 6 submits (0.74% rate). Google Ads drove 186 visits and 11 submits (5.91% rate). Instagram drove 1,402 visits and 4 submits (0.29% rate). The marketing read is: Google Ads is the most efficient channel per visit; Instagram drives traffic but few leads; organic is the steady mid-tier. Budget conversation now has data.
Example C — Campaign window post-mortem. A two-week Instagram push ended yesterday. Pull Attribution for the campaign window plus three days, outcome type Form Submit, filtered to social / instagram. The output is the exact number of form submissions Instagram drove during the push. Compare against the equivalent two-week window before the push for the lift number. Stakeholder report writes itself.
Reading the source taxonomy
Attribution organises sources by a two-part taxonomy that follows the UTM convention: source / medium. Common combinations and what they mean:
| Source / medium | What it usually means |
|---|---|
organic / google | Visitor arrived from a Google search result (no UTM, identified by referrer) |
organic / bing | Same as above for Bing |
cpc / google-ads | Paid click from Google Ads (UTM-tagged) |
cpc / meta-ads | Paid click from a Meta ads campaign |
social / instagram | Click from an Instagram link in bio, story, or post |
social / linkedin | Click from a LinkedIn post or profile |
email / | Click from a tagged email campaign |
referral / | Click from a link on another site (identified by referrer) |
(direct) | Visitor with no source signal — typed URL, bookmark, app, or stripped referrer |
(consent-not-granted) | Visitor declined tracking; source signals withheld |
(unknown) | Source signal arrived in a shape the parser could not classify |
utm_source / utm_medium combination your team uses on outbound links will surface as its own row in Attribution. Discipline pays off: every new combination is a new row to read, so consistent tagging across campaigns keeps the view legible.Consent and attribution
Tracking Consent is the upstream gate for any signal that requires consent. When a visitor declines, SGEN withholds the consent-gated signals and Attribution surfaces those visits in a labelled (consent-not-granted) bucket.
That bucket should be visible in the source breakdown, not hidden. The point is honest reporting — your reporting should reflect that some outcomes happened without a fully-attributed source, rather than silently mis-attributing them to (direct).
For sites in jurisdictions with strict consent rules (GDPR, ePrivacy variants, and similar), the (consent-not-granted) bucket will be larger. That is the expected operating cost of compliance, not a reporting bug.
Attribution and outcome-record retention
The attribution data lives on the outcome record itself — a form submission's attribution panel is part of that submission's record and follows the same retention policy. When form submissions are deleted, their attribution context goes with them. When Attribution surfaces a historic summary, it pulls from the records that still exist.
This is worth knowing for two reasons. First, aggressive retention policies on form submissions will eventually reduce the historical depth of Attribution reporting. Second, the attribution context cannot be recovered after the underlying outcome record is deleted — there is no separate attribution archive.
A practical attribution rhythm
Attribution is most useful when read at three cadences:
- Per-lead, on landing. When a notable lead comes in, open the record and read the attribution panel. This is debrief work — five seconds per important lead, and it builds your team's mental model of which sources produce which lead shapes.
- Weekly source-quality scan. Open the source breakdown for the prior 7 days, outcome type Form Submit. Note which sources punched above their weight (high rate, low volume) and which were drag (low rate, high volume).
- Quarterly stakeholder report. Pull a 90-day window. Export or screenshot the source breakdown table. Pair it with the corresponding Ads reporting summary for the cost-and-result picture alongside the source-and-result picture.
Where Attribution ends, deeper analytics start
Attribution covers site-native source-to-outcome reading. It does not cover:
- Lifetime value by source. Pair Attribution with the order or customer record to compute LTV by source. SGEN does not draw this view natively.
- Multi-touch attribution. Attribution defaults to the last-known source on the outcome record's session. Multi-touch models (first-touch, linear, time-decay, position-based) live in marketing-attribution tools outside SGEN.
- Cross-device journey reconstruction. When a visitor switches device between source and outcome, the attribution view captures only the outcome-side session. Cross-device identity stitching is outside scope.
- Predictive source modelling. Attribution is a descriptive read of what already happened. Predictive source-to-outcome models live in dedicated marketing-analytics products.
Related reading
- Analytics — the operations-level analytics surface that Attribution reads from.
- Ads reporting — paid-cost-and-result reporting that pairs with attribution for the full paid picture.
- Forms — the form-submission module that carries per-record attribution data.
- Phone Taps — the phone-tap module that carries per-record attribution data.
- Tracking Consent — the consent surface that gates attribution-eligible signals.
- Google Analytics 4 integration — the parallel marketing-attribution path through GA4.