TL;DR: Buy first, build second. HubSpot lists multi-touch revenue attribution at $3,600 per month, while the GA4 to BigQuery export costs nothing. Under $50K per month in paid spend, a bought tool wins. Above it, or past a six-month sales cycle, build the warehouse model.
What multi-touch attribution actually does for SaaS
Multi-touch attribution splits conversion credit across every touch in a buying journey instead of handing all of it to the last click. In SaaS, that gap gets expensive fast. A buyer reads a comparison post in March, clicks a branded search ad in May, and books a demo in June.
Last click gives June’s branded search 100% of the credit. You then cut the March content and the May prospecting spend that created the branded search in the first place.
The real decision is narrower than most vendor pages admit. Do you assemble attribution yourself from GA4, BigQuery, and your CRM, or do you pay a platform to do it for you?
We have managed $61M+ in ad spend for 250+ clients since 2017, and B2B SaaS is one of our three core verticals. Here is how the two paths compare on numbers you can check.
Build vs buy: the short version
| Factor | Buy (platform or vendor) | Build (GA4, BigQuery, CRM) |
|---|---|---|
| Entry price | $3,600 per month for HubSpot Marketing Hub Enterprise | $0 for the GA4 BigQuery daily export, plus query and storage |
| Time to first usable report | Two to six weeks | One to two quarters |
| Who maintains it | The vendor, on the vendor’s roadmap | Your analytics engineer or your agency |
| Historical data window | Plan dependent, often capped | As long as you keep paying for storage |
| Offline and CRM revenue | Native if the CRM is the platform | Native, because you own the join |
| Model transparency | Mostly a black box | SQL you can read line by line |
| Switching cost | High: your history stays with the vendor | Low: the warehouse is yours |
| Best fit | Under $50K per month in paid spend, no data team | $50K+ per month, or a sales cycle over six months |
Verdict: buy first, build second. Most SaaS teams under $50K per month in paid media get more out of a bought tool in week three than out of a warehouse model in month nine. Past $50K per month, or past a six-month sales cycle, the bought tool stops answering your questions and the build starts paying for itself.
Why SaaS attribution is harder than eCommerce attribution
An eCommerce purchase usually closes inside a week. A SaaS deal does not. That single difference breaks most default measurement setups.
Three things make it worse. Several people touch the account before anyone fills in a form. Much of the research happens on review sites and in private communities you cannot tag. And the revenue lands in the CRM months later, far away from the ad platform that started it.
Gartner’s research on B2B buying groups is the standard reference for how many people sit inside one complex purchase decision. Gartner’s research on B2B buying finds that the median buying group for a complex B2B solution involves six to 10 decision makers, each armed with four or five pieces of information they have gathered for themselves.
Trials complicate it further. A self-serve trial signup is a cheap conversion that any model can see. The paid upgrade 45 days later is the one that pays your bills, and it usually happens in a different session, on a different device, with no ad click anywhere near it. Optimize to the signup and you buy volume. Optimize to the upgrade and you need multi-touch data.
The cost of guessing wrong is not small. Across iClick-managed accounts in 2024 and 2025, SaaS and B2B software ran a $4.88 average CPC and a 3.80% conversion rate, with a $14.50 CPM on paid social. That is the second most expensive click of any vertical we run, behind personal injury law at $6.75.
At $4.88 a click, a misattributed campaign is not a rounding error. Kill the wrong top-of-funnel campaign and you feel it two quarters later, when pipeline dries up and nobody can point at the cause.
The buy path: what the money gets you
All-in-one platforms
If your CRM already holds the revenue, buying attribution from the CRM vendor is the shortest route. HubSpot lists Marketing Hub Enterprise at $3,600 per month with five Core Seats included, and $75 per month for each seat after that. Multi-touch revenue attribution sits at that Enterprise tier, tracking up to 10,000 logged interactions per contact.
Professional is $800 per month with three seats, and it does not include multi-touch revenue attribution. So the $2,800 per month step between the two tiers is the real price of the feature.
Salesforce and Adobe price in the same range or higher, and most enterprise vendors quote rather than publish.
Pure-play attribution vendors
Dreamdata, HockeyStack, and Factors sell B2B attribution as the whole product rather than as a module. Dreamdata publishes a free tier: five seats, three stage models, one sync connection, and two months of user history. Everything above that is custom pricing.
Read those two limits together. Two months of user history against a nine-month sales cycle means the free tier cannot see the touch that started the deal. That is not a knock on Dreamdata. It is the shape of the whole category, and it is the first question to ask any vendor on a demo call.
Where buying breaks
Bought tools fail in three predictable places.
- They model what they can see. Paid social clicks that never set a cookie stay invisible no matter what you pay.
- The history belongs to the vendor. Cancel, and 18 months of touch data leaves with them.
- The model is closed. When finance asks why paid search received 34% of the credit, “the algorithm decided” is not an answer.
The build path: GA4, BigQuery, and your CRM
What it actually costs
The starting piece is free, which surprises people. Google Analytics Help confirms the GA4 to BigQuery daily export carries no GA4 charge, with a limit of 1 million events per day on standard properties. Analytics 360 properties get up to 20 billion events per day. Streaming export removes the volume limit and costs $0.05 per gigabyte on top of standard BigQuery charges.
For a SaaS site under a million events a day, which is nearly all of them, the raw event feed costs nothing. You pay for storage, for queries, and for the person who writes the SQL. That last line is the one that decides the business case.
What the model needs to join
A working warehouse model needs three tables and one shared key. Ad platform cost by campaign and day. GA4 sessions with a stable user ID. CRM opportunities with a close date and a revenue amount.
The shared key is the hard part. Most SaaS sites pass a GA4 client ID into a hidden form field, store it on the contact record, and then push it back with the closed-won deal. Get that one field right and the SQL is a weekend of work. Get it wrong and no vendor, at any price, can rescue the numbers.
The 14-month wall
Here is the reason we push warehouse builds on longer sales cycles. Google’s own documentation caps event-level data retention at 14 months on standard GA4 properties, with two months as the only other option. Retention of 26, 38, or 50 months requires Analytics 360.
A nine-month sales cycle plus a year of year-over-year reporting does not fit inside 14 months. The BigQuery export gets around this, because rows you have already exported are yours to keep. Turn it on before you need the history, not after. Retention settings do not work backwards.
Where building breaks
Builds fail on people, not on technology. Someone has to own the identity join between an anonymous GA4 client ID, a form fill, and a closed-won deal in Salesforce. That person leaves, and the model rots inside a quarter.
Builds also fail on scope. Teams set out to answer one attribution question and end up rebuilding the whole reporting stack. Our measurement, tracking, and attribution playbook covers the sequencing that keeps a build contained.
What the ad platforms already decided for you
Two platform changes matter before you spend a dollar on either path.
First, Google Ads retired most rule-based attribution. Google Ads Help states that the first click, linear, time decay, and position-based models are no longer supported, and that conversion actions using them were upgraded to data-driven attribution. Data-driven is now the default for most conversion actions, and last click survives only as a manual override.
The multi-touch model inside Google Ads is no longer yours to choose. If you want a first-touch or position-based view, you are building it outside the platform. Our comparison of B2B attribution models from first touch to data-driven walks through what each model actually rewards.
Second, the cookie deadline everyone planned around went away. On April 22, 2025, Google’s Privacy Sandbox team wrote that it would “maintain our current approach to offering users third-party cookie choice in Chrome, and will not be rolling out a new standalone prompt for third-party cookies.”
That is not permission to relax. Safari and Firefox still block third-party cookies by default, and link decoration limits still strip click IDs. Server-side tracking is the fix either way. If your SaaS company also sells physical product, the same plumbing shows up in our guides to Shopify conversion tracking with GA4, enhanced conversions, and the Conversions API and the Meta Conversions API setup for Shopify stores.
How we decide, account by account
We use three rules.
- Under $20K per month in paid spend: buy nothing new. Fix conversion tracking, switch on the BigQuery export, and use GA4 data-driven attribution with a 90-day lookback window.
- $20K to $50K per month: buy. A CRM-native or pure-play tool answers the question faster than a build, and the license costs less than one month of an analyst’s salary.
- Over $50K per month, or a sales cycle longer than six months: build the warehouse model, and keep a bought tool for the dashboard your executives read.
iClick works with North American brands spending $20K+ per month, so most of our SaaS clients land in rule two or rule three.
The rules exist because attribution is a means, not the goal. On B2B SaaS accounts we push three numbers: CAC down, trial signups up, and demo volume up. Across that vertical we have cut CAC by 30%, lifted trial signups 60%, and doubled demos. Trademark Bob founder Brad Walz put it at a “30% CAC drop in under 90 days”. Talent Room CMO Oliver Auerbach said we “2x’d our demo requests without increasing spend” inside one quarter.
The hybrid that works most often
Most SaaS teams do not need to pick a side. The setup we run looks like this.
- GA4 with server-side tagging and enhanced conversions, exporting to BigQuery from day one.
- A bought tool for the weekly view your CMO reads, chosen for CRM fit rather than model sophistication.
- One SQL model in BigQuery that joins ad platform cost, GA4 sessions, and closed-won revenue on a single ID.
- A quarterly geo holdout or spend pause to check whether the model is telling the truth.
That last line matters more than the model choice. Attribution describes correlation. An incrementality test measures cause. Run one every quarter and you find out which channels were coasting on credit they never earned.
Our portfolio average is a 41% lift in conversion goals, and none of that came from a fancier attribution model. It came from measuring the right thing, then spending against it.
Get a second read on your tracking before you buy anything
Most of the SaaS accounts we audit do not have an attribution problem. They have a conversion tracking problem wearing an attribution costume. Duplicate conversions, missing enhanced conversions, and a CRM that never sends revenue back all produce numbers no model can fix.
Send us the account and we will check it. The free audit is a 47-point written PDF, delivered in five business days, with no sales call attached. Request your free written PPC audit and find out what your current data is actually telling you.
Sources
- HubSpot, Marketing Hub pricing (Enterprise $3,600/mo, multi-touch revenue attribution), checked September 2026
- Google Analytics Help, Data retention (standard properties capped at 14 months; 360 up to 50 months)
- Google Analytics Help, BigQuery Export (1 million events per day standard, $0.05/GB streaming)
- Google Ads Help, About attribution models (first click, linear, time decay, and position-based no longer supported)
- Google Privacy Sandbox, Next steps for Privacy Sandbox and tracking protections in Chrome, April 22, 2025
- Dreamdata pricing, free tier limits (two months of user history), checked September 2026
- iClick Advertising 2026 PPC benchmarks, from iClick-managed accounts 2024 to 2025
- Gartner, “Buyer Enablement: Win More B2B Sales Deals” / The B2B Buying Journey
Frequently asked questions
Is multi-touch attribution worth buying for a SaaS company under $20K per month in ad spend?
Usually not yet. At that level, fix conversion tracking first, switch on the free GA4 BigQuery export, and use Google’s data-driven attribution with a 90-day lookback. A $3,600 per month platform will not fix duplicate conversions or a CRM that never sends revenue back. Buy attribution after the tracking is clean.
Can I still use first-touch or time decay attribution in Google Ads?
No. Google Ads Help confirms the first click, linear, time decay, and position-based models are no longer supported, and that conversion actions using them were upgraded to data-driven attribution. Only data-driven and last click remain. If you want a first-touch view of your SaaS pipeline, you have to build it outside Google Ads.
How much does it cost to build multi-touch attribution in BigQuery?
The GA4 daily export is free for standard properties, up to 1 million events per day. Streaming export costs $0.05 per gigabyte on top of standard BigQuery charges. Your real costs are storage, queries, and analyst time. For most SaaS sites the cloud bill is small, and the people cost decides the business case.
Does GA4 keep enough history for a nine-month SaaS sales cycle?
Not on its own. Standard GA4 properties cap event-level data retention at 14 months, with two months as the only other option. Analytics 360 offers 26, 38, or 50 months. A nine-month cycle plus year-over-year comparison exceeds 14 months, so switch on the BigQuery export early. Retention settings do not apply retroactively.
Do we still need server-side tracking now that Chrome is keeping third-party cookies?
Yes. Google confirmed in April 2025 that it will keep its current third-party cookie approach in Chrome, but Safari and Firefox still block them by default, and link decoration limits still strip click IDs. Server-side tagging and enhanced conversions recover conversions that browser-side tags lose, whichever attribution path you choose.
