The complete guide to B2B marketing attribution

This guide is for B2B marketing leaders and RevOps people who are asked, every quarter, which channels produced pipeline and revenue, and who suspect the dashboard answer is shakier than it looks. It assumes your deals take months and involve several people.

After reading it you should be able to choose an attribution model and know what it hides, decide between what your CRM already does and a dedicated tool, add self-reported attribution without fooling yourself, and know when a holdout test or a marketing mix model is the better question to ask. You should also know what each option costs where prices are published.

It is not for self-serve products with short cycles and one buyer per purchase, where ad platform reporting and product analytics usually do the job. It is also not a substitute for a measurement plan agreed with finance. Attribution can inform a budget decision; it cannot settle an argument nobody has framed.

Summary

• B2B purchases involve large buying groups and long cycles, and buyers do most of their research before they talk to you. Any tracking system sees only part of that.
• Every rules-based model (first touch, last touch, linear, U-shaped, W-shaped, time decay) is a policy for splitting credit. None of them measures cause.
• GA4 now offers only data-driven and two last-click models. Its rules-based models were removed in 2023, so multi-touch B2B reporting usually lives in the CRM or a dedicated tool.
• Attribute to accounts and opportunities, not just leads. Salesforce and HubSpot both do this out of the box if contact roles and deal associations are kept clean.
• Ask “how did you hear about us” on high-intent forms. It catches podcasts, communities, peers and AI assistants that software misses.
• Cookie consent and Safari’s tracking limits shrink what any tool can see. Plan for gaps rather than pretending they are not there.
• Use lift tests and, at larger budgets, marketing mix modeling to check what the attribution report claims.

1. Why attribution is hard in B2B

Marketing attribution is the practice of assigning credit for a business outcome, such as a lead, an opportunity or closed revenue, to the marketing and sales touches that came before it. A touch, or touchpoint, is any recorded interaction: an ad click, a webinar registration, an email click, a page view, a sales meeting.

B2B paths differ from consumer ones on three counts.

The buyer is a group. Forrester’s State of Business Buying 2026 says the typical purchase decision involves 13 internal stakeholders and nine external influencers, and that the number rises for complex or strategic purchases.¹1 Source: Forrester, “Forrester’s 2026 The State Of Business Buying”, accessed Nov 2026. Your tracking may know two of those people.

The cycle is long. 6sense’s 2025 Buyer Experience Report, based on just under 4,000 respondents, put the average buying cycle at 10.1 months globally and 11.1 months in North America.²2 Source: CustomerThink, “Research Round-Up: 6sense Study Provides Critical Insights on B2B Buyer Behavior”, accessed Nov 2026, summarizing 6sense’s 2025 study. Sample size from 6sense, “2025 B2B Buyer Experience Report”, accessed Nov 2026. 6sense is a vendor; these are its own survey figures. Cookies expire and early touches drop out of view before the deal closes.

Most of the work happens before first contact. The same 6sense study found that buyers first contacted sellers when they were 61% through their buying process, that 79% of buyers initiated that contact themselves, and that 94% of buying groups had ranked their shortlist before talking to any vendor.²

Tracking sees the late, recorded part of the journey and has to guess about the rest. The models in the next chapter are different ways of guessing.

2. Attribution models and what each one assumes

A model is a rule, or an algorithm, that splits credit across the touches on a path. The choice changes which channels look good.

Single-touch models

First touch gives all credit to the first recorded interaction. It favors awareness channels such as paid social, content and events, and ignores what moved the deal forward.

Last touch gives all credit to the final interaction. It favors branded search, direct traffic and retargeting, which often capture demand created elsewhere. “Last non-direct click” skips direct visits.

Both are easy to audit, and wrong in a predictable direction.

Multi-touch models

Multi-touch attribution (MTA) spreads credit across several touches. The common rule-based versions, with the splits HubSpot documents for its reports, are:³3 Source: HubSpot Knowledge Base, “Understand attribution reporting”, accessed Nov 2026. HubSpot also offers J-shaped (20% first, 60% conversion) and inverse J-shaped (60% first, 20% conversion) models.

Model	How credit is split	Leans toward	Main blind spot
Linear	Equal across all touches	Busy, repetitive channels	Treats a page view like a demo
U-shaped	40% first, 40% lead creation, 20% rest	Discovery and form fills	Ignores the opportunity stage
W-shaped	30% first, 30% lead, 30% deal, 10% rest	Funnel milestones	Needs clean deal data
Full path	22.5% each to four milestones, 10% rest	Whole cycle incl. close	Overstates late sales touches
Time decay	More credit to recent touches	Late, bottom-funnel work	Undervalues early demand creation

Vendors do not all use the same splits. Adobe’s Marketo Measure, for example, defines its U-shaped model as 50% to first touch and 50% to lead creation, with nothing in between, while its W-shaped and full path models match the figures above.⁴4 Source: Adobe Experience League, “Marketo Measure Attribution Models”, accessed Nov 2026. HubSpot also offers J-shaped and inverse J-shaped models, which tilt credit toward the conversion or toward the first touch respectively, and runs its time decay model on a seven-day half-life.³ Check the definitions inside your own tool before comparing reports across systems.

Data-driven models

Data-driven attribution uses an algorithm to estimate each touch’s contribution from your own converting and non-converting paths, instead of fixed percentages. Google describes its version as using machine learning across both kinds of path.⁵5 Source: Google Analytics Help, “Get started with attribution”, accessed Nov 2026. Dreamdata says its data-driven model uses Markov chains, which score a channel by how much conversions fall when that channel is removed from the typical journey map. It also says that with little data the model ends up looking similar to a linear model.⁶6 Source: Dreamdata, “Dreamdata launches data-driven attribution for B2B”, accessed Nov 2026. Method description is the company’s own.

Data-driven models only reshuffle credit among tracked touches. They will not discover the podcast your buyer heard in the car.

What GA4 offers today

If your team still expects first-touch or linear reports in Google Analytics 4 (GA4), they are gone. Google’s documentation says the first click, linear, time decay and position-based models have not been available since November 2023.⁵ The remaining options are data-driven, paid and organic last click, and Google paid channels last click.⁵

GA4 also limits how far back it looks. The default key event lookback window is 90 days, with 30 and 60 days as alternatives; acquisition events such as first visit default to 30 days.⁷7 Source: Google Analytics Help, “Select attribution settings”, accessed Nov 2026. A ten-month B2B cycle does not fit in that window.

3. Lead-level versus account-level attribution

Most attribution tools started by tracking individuals: a cookie becomes a known contact at the first form fill, and that contact’s history becomes the path. This breaks down when revenue lands on an opportunity linked to five people, three of whom never filled in a form.

Account-level attribution groups every touch from every known person at a company onto the account, then credits touches against the opportunity. That matches how B2B buying works, how sales teams report, and how account-based marketing (ABM) plans.

The catch is that it depends on matching people to the right account and linking the right people to each opportunity. Both are data hygiene jobs that usually fall to RevOps.

Salesforce Customizable Campaign Influence

Salesforce’s native answer is Customizable Campaign Influence, available in Professional, Enterprise, Performance, Unlimited and Developer editions.⁸8 Source: Salesforce Help, “Customizable Campaign Influence”, accessed Nov 2026. It works by scanning active campaigns for members who also hold a contact role on an open opportunity, then creating influence records. Once an opportunity closes, no new influence records are created.⁹9 Source: Salesforce Help, “How Customizable Campaign Influence Works”, accessed Nov 2026. Check your org’s edition and licenses before relying on the additional models.

The default model, Primary Campaign Source, gives 100% of influence to the single campaign in that field on the opportunity. Additional first touch, last touch and even distribution models can be enabled, but Salesforce’s help pages limit some of them to customers who also use Account Engagement, its marketing automation product formerly called Pardot. Custom models with your own percentages are possible, but they do not auto-associate campaigns, so records must be added manually or through Apex code or the API.⁹

The practical weakness is contact roles. If sales reps do not add buyers to the opportunity as contact roles, campaign influence has nothing to match and marketing looks as though it did nothing.

HubSpot attribution reports

HubSpot’s attribution reports cover contact create, deal create and revenue attribution. Contact create reports are included in Marketing Hub or Content Hub Professional and Enterprise; deal create and revenue attribution reports require Marketing Hub Enterprise.¹⁰10 Source: HubSpot Knowledge Base, “Create attribution reports”, accessed Nov 2026. The tracked interaction types include ad clicks, event registrations and attendance, connected calls, CTA clicks, form submissions, email clicks, page views and social post clicks.³

Two limits matter for B2B. W-shaped and full path models return no value when a deal was created before its associated contacts, which happens when sales creates deals first and adds people later.³ And reports are capped at 20 million interactions after sampling, with page views and email interactions sampled first when volume is high.¹⁰

4. What software cannot see: dark social, AI search and self-reported attribution

“Dark social” is the industry’s loose name for sharing and discovery that leaves no referrer: links pasted into Slack or WhatsApp, a recommendation in a private community, a podcast mention, a LinkedIn post read without clicking. When those buyers finally visit, they type your URL or search your brand name, and the visit is filed as direct or branded search. Last-touch models then credit the brand search for demand it did not create.

AI search referrals

AI assistants add a newer version of the same problem. Since May 2026 GA4 has an AI Assistant channel in its default channel group. Google defines it as traffic from sources such as ChatGPT, Gemini, DeepSeek, Copilot or Grok, assigned when the referrer matches Google’s list of AI assistants.¹¹11 Source: Google Analytics Help, “Default channel group”, accessed Nov 2026. Google’s “What’s new” page dates the AI Assistant channel to May 13, 2026. Google’s own AI Overviews and AI Mode are excluded from that channel and counted as Organic Search.¹¹

Visits that pass no referrer still land in direct, and a buyer who reads an AI answer naming you and later searches your brand leaves no referral at all. Treat the AI Assistant channel as a floor.

Self-reported attribution

Self-reported attribution means asking the buyer, usually with a “How did you hear about us?” field on demo and contact forms. It is the cheapest way to see podcasts, peer referrals, communities and AI assistants. Some rules:

• Ask on high-intent forms only (demo, pricing, contact sales).
• Use free text, or a short list with an “other” option. Pick lists anchor people toward the first items.
• Ask one question about discovery. “What made you reach out now?” is a different question for the sales call.
• Code answers into fixed categories and keep the raw text.
• Report it next to software attribution, not merged into it. Software records the last clicks; buyers remember the first impression.

Self-reported data has its own biases. People forget, compress months into one memory, or leave the field blank. Use it to spot channels your tools miss, then test them.

For how buyers discover vendors in AI answers and how to track it, see the complete guide to AI search visibility.

5. Privacy and consent limits

Two mechanisms shrink what any tool can see.

Consent banners. In regions with consent requirements, visitors who decline analytics cookies cannot be tracked with those cookies. Google’s consent mode has two set-ups. In basic mode, Google tags are blocked until consent is given, and the property gets no modeled data. In advanced mode, tags load before the banner and send cookieless pings when consent is declined; no analytics cookies are set or read, and GA4 uses the pings to model the missing behavior.¹²12 Source: Google Analytics Help, “About consent mode”, accessed Nov 2026. Modeling has thresholds: a property needs at least 1,000 events a day with analytics storage denied for at least seven days, and at least 1,000 daily users with analytics storage granted on at least seven of the previous 28 days, and even then Google does not guarantee eligibility.¹³13 Source: Google Analytics Help, “Behavioral modeling for consent mode”, accessed Nov 2026. Many B2B sites, with modest traffic, never qualify.

Safari’s Intelligent Tracking Prevention (ITP). Since February 2019, WebKit has capped persistent cookies set through JavaScript at a seven-day expiry. In March 2020 it extended the rule to delete all script-writable storage after seven days of Safari use without user interaction on the site.¹⁴14 Source: WebKit, “Intelligent Tracking Prevention 2.1” (February 21, 2019) and WebKit, “Full Third-Party Cookie Blocking and More” (March 24, 2020), accessed Nov 2026. A buyer who reads your content on Safari, leaves for two weeks and then requests a demo can look like a brand-new visitor.

The practical consequences for B2B attribution:

• Early touches are the most likely to be lost, which tilts every model toward late channels.
• Server-side tracking helps but must still respect the consent a visitor gave. Be wary of tools that promise to “recover” declined users.
• CRM-recorded touches (meetings, event attendance, known-contact form fills) survive cookie loss, which is why they are the backbone of B2B attribution.

This chapter describes how the mechanisms work. It is not legal advice on consent requirements in your markets.

6. Tools by category, and what they cost

There are three tool categories: CRM-native attribution (already paid for, close to opportunity data, weak on web tracking), web analytics such as GA4 (good for channels and pages, short lookback, no opportunity data), and dedicated B2B platforms such as Adobe Marketo Measure (formerly Bizible), Dreamdata and HockeyStack, which combine web, ad and CRM data into account-level journeys. Most companies should exhaust the first two before buying the third.

The tools compared

Tool	Category	Models	List price, Nov 2026
HubSpot Marketing Hub	CRM-native	Nine rule-based	Revenue reports need Enterprise, from $3,600/mo
Salesforce Campaign Influence	CRM-native	Primary source, first, last, even, custom	Included in listed editions
Marketo Measure	Dedicated	First, lead creation, U, W, full path, custom	Not published
Dreamdata	Dedicated	Rule-based, data-driven, custom	Free tier; Advanced on quote
HockeyStack	Dedicated	Five rule-based, custom weights	Not published

Sources for each row are in the notes below.¹⁵15 Source: HubSpot, “Marketing Hub pricing”, accessed Nov 2026. List prices as of November 2026. The page shows both $800 and $890 a month for Professional without making clear which billing option each applies to; confirm with HubSpot. ⁸⁴¹⁶16 Source: Dreamdata, “Pricing”, accessed Nov 2026. List price as of November 2026; the page does not state the free plan’s monthly tracked user cap. ¹⁷17 Source: HockeyStack, “Pricing”, accessed Nov 2026. No prices are published.

Notes on each

HubSpot. Marketing Hub Professional starts at $800 a month (the page also shows $890) plus a required $3,000 onboarding fee; Enterprise starts at $3,600 a month plus a required $7,000 onboarding fee, list prices as of November 2026.¹⁵ Deal and revenue attribution are Enterprise-only.¹⁰ Limits: fixed models, and results depend on deals having contacts associated at the right time.

Salesforce. No extra license for the basic models, but the useful ones beyond Primary Campaign Source may require Account Engagement, and custom models need manual or coded association.⁹ Limits: no web tracking of its own; it only knows what is recorded as campaign membership.

Marketo Measure. Adobe publishes no price. Standard Marketo Measure supports one CRM, which must be Salesforce or Microsoft Dynamics; companies on other CRMs, or with several CRMs, need Marketo Measure Ultimate, according to Adobe’s community guidance.¹⁸18 Source: Adobe Experience League Communities, “Marketo Measure vs. Marketo Measure Ultimate: Which Is Right for My Organization?”, accessed Nov 2026. Adobe’s pricing page did not load when checked; we found no published Marketo Measure price. Its custom model lets you add your own stages and weights.⁴ Limits: built around Salesforce or Dynamics, and pricing is quote-only.

Dreamdata. Its pricing page lists a free plan at $0 a month, with two months of user history, five seats and a cap on monthly tracked users, and an Advanced plan at custom pricing that includes AI and custom attribution models, as of November 2026.¹⁶ Limits: the free plan’s two-month history is shorter than most B2B cycles, so it is a trial of the data model more than a working attribution setup, and the data-driven model needs volume to differ from linear.⁶

HockeyStack. Its pricing page lists no figures; plans are custom, with startup pricing mentioned.¹⁷ Its documentation lists first touch, last touch, linear, position-based and time decay models, with custom 0 to 1 weights per touchpoint type.¹⁹19 Source: HockeyStack Docs, “Advanced attribution models”, accessed Nov 2026. It also offers lift reports, which compare similar groups that did and did not see a touch; HockeyStack’s own docs say its control groups are “never perfect” and that lift does not work well where data is sparse.²⁰20 Source: HockeyStack Docs, “Lift analysis vs. multi-touch attribution”, accessed Nov 2026. Limits: no public price, and lift reports are an observational estimate, not a randomized test.

None of this tells you which tool is accurate. Ask each vendor to run on your historic data and show how many closed-won opportunities it connects to at least one marketing touch.

For where attribution sits next to your CRM, marketing automation and data warehouse, see the complete guide to building a B2B revenue stack.

7. Incrementality, holdouts and marketing mix modeling

Attribution distributes credit among things that happened. It cannot tell you what would have happened without a channel. That is the incrementality question, and it needs a different method.

Holdout and lift tests

A holdout test withholds a channel or campaign from a randomly chosen group and compares outcomes with the group that received it. The difference is the incremental lift. Ad platforms run versions of this for you.

Google Ads Conversion Lift based on users requires at least 1,000 observed conversions and a minimum campaign budget of $5,000, and it supports Search, Display, Video, Demand Gen, App and Performance Max campaigns. Google also says it is not available to every account.²¹21 Source: Google Ads Help, “Set up Conversion Lift based on users”, accessed Nov 2026. LinkedIn’s Conversion Lift Testing splits the target audience into test and control groups, shows ads only to the test group, and compares conversions tracked by the Insight Tag or Conversions API. LinkedIn notes that unconsented traffic is not measured and that conclusive results are not guaranteed.²²22 Source: LinkedIn Help, “Conversion Lift Testing”, accessed Nov 2026.

The B2B problem is volume. Many B2B advertisers lack 1,000 conversions of anything that matters. Two workarounds:

• Test on a higher-volume proxy, such as engaged visits from target accounts, then check whether the proxy moves pipeline.
• Run account-level holdouts yourself. Split a target account list at random, suppress a channel for half for a quarter, and compare pipeline created per account in each half. This needs a large list to say anything, and sales must not compensate by working the held-out half harder.

Marketing mix modeling

Marketing mix modeling (MMM) uses regression on aggregate data, such as spend and outcomes by channel over time, to estimate each channel’s contribution with no user-level tracking. That makes it immune to cookie loss. It also makes it slow, and it needs long histories with enough spend variation to learn from.

Two open-source options are free to use. Google’s Meridian is described in its repository as a framework based on Bayesian causal inference that handles geo-level data and supports calibrating the model with experiment results; it is Apache-2.0 licensed.²³23 Source: Google, “Meridian” on GitHub, accessed Nov 2026. Meta’s Robyn uses ridge regression with automated hyperparameter search and is MIT licensed.²⁴24 Source: Meta Marketing Science, “Robyn” on GitHub, accessed Nov 2026. Both need a data scientist, and both were built mainly for advertisers with large paid media budgets.

For LinkedIn-specific tracking, conversion set-up and testing, see the complete guide to LinkedIn Ads for B2B.

8. How to set it up, and the mistakes to avoid

A setup sequence that works

1. Agree the question with finance and sales. Usually: which programs create qualified pipeline and revenue, at what cost. Write down the outcome you will measure.
2. Fix the CRM first. Standard lifecycle stages, required contact roles or deal-contact associations, and a rule for creating opportunities. Without this, every tool produces confident nonsense.
3. Standardize campaign tracking. A short, enforced UTM naming convention (source, medium, campaign) and a CRM campaign record for every program, including events.
4. Pick two views. A first-touch or self-reported view for demand creation and a W-shaped view for pipeline progression, side by side.
5. Add the “how did you hear about us” field and code the answers monthly.
6. Roll up to accounts. Report pipeline influenced per account, and compare target accounts with the rest.
7. Schedule a test. Run one holdout per half-year on the largest budget line you are least sure about.
8. Review quarterly. Disagreements between views are where the useful questions are.

Buy a dedicated platform only once steps 2 and 3 hold and your current tools cannot answer the question.

Mistakes that sink attribution programs

• Treating the model as the truth. Every model is a policy for splitting credit. Switching from last touch to linear moves budget arguments, not reality.
• Crediting branded search and direct traffic. They are where demand shows up, not where it is created.
• Measuring leads instead of opportunities. A channel that produces cheap leads that never convert will win on lead-level attribution.
• Short lookbacks. A 30- or 90-day window on a ten-month cycle drops the touches that started the deal.
• Buying the tool before fixing contact roles. A dedicated platform cannot attribute to people who are never linked to the deal.

For the team and process that own this data, see the complete guide to revenue operations (RevOps).

9. The verdict

B2B attribution shows which recorded touches preceded pipeline, split by a rule you chose. That helps spot programs that never appear near revenue. It is weak at valuing demand creation, dark social and AI discovery, and it cannot measure cause.

The setup that holds up combines three partial views: account-level CRM attribution with a fixed model, self-reported attribution on high-intent forms, and an occasional holdout test on the biggest uncertain budget line. Teams with large paid budgets and long, clean histories can add MMM. Dedicated platforms earn their cost once the CRM is clean and native reports run out.

Frequently asked questions

What is the best attribution model for B2B?

No single model is best. A W-shaped or similar milestone model fits B2B funnels because it credits first touch, lead creation and opportunity creation, but it still undervalues untracked early influence. Pair it with a first-touch or self-reported view, and test the channels where they disagree.

Can I still use first-touch or linear attribution in GA4?

No. Google removed first click, linear, time decay and position-based models from GA4, and they have not been available since November 2023.⁵ You can export GA4 data or use your CRM or a dedicated tool for those views.

Is self-reported attribution reliable?

It is reliable for spotting channels your tools cannot see, and unreliable as a precise measure because people forget and simplify. Use it next to software attribution and test what it suggests.

How do I track traffic from ChatGPT and other AI assistants?

GA4’s default channel group now includes an AI Assistant channel for referrals from tools such as ChatGPT, Gemini and Copilot.¹¹ It misses visits that arrive without a referrer and all later brand searches, so add an AI assistant option to your “how did you hear about us” field as well.

Do I need a dedicated attribution tool?

Not at first. Most teams get further by fixing contact roles, UTMs and campaign records and using Salesforce Campaign Influence or HubSpot reports. A dedicated tool makes sense once the CRM is clean and you need account-level journeys across many channels.

What is the difference between attribution and incrementality?

Attribution divides credit among touches before a conversion. Incrementality measures how many conversions would not have happened without a channel, usually through a randomized holdout. The first is continuous and cheap; the second is periodic and more trustworthy.

Sources and notes

1. Source: Forrester, “Forrester’s 2026 The State Of Business Buying”, accessed Nov 2026. ↩

2. Source: CustomerThink, “Research Round-Up: 6sense Study Provides Critical Insights on B2B Buyer Behavior”, accessed Nov 2026, summarizing 6sense’s 2025 study. Sample size from 6sense, “2025 B2B Buyer Experience Report”, accessed Nov 2026. 6sense is a vendor; these are its own survey figures. ↩ ↩² ↩³

3. Source: HubSpot Knowledge Base, “Understand attribution reporting”, accessed Nov 2026. HubSpot also offers J-shaped (20% first, 60% conversion) and inverse J-shaped (60% first, 20% conversion) models. ↩ ↩² ↩³ ↩⁴

4. Source: Adobe Experience League, “Marketo Measure Attribution Models”, accessed Nov 2026. ↩ ↩² ↩³

5. Source: Google Analytics Help, “Get started with attribution”, accessed Nov 2026. ↩ ↩² ↩³ ↩⁴ ↩⁵

6. Source: Dreamdata, “Dreamdata launches data-driven attribution for B2B”, accessed Nov 2026. Method description is the company’s own. ↩ ↩²

7. Source: Google Analytics Help, “Select attribution settings”, accessed Nov 2026. ↩ ↩²

8. Source: Salesforce Help, “Customizable Campaign Influence”, accessed Nov 2026. ↩ ↩²

9. Source: Salesforce Help, “How Customizable Campaign Influence Works”, accessed Nov 2026. Check your org’s edition and licenses before relying on the additional models. ↩ ↩² ↩³

10. Source: HubSpot Knowledge Base, “Create attribution reports”, accessed Nov 2026. ↩ ↩² ↩³

11. Source: Google Analytics Help, “Default channel group”, accessed Nov 2026. Google’s “What’s new” page dates the AI Assistant channel to May 13, 2026. ↩ ↩² ↩³

12. Source: Google Analytics Help, “About consent mode”, accessed Nov 2026. ↩

13. Source: Google Analytics Help, “Behavioral modeling for consent mode”, accessed Nov 2026. ↩

14. Source: WebKit, “Intelligent Tracking Prevention 2.1” (February 21, 2019) and WebKit, “Full Third-Party Cookie Blocking and More” (March 24, 2020), accessed Nov 2026. ↩

15. Source: HubSpot, “Marketing Hub pricing”, accessed Nov 2026. List prices as of November 2026. The page shows both $800 and $890 a month for Professional without making clear which billing option each applies to; confirm with HubSpot. ↩ ↩²

16. Source: Dreamdata, “Pricing”, accessed Nov 2026. List price as of November 2026; the page does not state the free plan’s monthly tracked user cap. ↩ ↩²

17. Source: HockeyStack, “Pricing”, accessed Nov 2026. No prices are published. ↩ ↩²

18. Source: Adobe Experience League Communities, “Marketo Measure vs. Marketo Measure Ultimate: Which Is Right for My Organization?”, accessed Nov 2026. Adobe’s pricing page did not load when checked; we found no published Marketo Measure price. ↩

19. Source: HockeyStack Docs, “Advanced attribution models”, accessed Nov 2026. ↩

20. Source: HockeyStack Docs, “Lift analysis vs. multi-touch attribution”, accessed Nov 2026. ↩

21. Source: Google Ads Help, “Set up Conversion Lift based on users”, accessed Nov 2026. ↩ ↩²

22. Source: LinkedIn Help, “Conversion Lift Testing”, accessed Nov 2026. ↩

23. Source: Google, “Meridian” on GitHub, accessed Nov 2026. ↩

24. Source: Meta Marketing Science, “Robyn” on GitHub, accessed Nov 2026. ↩