# How to Measure Visitor Identification ROI

A plan for measuring visitor identification ROI that separates incremental pipeline from pipeline the tool only touched, with a holdout test, full running cost and a Salesforce, GA4 and Sheets setup.

Influenced pipeline is the biggest number in a visitor identification renewal case, and the weakest one. This plan measures the pipeline the tool actually caused, prices everything it takes to run, and shows where each number lives in Salesforce, GA4 and Google Sheets.

- Judge visitor identification on incremental pipeline from a holdout test, not on influenced pipeline totals.
- ROI is incremental closed-won revenue minus full running cost, divided by full running cost.
- Full running cost covers licence, usage meter, paid add-ons, admin time and rep follow-up time.
- Track pipeline in Salesforce, check traffic in GA4 and run the maths in Google Sheets.

This works when enough target accounts visit the site each month to split them into a treated group and a holdout group, and when identified visits reach Salesforce as records reps work. On a low-traffic site a holdout is too small to read. Report sourced pipeline instead, add influenced pipeline at a conservative weighting agreed in advance, and label the result as an estimate.

## What Is The Difference Between Sourced, Influenced And Incremental Pipeline?

Sourced pipeline is opportunities where an identified visit was the first recorded touch, influenced pipeline is any opportunity whose account was identified before it closed, and incremental pipeline is only the share that a holdout comparison shows would not exist without visitor identification. Only the third number answers the ROI question.

The three numbers come from the same data. A [website visitor identification](https://growthtribune.com/website-conversion/visitor-identification/) tool names an account, the account reaches the CRM, and an opportunity opens or moves. What changes is how much credit the tool takes for that opportunity.

- **Sourced pipeline** is the strongest claim the CRM can make on its own. The identified visit came first, before any form fill, meeting or outbound touch. It still includes accounts that were about to raise their hand anyway.
- **Influenced pipeline** is the largest number and the weakest. Any opportunity whose account appeared in the tool before close counts, including deals sales had been working for months.
- **Incremental pipeline** is the only number that shows cause. It compares accounts whose visits reached sales with similar accounts whose visits did not, and counts only the difference.

> **Incremental pipeline:** Pipeline that exists only because identified visits reached sales, measured as the gap in pipeline per account between a randomly assigned treated group and a holdout group over the same window.

Influenced pipeline overstates the tool for a simple reason. The accounts that visit pricing and demo pages are the accounts already closest to buying, so a tool that watches those pages tends to sit near deals that were already close to closing. That is a selection problem. Randomised field experiments in advertising show how far observational credit can drift from the true effect, and the holdout section later in this plan sets out that evidence and the method that removes the bias.

Report all three numbers, labelled, and never add them together. Sourced and influenced pipeline describe reach. Incremental pipeline describes return.

## How Do You Calculate Visitor Identification ROI?

Visitor identification ROI is incremental closed-won revenue minus the tool's full running cost, divided by that running cost, measured over one fixed window that the team agrees before the test starts. A result above zero means the tool returned more revenue than it cost to run in that window.

Written out, the formula is:

- **ROI** = (incremental closed-won revenue − full running cost) ÷ full running cost
- **Incremental closed-won revenue** = (closed-won revenue per treated account − closed-won revenue per holdout account) × number of treated accounts
- **Full running cost** = licence + usage meter + paid add-ons + (admin hours + rep hours) × loaded hourly cost

Each input has one source and one owner. Agree both before the window opens, so nobody chooses the inputs after seeing the result.

**Inputs to the visitor identification ROI formula**

| Input | Where it comes from | Owner |
| --- | --- | --- |
| Closed-won revenue per account, by group | Salesforce opportunity report, filtered to the window | RevOps |
| Treated and holdout account lists | Random split of the target-account list | RevOps |
| Licence cost | Signed order form for the window | Finance |
| Usage meter cost | Credit or company-tier overage and top-ups bought | RevOps |
| Paid add-on cost | Order form line items | Finance |
| Admin hours | Time log for setup, field upkeep and fixes | Salesforce admin |
| Rep hours | Time spent working alerts and tasks | Sales leadership |
| Loaded hourly cost | Salary plus overhead per role | Finance |
| Measurement window | Fixed in advance, at least one sales cycle | RevOps and finance |

Two rules keep the number credible.

**Show the arithmetic.** Put every input, its source report and the formula in one sheet, so finance can rerun the result from the same Salesforce data. A single headline percentage with no workings gets discounted in budget review.

**Keep ROI and pipeline multiples apart.** A pipeline multiple divides incremental pipeline by running cost. It is useful early, before deals close, but pipeline is not revenue, and a multiple built on influenced pipeline is not a return at all. Use the pipeline multiple as a leading indicator during the window and ROI on closed-won revenue as the renewal number.

Use the same fixed window for revenue and cost. Salesforce Campaign Influence has its own optional time frame, which is a different setting: it caps the days between a campaign's first association and an opportunity's creation, at up to 9,999 days. Campaigns associated before that time frame do not count, and a blank time frame means no cap. Choose it deliberately before the test starts, and do not change it mid-test.

Do not borrow inputs from a vendor ROI calculator. Its match rates and conversion rates describe the vendor's customer base, not your traffic, and the holdout replaces them with your own numbers.

## What Costs Should A Visitor Identification ROI Model Include?

The full cost of visitor identification is the licence plus its usage meter, paid add-ons, admin time and rep follow-up time, because Warmly prices its plans on a monthly credit allowance and Leadfeeder prices its plans on identified-company tiers and monthly credits. The licence alone understates the cost side of the formula.

Start with the licence each vendor lists.

**Warmly pricing**

| Plan | Price | Billing | Notes |
| --- | --- | --- | --- |
| AI Web-Deanonymization (AI Web-Deanonymization) | $10,000 | per year |  |
| Inbound Chat (Inbound Chat) | $20,000 | per year |  |
| AI Inbound Autopilot (AI Inbound Autopilot) | $30,000 | per year |  |

Warmly lists its plans from a 10,000-credit monthly allowance, so the credit meter is part of the plan you buy. It sells the GTM Signals Package and Warm Experiences as separate annual add-ons, and prices the AI 24/7 Video Chat Agent through sales. A setup that needs signals or personalised site experiences carries those add-on lines on top of the plan.

**Leadfeeder pricing**

| Plan | Price | Billing | Notes |
| --- | --- | --- | --- |
| Lite | €0 | per month |  |
| Discover | €79 | per month, billed annually |  |
| Activate | €369 | per month, billed annually |  |
| Scale | €599 | per month, billed annually |  |
| Enterprise | Custom pricing | per custom |  |

Leadfeeder (formerly Dealfront), whose visitor product is now called Web Visitors, tiers its paid plans by identified companies per month and, on Activate and Scale, by credits per month. Extra credits come as top-ups in blocks of 100 to 5,000 that do not renew, so a month of heavy enrichment is a one-off cost rather than a new baseline. Paid plans include unlimited users, which keeps seat growth out of the model. Scale bills annually upfront or quarterly, which affects when the cost lands in the window.

Then add the internal costs that no pricing page shows:

- **Admin setup and upkeep.** Field mapping, segment rules, fixes when a required field is renamed, and the monthly report build. Log the hours.
- **Rep follow-up time.** Every alert a rep opens, researches and acts on is selling time. Multiply alerts worked by average minutes per alert.
- **Overage and top-ups.** Credits or company tiers bought mid-window belong to the window.
- **Paid add-ons.** Anything on the order form beyond the base plan.

To set the meter against alternatives on the same basis, you can [compare website visitor identification tools](https://growthtribune.com/best-website-visitor-identification-tools/) by what drives each bill. For this model, the rule is narrower. Cost everything the treated group needed, for the whole window, at loaded hourly rates.

## How Do You Track Visitor Identification Pipeline In Salesforce?

Salesforce credits a campaign with influence on an opportunity only when the contact is a Campaign Member and has been added to that opportunity as a Contact Role, so visitor identification pipeline stays missing from Campaign Influence reports until both links exist. Tracking starts with what each tool writes, then adds the campaign layer.

This section assumes the tool is already connected to Salesforce. A guide to [visitor identification tools for Salesforce](https://growthtribune.com/best-visitor-identification-tools-for-salesforce/) compares what each tool writes.

**What to set up:** A dedicated lead or account source value for identified visits, plus one Salesforce campaign that holds identified contacts as Campaign Members. Turn on Customizable Campaign Influence with auto-association, which can link several influential campaigns to one opportunity. Choose the optional influence time frame, which caps the days between a campaign's first association and an opportunity's creation, before the test starts. Leaving it blank means no time frame at all.

**Permission needed:** For Warmly, the connecting user needs the Salesforce permissions listed in the next section, including Read, Create and Edit on Accounts, Contacts, Leads, Opportunities and Campaigns. Adding identified contacts as Campaign Members, and as Contact Roles on opportunities, is your own workflow.

**What it records:**

- **Leadfeeder Web Visitors.** Leadfeeder can sync accounts, contacts, leads, tasks and opportunities. It sends the latest Web Visitors visit, a link to Web Visitors and the organisation name, plus firmographic fields on accounts and leads. It does not overwrite existing Salesforce values except the Web Visitors link and latest visit, and it syncs from Salesforce once a day. Campaign membership is set up in your own workflow.
- **Warmly.** Warmly writes to Lead, Contact and Account objects, using custom fields it creates with a Warmly prefix plus your mapped fields. Its triggers are website visitor identification, the inbound chat agent and outbound agent orchestration. Without segments, every identified visitor is written to the CRM. If any required field is missing or deleted, the whole write for that record fails.
- **Campaign Influence.** The CampaignInfluence object stores each link between a campaign and an opportunity, and it is available in API version 37.0 and later, so a report or SOQL query can pull influence records directly.

**Limitation:** Customizable Campaign Influence picks up a contact only when that contact is a Campaign Member and a Contact Role on the opportunity, and Campaign Member dates set first and last touch. Primary Campaign Source gives 100% credit to one campaign. The Even Distribution, First Touch and Last Touch models are enabled as Additional Campaign Influence Models, which are described as available to customers with both Salesforce and Account Engagement (formerly Pardot) licences.

> **Limitation:** Campaign Influence is only as complete as Contact Role discipline. If reps open opportunities without adding the contacts who visited, those deals vanish from influence reports while still appearing in the tool. Make a Contact Role mandatory at a set opportunity stage, and audit opportunities with no Contact Roles every month.

## What Permissions Does A Visitor Identification Workflow Need?

Warmly's Salesforce connection asks for API Enabled, View All Data and Modify All on Accounts, Contacts, Leads, Opportunities and Campaigns, while GA4 needs the Marketer role to change attribution settings and the Editor role to change data retention. The Google Sheets connector adds its own access request on top.

**Permissions the measurement workflow needs**

| Tool and scope | Permission or role | What it allows |
| --- | --- | --- |
| Warmly in Salesforce, system | API Enabled, APEX Rest Services, Customize Application, Modify Metadata through Metadata API functions | Connects and creates Warmly fields |
| Warmly in Salesforce, data | View All Data, or View All Users when View All Data is off, plus Edit Tasks | Reads records and logs tasks |
| Warmly in Salesforce, objects | Read, Create, Edit, View All, Modify All and View All Fields on Accounts, Contacts, Leads, Opportunities and Campaigns | Writes visits and matches records |
| Warmly with partial access | Continue with limited functionality, then create the Warmly Created, Prospected and Touched fields by hand | Runs without metadata rights |
| GA4 attribution settings | Marketer role or above at property level | Locks the model and lookback window |
| GA4 data retention | Editor role | Sets event retention to 14 months |
| GA4 revenue reporting | No No Revenue Metrics restriction on the analyst | Shows revenue in reports |
| Google Sheets Salesforce connector | View and manage the spreadsheet, connect to an external service, run when you are not present, see your primary email | Imports and refreshes Salesforce reports |

Run the Salesforce connections through a dedicated integration user rather than an admin's own login. Permissions then sit on one auditable profile, the connection survives staff changes, and changes the tool makes are easy to tell apart from human edits.

Watch the GA4 data restrictions. GA4's No Cost Metrics and No Revenue Metrics restrictions hide those metrics from a user, and an analyst with the revenue restriction set will see a test report with the revenue missing. Check each analyst's role before the window opens.

## How Do You Run A Holdout Test For Visitor Identification?

A holdout test assigns target accounts at random to a group whose identified visits reach sales and a group whose alerts are suppressed, because observational comparisons [missed the true lift by a factor of three in half of 15 Facebook field experiments](https://www.kellogg.northwestern.edu/faculty/gordon_b/files/fb_comparison.pdf). The gap between the groups is the incremental pipeline.

**3 x** (2019, 15 Facebook advertising field experiments, 500 million user-experiment observations): In half of the studies, the observational estimate of the percentage increase in purchases was off by a factor of three across all methods compared with the randomised result. Source: [A Comparison of Approaches to Advertising Measurement: Evidence from Big Field Experiments at Facebook](https://www.kellogg.northwestern.edu/faculty/gordon_b/files/fb_comparison.pdf).

Gordon, Zettelmeyer, Bhargava and Chapsky studied advertising, not visitor signals, and their observational methods sometimes underestimated the effect too. The lesson carries over. Comparing accounts that were alerted with accounts that were not, after the fact, mixes the tool's effect with who chose to visit.

**What to set up:** Run the test in this order.

1. Take the ICP target-account list that sales already works.
2. Assign each account at random to a treated group or a holdout group before the window opens. A random number column in a sheet is enough. Do not hand-pick.
3. Route identified visits from treated accounts to reps as normal, using the [actionable visitor signal](https://growthtribune.com/what-revenue-teams-consider-an-actionable-visitor-signal/) rules your team already applies to decide which visits count.
4. Suppress alerts, tasks and CRM writes for holdout accounts. Keep everything else equal, including ads, outbound and inbound routing.
5. Leave the setup alone for a window at least one sales cycle long.
6. Compare opportunity rate and pipeline per account between the groups. The difference, multiplied by the treated accounts, is incremental pipeline.

**Permission needed:** Suppression is a CRM write rule, so it needs whoever controls the tool's sync. In Warmly, segments control which identified visitors are written to the CRM, so a segment that excludes holdout accounts keeps them out. With Leadfeeder Web Visitors, keep holdout accounts out of the alert and task routing your team uses.

**What it records:** Two comparable sets of Salesforce accounts with the same opportunity fields, one exposed to identified-visit follow-up and one not. Identification itself still runs for both, which lets you confirm both groups visited at similar levels.

**Limitation:** The method only removes selection bias when assignment is random and the groups stay untouched. A rep who works a holdout account they spotted elsewhere contaminates the test.

> **Warning:** Small groups give noisy results, and a gap that looks large in the first weeks can disappear by the end. Fix the window and the comparison in advance, do not check the result mid-test and stop early when it looks good, and treat a result from a small target list as directional.

## How Do You Use GA4 To Measure Visitor Identification?

GA4 can show whether key events rose on the pages that trigger visitor alerts, but its default lookback window is 90 days for most key events and 30 days for acquisition key events, and a changed lookback applies only to data collected afterwards. Use GA4 as the traffic check, not the pipeline source.

GA4 does not name visiting companies. Its job in this plan is to confirm that traffic and intent on alert pages held steady across the window, so a change in pipeline is not just a change in visits.

**What to set up:** Track demo requests and pricing-page views as key events on the pages that trigger alerts. Then lock the attribution settings. GA4 offers data-driven and rules-based models, after first click, linear, time decay and position-based were removed in November 2023. The acquisition key event lookback defaults to 30 days, with 7 days optional, and other key events default to 90 days, with 30 or 60 optional.

**Permission needed:** The Marketer role or above at property level changes attribution settings, and the Editor role changes data retention. Marketers can also create and edit key events.

**What it records:** Key events by page and channel over the window, which you compare against the Salesforce pipeline for the same dates.

**Limitation:** A model change rewrites historical and future data, while a lookback change applies only going forward, so changing either mid-test breaks the comparison. Standard properties keep user-level and event data for 2 or 14 months, with key events following the user-level setting. The setting affects explorations and funnel reports only, and at 2 months an exploration loses the start of a long test. Analytics 360 can keep event data for up to 50 months.

> **Tip:** Before the window opens, set event data retention to 14 months, record the attribution model and both lookback windows in the test plan, and remove edit rights from anyone who does not need them until the test ends.

## How Do You Build A Visitor Identification ROI Sheet In Google Sheets?

Google's Salesforce connector for Sheets imports Salesforce reports or SOQL queries into a sheet and refreshes them every 4, 8 or 24 hours, which keeps a visitor identification ROI sheet current without manual exports. The sheet then holds the formula, the inputs and the workings finance asks for.

**What to set up:** Install Google's Data connector for Salesforce in Google Sheets. Import the opportunity report filtered to the target-account list and window, or build a query with the query builder or SOQL using up to five source objects with their fields and filters. Schedule the refresh every 4, 8 or 24 hours, or refresh manually. The connector is English only.

**Permission needed:** The add-on asks to view and manage the spreadsheets where it is installed, connect to an external service, run when you are not present and see your primary Google Account email. It also uses the Salesforce access of the person who connects it.

**What it records:** A live copy of the Salesforce rows the model needs, one row per account, alongside the cost inputs you enter by hand.

**Visitor identification ROI sheet layout**

```
TAB 1 - ACCOUNTS (Salesforce import, read-only)
{{account-id}} | {{group}} | {{first-identified-date}} | {{opportunity-created-date}} | {{pipeline-amount}} | {{closed-won-amount}}

TAB 2 - COSTS (entered by finance and RevOps)
{{licence-cost}} | {{meter-cost}} | {{add-on-cost}} | {{admin-hours}} | {{rep-hours}} | {{hourly-cost}}
Full running cost = {{licence-cost}} + {{meter-cost}} + {{add-on-cost}} + ({{admin-hours}} + {{rep-hours}}) * {{hourly-cost}}

TAB 3 - RESULTS
Opportunity rate per group = accounts with an opportunity in {{group}} / accounts in {{group}}
Pipeline per account per group = sum of {{pipeline-amount}} in {{group}} / accounts in {{group}}
Incremental closed-won = (closed-won per treated account - closed-won per holdout account) * treated accounts
ROI = (incremental closed-won - full running cost) / full running cost
```

**Limitation:** A spreadsheet created in or converted to Sheets holds up to 20 million cells or 100MB, so import account-level summaries rather than every visit row.

> **Warning:** The same connector can push sheet edits back to Salesforce as new or updated records, and can delete highlighted records from Salesforce. Google itself warns to use that with caution. Analysts building the ROI sheet should import only and never use the update or delete actions.

## Which Visitor Identification Metrics Should You Report To Leadership?

Leadership needs incremental pipeline, cost per incremental opportunity and sourced pipeline from visitor identification, because more of the buying journey happens without sales and [67% of 646 B2B buyers surveyed by Gartner prefer a rep-free experience](https://gcomdr.pdo.aws.gartner.com/en/newsroom/press-releases/2026-03-09-gartner-sales-survey-finds-67-percent-of-b2b-buyers-prefer-a-rep-free-experience). Web signals are one of the few views into that journey.

Gartner fielded that survey in 2025, and it also found that 45% of buyers used AI during a recent purchase. When buyers research alone, a visit to pricing or demo pages may be the first sign of intent sales gets. That raises the value of the signal, and it raises the cost of crediting it wrongly.

- [ ] Incremental pipeline and incremental closed-won revenue from the holdout comparison
- [ ] ROI on incremental closed-won revenue, with the workings attached
- [ ] Cost per incremental opportunity, from the full running cost
- [ ] Sourced pipeline, labelled as sourced
- [ ] Influenced pipeline, shown separately and labelled as influenced
- [ ] Speed to first touch on alerts
- [ ] Share of alerts reps worked

Report monthly, in the same pipeline review as every other channel, using the same definitions. A visitor identification number presented in its own meeting, on its own terms, invites the suspicion that it would not survive the shared review.

Add one quality check. A match on the wrong company inflates every number above, so test [match accuracy on known accounts](https://growthtribune.com/original-study-visitor-match-accuracy-on-known-accounts/) before trusting the pipeline the tool reports.

## Common Mistakes When Measuring Visitor Identification ROI

Reporting influenced pipeline as if visitor identification created it credits the tool with deals that would have closed without any alert. Each mistake below has a fix that is cheap to put in place before the window opens.

> **Warning:** Influenced pipeline overstates the tool, and observational comparisons can miss the true effect by a large margin. Every fix below exists to keep the reported number honest.

- **Reporting influenced pipeline as created by the tool.** Show it separately, labelled, and lead with incremental pipeline from the holdout.
- **Using a vendor calculator's match and conversion rates.** Replace them with your own treated and holdout results.
- **Costing only the licence.** Add the usage meter, top-ups, paid add-ons, admin hours and rep hours.
- **Changing settings mid-test.** A new GA4 lookback applies only going forward and a model change rewrites history. Lock both before the window opens, and leave the Salesforce influence time frame unchanged during the test.
- **Missing Contact Roles.** Campaign Influence ignores contacts who are not Contact Roles on the opportunity. Make them mandatory at a set stage.
- **Letting every identified visitor flood the CRM.** Without segments, Warmly writes every identified visitor to the CRM. Set segments so only target accounts reach reps, which also makes the holdout possible.
- **Reading First Touch or Last Touch models you cannot use.** The additional influence models depend on extra licensing, so confirm which models the org has before promising them in a report.

## Frequently Asked Questions

**How long should a visitor identification ROI test run?**

A visitor identification ROI test should run for at least one full sales cycle, so opportunities from identified visits can reach a closed outcome before the treated and holdout groups are compared. Choose the Salesforce Campaign Influence time frame, which caps the days from a campaign's first association to opportunity creation, before the test, and leave it unchanged.

**Is a free visitor identification plan enough to measure ROI?**

Any plan, free or paid, is enough only when it supports what a holdout needs: enough target accounts to split into treated and holdout groups, identified visits that reach Salesforce as records reps work, and a window of at least one full sales cycle. Check each tool's pricing table above against those three needs before the test starts.

**Does Warmly joining HubSpot change how it works with Salesforce?**

Not for now, according to Warmly. Warmly announced on June 30, 2026 that it agreed to be acquired by HubSpot, and it states that existing contracts, pricing, account teams, product experience and integrations remain unchanged, so a Salesforce connection keeps working as before. Plans to connect Warmly across HubSpot's platform are announced, not available.

**Does the Dealfront rebrand change Leadfeeder data or pricing?**

The rebrand leaves Leadfeeder's pricing unchanged. Dealfront is now Leadfeeder, and Leadfeeder says plans, features and pricing remain the same and no functionality is removed. What changes is the naming: the visitor product is now called Web Visitors, Promote becomes Campaigns and the Dealfront API becomes the Leadfeeder API.

## Sources

- [A Comparison of Approaches to Advertising Measurement: Evidence from Big Field Experiments at Facebook](https://www.kellogg.northwestern.edu/faculty/gordon_b/files/fb_comparison.pdf), kellogg.northwestern.edu
- [Gartner Sales Survey Finds 67% of B2B Buyers Prefer a Rep-Free Experience](https://gcomdr.pdo.aws.gartner.com/en/newsroom/press-releases/2026-03-09-gartner-sales-survey-finds-67-percent-of-b2b-buyers-prefer-a-rep-free-experience), gcomdr.pdo.aws.gartner.com

---
Source: https://growthtribune.com/how-to-measure-visitor-identification-roi/
