B2B Marketing Attribution: Models, Data and Revenue (2026)

It is written for the people who have to defend a B2B marketing budget with CRM numbers: marketing leaders, demand generation and RevOps teams at software and services companies, and the agencies that run paid channels for them. If you already run multi-touch attribution and want the models compared, the B2B multi-touch guide is the shorter read. If your question is why the pipeline report and the ad dashboards disagree at all, start here.

B2B marketing attribution is the method of assigning credit for a business-to-business sale to the marketing touchpoints that contributed to it: the ads, searches, content, emails and events a buyer encountered between first hearing of a vendor and signing a contract. The definition is the same as in any other market. What changes in B2B is the thing being credited. It is not a purchase made in one session by one person, it is a CRM deal, reached through a pipeline of stages, by a group of people, over a period measured in months.

That changes the mechanics completely. A shop can attribute an order to a session because the order happens in the session. A B2B company has to attribute a deal to sessions that happened a quarter earlier, on several devices, by several colleagues, most of whom never filled in a form. The credit has to travel from an ad platform, through a website, into a CRM, and wait there until the deal closes. Every method and every tool in this guide is a way of making that journey hold.

Three phrases get typed into Google for this, and they are three emphases on one discipline rather than three disciplines. It is worth being precise about them, because a tool that is strong on one is often silent on another, and a buyer who asks for the wrong one gets a demo of the wrong report.

• B2B marketing attributionCredits the marketing touchpoints: the ad click, the organic search, the webinar, the email. The question is which marketing activity earned the deal, and the output is a channel and campaign report.
• B2B revenue attributionPuts the closed amount at the centre. The unit credited is euros from won deals rather than leads or conversions, so a channel with cheap leads and no revenue looks like what it is.
• B2B sales attributionCounts sales touches as touchpoints too: the discovery call, the demo, the email sequence, the meeting. The question is which combination of marketing and sales activity moved the deal to closed-won.

Most attribution writing assumes an e-commerce shape: one buyer, one session, one order, a conversion pixel that fires when the money moves. B2B breaks every one of those assumptions, and the breakages are what make the discipline its own subject rather than a variant of conversion tracking. Four of them matter more than the rest, and each one removes a shortcut that works perfectly well for a shop.

• A committee buys, not a person

  Gartner's research puts the typical B2B buying group at six to ten people. One of them fills in the form. The others read, compare and object without ever identifying themselves.

• The cycle runs for months

  The click that started a deal and the signature that ended it can be a quarter apart. A seven day browser cookie has forgotten the first touch long before the second pipeline stage.

• Half the touches are offline

  Calls, demos, meetings, a conversation at an event, a referral from a customer. None of them fires a pixel, and all of them move deals. The record has to hold them beside the clicks.

• The outcome is a deal, not a form

  A form fill is where the anonymous visitor becomes a lead, not where the value is created. Value arrives at closed-won in the CRM, months later, at an amount no ad platform ever sees.

The last one has a consequence that the rest of this guide keeps returning to: the CRM is the system of record. Not the ad platform, whose conversion is a form fill inside its own window. Not the analytics tool, whose user is a browser. The CRM is the only system that knows a deal was qualified, opened, negotiated and won, and what it was worth, so any attribution that does not end in the CRM is attributing a proxy for the thing that pays the bills.

None of this is an argument about last click. The two multi-touch guides here already make that case, and it is a true one. The B2B argument is one step earlier: before any model can split the credit, the record has to reach from the first anonymous click to the closed deal, across the people, the months and the offline touches in between. A perfect model over a record that stops at the form is a perfect model of the wrong thing.

Every B2B attribution product makes one decision before any other, and it shapes what the product can and cannot see: the unit it tracks is either a named lead or an account. Vendors do not always say which, and buyers rarely ask, so a great deal of the disappointment in this category comes from buying one architecture while needing the other. It is worth settling before the first demo.

Lead-level attribution follows a person. A visitor arrives from a click, is tracked first-party as an anonymous visitor, and becomes a named lead at the form, the call or the booking. Everything the visitor did before that moment is attached to the person, and the person is attached to the CRM deal they sit on, so the deal inherits the journey. Account-level attribution follows a company. It identifies the organisation behind anonymous visits, puts every stakeholder's touches on one company timeline, and credits the opportunity on that account from the combined record, including people who never filled in anything.

Account-level is the right answer more often than a lead-level vendor would like. If your deals are six figures, your buying committee is large and mostly anonymous, and marketing runs account-based programmes against a named target list, then the thing you need to see is the account's timeline, and only a product that identifies companies can draw it. That is a real capability and it is worth paying for when the deal shape demands it.

Four products are built that way. Dreamdata models the whole account journey and identifies the companies behind anonymous visits, with audiences built from that model and synced to the ad platforms. HockeyStack resolves an enterprise's CRM, ad, ABM and warehouse sources into one dataset, with stakeholder maps, account scoring and an AI analyst inside Slack. Factors.ai identifies visiting companies, scores them and steers LinkedIn and Google Ads at them. Adobe Marketo Measure writes every-touch attribution into Salesforce or Microsoft Dynamics, with BDR and sales activities as touchpoints. Three of the four sell on quote to organisations with a RevOps team; Factors.ai publishes a price list by tracked users from $199 a month.

Lead-level is the right answer for the larger number of B2B companies whose buyer does identify. If demand arrives through paid search, paid social, content and referrals, and the decisive moment is a demo request, a trial signup or an inbound call, then the person who converted is the person whose journey matters, and the deal they open in the CRM is the outcome. That architecture needs no company identification, sets up in an afternoon rather than a quarter, and is what public per-month pricing is possible on. Its limit is stated plainly in the LeadJourney section below: it sees the named lead's journey, not the anonymous colleague's.

The test is one question: at the moment your CRM opens an opportunity, how many of the people who influenced it have already identified themselves on your site? If the answer is most of them, track leads. If the answer is one out of eight, track accounts. The comparison pages for Dreamdata, HockeyStack, Factors.ai and Adobe Marketo Measure run that question against each product row by row, and say where the account-level product wins.

In e-commerce there is one conversion and it is the order. In B2B there are four, and they are the stages of the pipeline: the lead (a person identified at a form, call or booking), the qualified lead (sales agreed they fit), the opportunity (a deal opened with an amount) and closed-won (the contract signed and the amount real). Each is a conversion in the strict sense, an event with a date that a marketing touch can be credited for, and each one is set by a different person in a different week.

Treating those stages as the conversion ladder is what turns attribution from a marketing report into revenue analytics. The report is the funnel by channel with money at both ends: what each channel cost on the left, what its leads became stage by stage across the middle, and what they were worth at closed-won on the right. Here is the quarter from the introduction laid out that way, each lead credited to the channel of its first click.

Read the Meta and LinkedIn rows against each other. Meta produced 74 leads for €9,000 and LinkedIn 71 for twice the money, so on a cost per lead report Meta is the channel to scale. Then follow the rows to the right. Fifteen of Meta's leads qualified and one closed; thirty-one of LinkedIn's qualified and six closed, for seven times the revenue. The lead count told the opposite of the truth, and the only way to know was to carry each lead through the stages it reached.

Dividing the spend by each stage gives the unit economics per channel, and the further right you go the more the ranking changes. Every figure here is the spend divided by the count in the table above, rounded to the euro, and the last column is the revenue divided by the spend.

Meta has the cheapest lead and the most expensive closed deal on the account, three times LinkedIn's. Microsoft, the smallest line item, has the cheapest qualified lead and returns more than nine euros per euro. Neither fact is visible in any ad platform, because no ad platform knows which of its conversions qualified, opened or closed. This is the report the phrase B2B revenue analytics is reaching for: not a dashboard of more metrics, but the same funnel your sales team runs, with the cost of each channel at the top of it and its closed revenue at the bottom.

Two practical rules follow. First, the stage names are the CRM's, not the tool's: if your pipeline says Discovery, Proposal and Negotiation, those are the conversions, and a tool that makes you rename them to its own vocabulary will be maintained by nobody. Second, the deal amount travels with the stage. A closed-won conversion without its value is a count, and the whole point of the ladder is that the right hand column is in euros. Lead quality by channel is the shorter statement of the same argument, and the true cost per lead works the arithmetic for a single channel.

0:40

The funnel above credits two different kinds of money, and they answer different questions at different times. Pipeline attribution credits the value of open opportunities to the touchpoints that produced them, as soon as the opportunity is opened. Revenue attribution credits the closed amount, and only once it has closed. In the quarter above, pipeline attribution has €1,148,000 to distribute across 41 opportunities the moment they are created; revenue attribution has €392,000 across 14 deals, and can only hand it out at the end.

During a long cycle, pipeline attribution is the operating report. A channel whose clicks in January become opportunities in February is doing its job, and a team that waits for closed revenue before judging it will judge it in May, two budget cycles too late. By first click, the €1,148,000 of pipeline in the example splits into €510,000 opened by LinkedIn Ads, €468,000 by Google Ads, €95,000 by Meta and €75,000 by Microsoft, and that ranking is available in week six of the quarter rather than week thirteen.

Revenue attribution is the audit. Pipeline is a promise, and some channels produce promises that do not close: a channel that opens large opportunities with a low win rate looks excellent on pipeline attribution and ordinary on revenue attribution, and only the second report catches it. The disciplined setup runs both on the same record, reads pipeline weekly to steer and revenue quarterly to correct, and keeps the deal amount on the record so the two reports are the same money at two dates rather than two estimates. The glossary entry on revenue attribution has the compact definition.

Revenue attribution also has a dimension pipeline attribution mostly lacks: the model. Once €392,000 has closed, the question of which touchpoints earned it has several defensible answers, and a B2B journey with five or more touches makes them diverge sharply. Here is the same closed revenue credited three ways. The first click column is the funnel table's revenue column, because that table credited each lead to the channel that opened its journey. Organic search and direct appear as a row, because in the example every journey was opened by a paid click but many ended on a brand search or a typed URL.

LinkedIn earns €182,000 under first click and €74,000 under last click, because it opens journeys that Google's brand search and a direct visit close. Google Ads moves the other way, from €156,000 to €198,000, for the same reason seen from the other end. Linear sits between them and is the honest default when nobody can defend a weighting. What matters is not which column is right but that the total is the same in all three: once the credit is split from one record, the argument moves from whose number is true to which split of one true number best describes the journey.

In the funnel above, marketing's influence does not stop at the lead. Between the qualified lead and the closed deal sit a discovery call, a demo, two proposals, an email sequence and probably a conversation nobody logged, and each of those is a touchpoint in the same sense as the ad click. B2B sales attribution is the practice of counting them: putting the calls, meetings, emails and sequences on the same timeline as the marketing touches, so the record shows what moved the deal after it became a deal.

The enterprise tools have done this for a long time. Adobe Marketo Measure counts BDR and sales activities, calls, emails and meetings, as touchpoints alongside paid media, webinars and events, written into Salesforce or Dynamics. HubSpot's attribution reports count calls and meetings logged in the CRM among the interactions they credit. The value is real: a channel whose leads need three demos to close is a more expensive channel than one whose leads close after one, and only a record that holds the sales touches can show that.

The trap is as old as the practice. The last touch before closed-won is nearly always a sales touch, because the signature follows a meeting, so any model that leans on recency hands the credit to the rep who happened to run the final call. Under last touch, marketing's contribution to a six month deal evaporates into the closing meeting. Under a time decay model with a short half-life, it fades to a rounding error. The rep did their job; the model simply stopped looking at the four months before the rep was involved.

The fix is structural rather than a matter of modelling: keep the marketing and sales touches on one record, and report them as two layers rather than one contest. The marketing layer answers which channels produced and progressed the lead; the sales layer answers which activities moved the opportunity; and a model that credits across both should be read as a description, not a verdict on either team. In practice the deal record carries the source, the campaign and the first click beside the logged calls and meetings, and neither field overwrites the other when a rep updates the stage. The per-lead journey view is where that record is read: one person, every touchpoint of either kind, with its date.

Sales attribution also has the most honest limit in the discipline. Many of the sales touches that matter, the referral at a dinner, the call from a mobile, the colleague who forwarded the proposal, never reach any system. A record can hold what was logged; it cannot hold what was not. Getting reps to log calls and meetings is therefore an attribution project as much as a sales operations one, and the tools that make it automatic, call tracking with a number per campaign, a booking link that writes to the CRM, earn their place on that ground alone.

The five standard models are the same in B2B as anywhere else, and two guides here cover them in depth: multi-touch attribution for B2B works through each one against long cycles, and multi-touch attribution shows them as a switch on a report. What belongs in this guide is the question each model answers in a pipeline that runs for months, because in B2B the choice of model is really a choice of question.

• [First click](/glossary/first-click-attribution) answers where demand was created. In a committee sale it is the only model that consistently credits the non-brand search or the LinkedIn campaign that made the account aware of you, months before anyone converted.
• [Last click](/glossary/last-click-attribution) answers what the decided buyer touched on the way in. It is useful for landing page and form work and misleading for budget, because in B2B the last click is almost always brand search or direct.
• [Linear](/glossary/linear-attribution) answers which channels were present in journeys that closed, weighting none of them. It is the honest default over a long cycle, and the one to read first when nobody can defend a weighting.
• [Position-based](/glossary/position-based-attribution) answers what opened and what closed, with 40% each on those two and 20% across the middle. HubSpot's U shaped model is this shape, and it fits a funnel where creation and conversion are the two expensive moments.
• [Time decay](/glossary/time-decay-attribution) answers what was recent, on a half-life of seven days in HubSpot's version. Over a six month cycle it starves the first quarter of the journey, so it suits short cycles and retargeting analysis more than budget allocation.

The enterprise tools add shapes built around the pipeline milestones themselves. HubSpot's W shaped model puts 30% on each of three milestones and 10% across the rest, its full path model 22.5% on each of four with 10% between them, and its J shaped and inverse J models, at 20/60/20 and 60/20/20, move most of the weight to one milestone. Adobe Marketo Measure runs U shaped, W shaped, full path, first touch and lead creation simultaneously and adds custom weightings. These answer a stage question, which touch preceded each milestone, and they only work on a record that holds the milestones, which brings the discipline back to where it started.

Whatever the model, the useful reading is the same: switch it and watch which channels change rank. A channel that holds under every model is genuinely performing, and one that only wins under last click is being paid for demand another channel created. That comparison is only made if switching costs nothing, which is an argument about the report rather than about the model. See attribution model for the compact definitions of all of them.

Before buying anything, it is worth knowing exactly what the systems you already pay for will attribute, because the honest answer is more than nothing and less than the sales deck suggests. Four of them come up in every B2B evaluation, and each attributes what it can see and nothing beyond it.

HubSpot

HubSpot has the most complete attribution of the three CRMs. Marketing Hub ships nine models (linear, first interaction, last interaction, U shaped, W shaped, full path, J shaped, inverse J and time decay on a seven day half-life), and contact create attribution is available from Marketing Hub Professional. Deal create attribution and revenue attribution, the reports this guide is about, need Marketing Hub Enterprise. Every report buckets interactions by the session HubSpot's tracking code recorded in the browser, so a visitor who declined the analytics cookie category contributes nothing.

Two more limits matter for paid channels. Only the gclid is captured natively; fbclid, li_fat_id and msclkid need a hidden field and your own script. Ads ROI is estimated from an average sale price and a contact-to-customer rate rather than measured from deals, and conversions sent back to the platforms carry a fixed value you type per event, not the deal amount. HubSpot Marketing Hub attribution explained has the detail, and the best attribution tools for HubSpot covers what sits in front of it.

Salesforce Campaign Influence

Salesforce answers a different question. Customizable Campaign Influence, in the Professional, Enterprise, Performance, Unlimited and Developer editions, credits an opportunity across the campaigns its contact roles are members of, on a predefined model or on percentages you enter, within an auto-association time frame you set. It is a good tool for the question it asks. It knows nothing about a click that never became a campaign member, so the ad, the organic search and the anonymous visit are outside it unless something else wrote them onto a campaign. Salesforce Campaign Influence explained covers the setup and the best attribution tools for Salesforce the tools that feed it.

Pipedrive

Pipedrive has no attribution models and no ad spend anywhere in the product. Insights will report won value by any field on the deal, including a source field a rep picks from a list, which is a real report if the field is maintained and a fiction if it is not. Everything else comes from the tool placed in front of it, and the best attribution tools for Pipedrive ranks them by what they write onto the deal.

LinkedIn's Revenue Attribution Report

Not a CRM, but it appears in the same conversations. The Revenue Attribution Report lives in LinkedIn Business Manager, needs a Business Manager admin, connects Salesforce, Dynamics 365 or HubSpot by OAuth, and reports revenue won, return on ad spend, pipeline amount, leads, open and closed-won opportunities, win rate, average deal size and average days to close. Its influence is impression-based or engagement-based, over a lookback window of 30 to 365 days (180 by default) and a configurable number of touchpoints.

Two things to understand before quoting it in a budget meeting. It reports LinkedIn's influence only, with no view of Google, Meta or organic. And influenced means the account touched LinkedIn inside the window, not a split of credit: a deal can be influenced by LinkedIn under this report and be credited, by every model in the previous section, to something else. The LinkedIn Ads integration is the version of this loop where the deal value goes back to LinkedIn matched on the click.

Every B2B company with a data team considers building this. The plan is always the same: the GA4 BigQuery export for the touchpoints, the CRM's own export for the stages and amounts, a cost pipeline from the four ad platforms, and dbt to join them into a model the BI tool can read. It is a reasonable plan, and it is worth knowing where it runs into the ground before committing a quarter of an analyst to it.

The GA4 export is the weak link, for reasons the BigQuery attribution guide works through in SQL. It carries no cost or spend field, so cost per qualified lead has nothing to divide by until a second pipeline pulls spend from Google, Meta, LinkedIn and Microsoft on their own schedules and joins it on campaign names as each platform spells them. Its conversions are, in Google's own description, last click observed with no modelling, so the modelled conversions in the GA4 dashboard never reach the warehouse and the two will never agree. Its identity is a browser rather than a person, and it has never heard of your CRM.

There are teams that should build anyway. If you already run a warehouse with the CRM and the ad platforms in it, have an analyst who owns the models, need attribution joined to product usage or billing data no vendor holds, or sell in a way no packaged tool's stage model fits, then the build is a few weeks on top of infrastructure you already pay for and the result is yours. The honest test is whether the warehouse and the pipelines exist today. If the attribution project is the reason to build them, the platform route is cheaper by an order of magnitude and live this week.

The two routes are no longer exclusive. A platform that streams its own rows into your warehouse gives the data team the raw material without the collection problem. LeadJourney's BigQuery export writes raw clicks, conversions and leads into your own dataset, each click already carrying the campaign, the click id and the conversion it reached, so the join the GA4 export makes impossible is already done on the row. The marketing team gets the reports on Monday; the analysts get the same rows for the questions no report answers.

Seven steps, in the order the data depends on itself. A team that starts at step six builds a model over a record that cannot carry it, and the touches not collected in the first quarter are not available in the second. Each step is two sentences here and a page of its own elsewhere on the site.

1. Capture the journey server-side, on your own domain. A browser pixel loses a share of every channel's clicks to Safari, iOS, ad blockers and consent tooling, and loses them unevenly, which biases every comparison downstream. Server-side tracking with an identifier set by your own domain and kept on the server is what puts the accuracy in the nineties.
2. Read the click ids at the first visit and keep them. gclid with gbraid and wbraid, fbclid, li_fat_id, msclkid and ttclid, plus the UTM parameters and the landing page, read once at the first visit and stored against one visitor id that survives weeks and months. How the gclid is captured and stored is the long version.
3. Resolve the visitor to a person at the form, the call or the booking. That is the moment the anonymous journey gets a name, and everything the same visitor did before it, on any device, attaches to the lead. Without this step there is a journey and a lead and no link between them.
4. Write the source onto the CRM contact and the deal. The channel, campaign, ad and landing page belong in fields on the record in HubSpot, Salesforce, Pipedrive or whatever you run, where a sales manager can filter them. A source that lives only in a reporting tool is not trusted by the people who close the deals, and the CRM lead source is where that trust is won or lost.
5. Make the pipeline stages the conversions, with the deal amount on each. Qualified, opportunity and closed-won are the conversions the model credits, named as your CRM names them, and the amount travels with the stage. This is the step that turns a cost per lead report into revenue attribution.
6. Choose a model per question, and keep it switchable. First click for where demand was created, linear as the default, position-based for the two expensive moments, on the same report with a switch rather than a re-import. The comparison between models is where the insight is, and it only happens if switching costs nothing.
7. Send the closed stages back to the platforms. Once the record knows which leads qualified and closed, those stages go back to Meta, Google, LinkedIn and Microsoft as conversions with the deal value, so the platforms optimise toward deals rather than form fills. Optimising on revenue is what this step buys, and it is the one that makes the rest pay for itself.

Steps one to four are the record and cannot be retrofitted: the data you did not collect in March is not available in June. Steps five to seven are the reporting and can be changed in an afternoon. The order is the whole method, and it is also why a tool should be judged on how it does the first four before anyone looks at the models in the sixth.

Above a certain size the questions change. A group with several brands, markets or business units does not need attribution; it needs one attribution standard, so that a euro of pipeline in the German subsidiary means the same thing as a euro in the British one and the board sees one report rather than five. That means one stage model across the units, one set of models switchable on every report, and one place where a RevOps team can see every unit without logging into each.

Then the procurement questions, which in the EU come before the technical ones. Where is the data hosted: LeadJourney's is in Frankfurt. Is there an Article 28 data processing agreement: yes, and the vendor security review a procurement team runs is supported. Of the four account-level platforms in the earlier table, Dreamdata publishes data centres within the European Union, Factors.ai publishes a hosting zone in the United States, and the other two publish no region on their sites, so the question has to be asked rather than assumed. Ask it of every vendor, including us.

The third set of questions comes from the BI and RevOps team, who will not read attribution in a vendor's dashboard: they will pull it into their own. That needs an API, a warehouse export and, increasingly, an MCP connection so an analyst can ask a model to query the workspace directly. LeadJourney's BigQuery export streams raw clicks, conversions and leads into the customer's own dataset, and the MCP server lets Claude or ChatGPT query a workspace. Snowflake is on the roadmap and is not claimed here.

Finally the contract and the SLA, which a public per-month price does not by itself provide. The enterprise page sets out how a multi-entity standard, the DPA, the security review, the API and MCP access and the contract fit together. What it does not claim is a SOC 2 report, an ISO 27001 certificate or a HIPAA business associate agreement; a procurement team that needs one of those should say so at the first call rather than the last.

LeadJourney is an all-in-one attribution platform for lead generation, B2B SaaS, e-commerce and the agencies that run marketing for them; this section is about what it does for a B2B cycle. Tracking runs server-side on your own domain at 95%+ accuracy. The click ids are read at the first visit and kept, gclid with gbraid and wbraid, fbclid, li_fat_id, msclkid and ttclid, together with the UTM parameters and the landing page, all stored on the LeadJourney Click ID: an identifier of our own for one visitor, set on the first visit, kept in the browser for weeks or months, with every later session appended to it. That is what lets a click in January and a form in March be the same person.

At the form fill, the call or the booking the visitor becomes a named lead, and from there the record follows your CRM stages to the closed deal: natively on HubSpot, Salesforce, Pipedrive, Close, Attio, GoHighLevel, ActiveCampaign and Odoo, and by webhook, Zapier, Make, n8n or the API on anything else. Closed revenue is credited back across the touchpoints under first click, last click, linear, position-based or time decay, switchable on the channel report, the campaign report and the landing page report, on every plan. The per-lead journey view shows one person's touchpoints with their dates, the AI search engines are sources in their own right, and Atlas, the AI analyst, answers questions about the workspace.

The loop closes the other way as well. CRM stages and closed deals go back to Meta, Google, LinkedIn and Microsoft as conversions with the deal value, on every plan, so the platforms optimise toward the leads that closed. And the honest sentence this guide promised: LeadJourney is lead-level attribution, not account-level. It follows the named person who identified and joins them to the deal; where a deal has several contacts, each contact's journey sits on the deal's record, but it does not identify anonymous companies, model a buying committee across stakeholders who never converted, or run marketing mix or incrementality models.

For an inbound or paid B2B funnel where the buyer identifies at the demo request, that is the architecture that fits; for a mostly anonymous committee, the account-level section above names the products built for it. Setup takes about 21 minutes, from €129 a month, with a 14-day free trial and no credit card, and the whole product is clickable on demo data in the live demo before anyone talks to sales.

The models in depth: multi-touch attribution for B2B and multi-touch attribution for B2B lead generation. The channel question: cross-channel attribution and the best cross-channel attribution tools. The buyer's side of the record: the B2B buyer journey.

What the CRMs do and what sits in front of them: attribution tools for HubSpot, for Salesforce and for Pipedrive, with HubSpot Marketing Hub attribution and Salesforce Campaign Influence explained. The account-level products, row by row: Dreamdata, HockeyStack, Factors.ai and Adobe Marketo Measure. On the product side, attribution for enterprises, attribution for SaaS companies and the marketing attribution software category page.