What is marketing attribution?
Marketing attribution is the practice of assigning credit for a conversion to the marketing touchpoints that preceded it.
Someone sees a paid social ad, searches your brand a week later, clicks an email, and buys. Attribution decides how much of that sale belongs to each step. That is the whole concept, and it is not controversial.
What makes the term slippery is that it names a job rather than a kind of software. The core definition is relatively settled; the disagreement is over which system should answer it — and buyers disagree with themselves, often inside a single interview. The word gets attached to web analytics platforms, ad-management layers, CRMs, mobile measurement partners and dedicated measurement vendors, and all of those uses are ordinary.
The short version. Attribution is a question your stack answers, not a box on an architecture diagram. Before evaluating tools that promise to do it, it is worth knowing which tool is doing it today — because in many stacks, something already is.
Which tools do buyers use for marketing attribution?
Google Analytics is named more often than any other product for attribution in these interviews, even though it is a web analytics platform rather than a dedicated attribution product.
In the neighbouring categories, the most-named product is usually purpose-built for that category. Email platforms lead the email category. CRMs lead CRM. Dedicated incrementality and mix-modeling vendors lead their own. Attribution is the exception here — the tool named most is a web analytics platform rather than a dedicated attribution product.
The most-named product in each category, from verified buyer interviews. Attribution is the one whose leader belongs to a different category.
You can watch the word stretch inside one interview. A hospitality technology company describes using GA4 for web analytics and simple attribution, then says the same platform has significant gaps in attribution and data capture. A few questions later the same buyer describes their search-ads management layer as a tool that provides us with attribution, praising its robust search attribution. Two products, one word, two different jobs — which shows how broadly buyers use the term.
Others stretch it further. A consumer-internet company whose marketing targets professional users is working on measuring and attributing product adoption, and reports that teams struggle to agree on how to define and gauge tool usage. A large B2B software company frames the goal as building a more robust attribution system, whether in-house or outsourced — a system to be constructed, not a product to be bought.
Why the analytics tool struggles with the job
The buyers asking most of it are the least satisfied with it.
GA4 averages 5.8 among buyers who are in market for attribution. Among everyone else in the corpus, the same product averages 6.6. That gap is the clearest signal on this page: it is not that GA4 is badly regarded generally, it is that the people leaning on it for this specific job rate it lower than the people who are not.
The reasons buyers give are structural rather than cosmetic. A travel operator whose customers take a year or more to book runs attribution out of their CRM on hand-built tracking parameters, and describes the result plainly: the system credits most success to our website due to last-click attribution, masking true channel performance. The same buyer names a disconnect between their ad platform and their CRM that makes results hard to track and attribute at all.
That is the shape of the problem. The analytics platform is built around the web journey it observes, and it is weakest on the channels and outcomes that do not naturally appear there — offline purchases, non-click media, or a customer returning months later under a different identifier. Asked for a report it can produce, it does well. Asked which marketing worked, it answers from the touchpoints it can observe, which can overweight the parts of the journey closest to conversion.
The counter-case is instructive. A DTC apparel brand that moved attribution onto a dedicated platform rates that platform a ten, and rates GA4 an eight — explicitly for its reliability and its integration with the attribution tool, into which it now flows. Relieved of the job, the analytics tool scored well. That example suggests some of the dissatisfaction comes from asking the analytics platform to carry an attribution job beyond its natural scope.
What are the main attribution models?
The vendor taxonomy contains several attribution models, but buyer language is concentrated heavily in two: last-click and first-click.
All credit to the final touchpoint before the conversion. The model buyers mention far more often than any other, and the one they most often describe being stuck with rather than choosing.
All credit to the first known touchpoint. Flatters discovery channels for the same reason last-click flatters closing ones.
Credit split evenly across every touch in the path. Simple and easy to explain, but it assigns equal credit to touchpoints regardless of their actual influence.
More credit to touches nearer the conversion. A middle position between linear and last-click.
Weighted towards the first and last touches, with the middle sharing the remainder.
The umbrella term for any model sharing credit across more than one touch — and the name of the product category built around doing it.
That list is the vendor taxonomy. Buyer language is far more lopsided: last-click appears far more often than any other named model, first-click appears next, and linear, time-decay and position-based are close to absent — between them they are named in a handful of interviews across the whole corpus. If you are comparing models on a vendor page, you are reading a menu that buyers mostly do not order from.
What buyers describe instead is running on last-click without having picked it. We currently are working in very old school last click attribution modeling right now, says one. We are very much last click, says another. A third puts the two halves of this page together in a single sentence: we use GA for last-click attribution, and so that's our only source of truth in terms of our marketing effectiveness.
Which model you use matters less than whether you chose it.
How do buyers rate the tools they use for attribution and measurement?
GA4 sits at the bottom of this mixed measurement set and several specialist tools rate higher — though they are not direct substitutes for it or for each other.
These are not all products in the same software category. They are products buyers put into the same measurement conversation, which is the whole point of this page.
| Product | Rating | What it means here |
|---|---|---|
| Haus | 8.0 | The highest-rated product here, though it is an incrementality tool rather than a conventional attribution platform. |
| Triple Whale | 7.8 | The most-rated purpose-built tool in this set. The platform in the counter-case above, where attribution moved off analytics. |
| Northbeam | 7.7 | Frequently named alongside Triple Whale when buyers describe evaluating options, yet the two almost never appear in the same running stack. Typically an alternative to it rather than a complement. |
| AppsFlyer | 7.5 | A mobile measurement partner. In the table because buyers file it under attribution — evidence for how wide the label runs. |
| Measured | around 7.5 | Too few buyers rated it for a decimal. A mix-modeling vendor, reached for when the question outgrows click data. |
| Adobe Analytics | 7.2 | The enterprise counterpart to GA4, and averages higher than GA4 in this corpus — but buyers still name its attribution models as the part with room to improve. |
| Rockerbox | 6.7 | The lowest-rated dedicated platform here. Buying purpose-built is not on its own a fix. |
| Google Analytics 4 | 6.6 | The most-named tool doing this job, and the lowest-rated in the table — 5.8 among the buyers actually in market for attribution. |
| Recast | not rated | Too few buyers rated it to publish a number. Named in mix-modeling conversations rather than attribution ones. |
Want this read against your own stack?
Get my read →Do you need a dedicated attribution tool?
Answer a narrower question first: what is your analytics platform failing to do, specifically?
The buyers here who move to a dedicated tool name concrete failures rather than general dissatisfaction. Three recur:
Everything lands on the last click, and channels doing real upper-funnel work show almost nothing. One buyer tested the attributed result against a control and found the credited spend wasn't generating extra sales — attributed credit is not the same thing as causal lift.
Meaningful spend goes somewhere the analytics tool cannot observe — retail, connected TV, offline, or any channel without a click to follow.
The gap between first touch and purchase is long enough that sessions, cookies and identifiers expire before the sale lands.
Where none of those problems is present, the case for a dedicated tool is weaker. And where the real problem is that nobody reads the reports or acts on them, our recommendation — not a practice buyers describe — is to fix that before buying anything, because a second dashboard inherits the first one's audience.
It is also worth knowing that attribution is not the only instrument, and for some questions it is the wrong one. Incrementality testing deliberately varies exposure between test and control groups to measure causal lift, and marketing mix modeling estimates channel contribution statistically, including for channels with no click to track. Buyers in these interviews often describe these methods alongside attribution rather than as simple replacements for it. On why confidence in multi-touch attribution fell and what budgets moved to, we have a separate page; the recurring complaints about these tools and what buyers wish they had known have their own too.
Common questions
What is marketing attribution?
Marketing attribution is the practice of assigning credit for a conversion to the marketing touchpoints that preceded it. The working definition is uncontroversial. What makes the term slippery is that it names a job rather than a kind of software: buyers apply the word to their web analytics platform, their ad-management layer, their CRM and their dedicated measurement vendor, sometimes in the same conversation. So the useful question is rarely what attribution means. It is which system in your stack is currently answering it, and how well.
Which tool does marketing attribution?
More often than any other product, Google Analytics. It is the most-named attribution tool in our corpus, which makes attribution unusual: in the neighbouring categories the leading product is one built for the category, but in attribution the leader is a web analytics platform rather than a dedicated attribution product. Ad-management layers, CRMs and mobile measurement partners all take on the job too. The dedicated platforms buyers name most are Triple Whale, Rockerbox and Northbeam.
What are the main attribution models?
Last-click gives all credit to the final touchpoint before conversion, and first-click gives it to the first. Linear splits credit evenly across every touch, time-decay weights the touches nearer the conversion, and position-based weights the first and last most heavily. Multi-touch attribution is the umbrella term for any model that shares credit across more than one touch. Buyers describe last-click as the default they inherit rather than choose, and as the one that flatters whichever channel sits closest to the sale.
Is Google Analytics good enough for attribution?
It depends on how much you are asking it to carry, and the ratings suggest the answer changes when you ask for more. Buyers in market for attribution rate GA4 5.8 out of 10. Everyone else in the corpus rates the same product 6.6. The people leaning on it hardest for this job are the least satisfied with it, which is consistent with a general analytics platform being asked to carry a broader attribution job than some buyers think it handles well. One buyer rated it highly precisely because attribution had been moved off it onto a dedicated platform, leaving analytics to do analytics.
Do you need a dedicated attribution tool?
The question worth answering first is what your analytics platform is failing to do, in specific terms. Buyers who move to a dedicated tool tend to describe a concrete failure rather than a general dissatisfaction: credit collapsing onto the last click, channels that the analytics tool cannot see, or a journey long enough that sessions expire before the purchase happens. Where the honest complaint is that nobody reads the reports, a new platform is unlikely to fix it.
Is attribution the same as marketing measurement?
No. Attribution is one method inside marketing measurement, and it is the one that assigns credit to touchpoints it can observe. Marketing mix modeling estimates channel contribution statistically, including for channels with no click to track, and incrementality testing deliberately varies exposure between test and control groups to measure causal lift. Buyers in these interviews often describe these methods alongside attribution rather than as simple replacements for it, because each answers a different question and each has blind spots the others do not.
What is the difference between attribution and incrementality?
Attribution assigns credit among the marketing touchpoints associated with a conversion. Incrementality asks whether the marketing caused additional conversions that would not otherwise have happened. Conventional attribution is observational; incrementality estimates causal lift through experiments or quasi-experiments. Buyers in these interviews use incrementality to challenge or validate attribution rather than simply replace it.
Get this research made for your stack
Whether you need a dedicated attribution tool depends on how many channels you run and what your analytics platform is already failing to answer. Do a 15-minute interview and get the read for your stack — what peers your size measure with, and what they moved to when the reporting stopped holding up.
Get my personalized read — 15-min interviewNo password · your interview is anonymized before it ever informs a page like this one.