Attribution models.<\/p>\n\n\n\n
If you\u2019re in B2B, you\u2019ve heard of them, used them, maybe even fought over them in a boardroom. You\u2019re probably even tired of talking about them. <\/p>\n\n\n\n
But here\u2019s the thing: attribution models aren\u2019t inherently the problem. It\u2019s how we use them that leads us astray. <\/p>\n\n\n\n
Ask anyone who works in CPG if, especially those with a marketing mix that includes dozens of both online and offline channels, if they use attribution (or versions like MMM<\/a> or incrementality<\/a> – which as an aside, the author has to admit his preference for incrementality in many cases. Very underrated). It\u2019s not even a question – of course they do! <\/p>\n\n\n\n When you\u2019re in charge of $100M+ in spend, you want to know, at least to some degree of accuracy, that you\u2019re spending it wisely. <\/p>\n\n\n\n B2B brands operate in a different landscape however, especially those enterprise software brands who have a complex suite of products and a long sales cycle with many stakeholders in the buying committee. Channel concentrate is typically higher, with less overall digital spend, and many touchpoints being somewhat opaque (…dark social?). <\/p>\n\n\n\n So, we end up applying, naively, similarly complex mu<\/a>lti-touch attribution (MTA) models<\/a> to enterprises with a narrow TAM and opaque customer journey. <\/p>\n\n\n\n Let me explain with an analogy from Nassim Taleb:<\/p>\n\n\n\n \u201cAccording to Wittgenstein\u2019s ruler: Unless you have confidence in the ruler\u2019s reliability, if you use a ruler to measure a table you may also be using the table to measure the ruler. The less you trust the ruler\u2019s reliability, the more information you are getting about the ruler and the less about the table.<\/em><\/p>\n\n\n\n Attribution models are our rulers. They often tell us more about the model than the reality of our marketing, measuring our own biases, which go on to fuel our decision-making (which is why you\u2019re probably overspending on Google and Meta). <\/p>\n\n\n\n Worse, we then use these flawed insights to make business decisions, shuffle credit, and justify budgets, ignoring the bigger picture: is the business actually growing?<\/p>\n\n\n\n I\u2019ve sat in too many meetings where leaders argue over attribution. One team wants to switch from W-shaped to linear. Many haven\u2019t even figured out how organic search reasonably fits into their go-to-market, so they\u2019ll pick a simple model like last non-direct click or first click and run with it. <\/p>\n\n\n\n When the model makes the results look bad, they switch the model. And everyone believes this change will magically fix the problem.<\/p>\n\n\n\n Spoiler: it won\u2019t.<\/p>\n\n\n\n Attribution models often become credit assignment systems\u2014political tools to justify departmental budgets. SDRs argue they sourced the lead. Marketing claims credit for nurturing it. Sales insists they closed it. And instead of asking why the business isn\u2019t hitting its targets, we waste time debating who deserves the gold star (and more headcount, which is the currency of power in large organizations).<\/p>\n\n\n\n Meanwhile, win rates plummet. Pipelines dry up. And we\u2019re left wondering why shifting attribution shapes didn\u2019t save the day.<\/p>\n\n\n\n Here\u2019s the truth: the problem isn\u2019t the model; it\u2019s how we\u2019re using it.<\/p>\n\n\n\n In my recent podcast episode with Chris Walker (out this week, find it here<\/a>), he clarified his position on attribution models, which is remarkably similar to how I view them. He proposed a thought experiment. Imagine your win rate drops from 16% to 6% over 12 quarters. Is the issue:<\/p>\n\n\n\n The real question isn\u2019t \u201cwhat shape is our attribution model?\u201d It\u2019s \u201chow do we fix our pipeline?\u201d<\/p>\n\n\n\n Attribution should answer that question\u2014not distract from it. It\u2019s not about assigning credit; it\u2019s about uncovering insights to solve specific business problems.<\/p>\n\n\n\n When we narrow in on the domain of SEO and its contribution to pipeline and revenue, the problem becomes even clearer. <\/p>\n\n\n\n Here\u2019s an example:<\/p>\n\n\n\n For enterprise software co<\/a>m<\/a>panies<\/a>, focusing solely on last-touch attribution is a recipe for disaster. Why? Because not all leads are equal. One might close a $50k deal that falls through; another, a $1M contract. A nuanced model that prioritizes high-value pipeline insights can shape a more effective strategy than generic website-based lead attribution ever will.<\/p>\n\n\n\n We\u2019ve written in the past about how we often default to a last click attribution model<\/a>.<\/p>\n\n\n\n That\u2019s because many brands we work with are product-led growth companies who, whether you choose a first, last, or linear model, are looking to drive a low friction free user into the top of the product funnel. There\u2019s not a lot of room, speaking of unit economics and channel CAC, for a whole lot of nurturing and sales-led conversations to get them into this stage. Thus, we chose the most difficult one to force us into driving more business value.<\/p>\n\n\n\n However, we\u2019ve since begun working with very complex enterprises (some of the logos are on our website, many are not due to NDAs). We\u2019re not using last click models with them. We\u2019re digging into HubSpot and Bizible to reverse engineer the touchpoints and pages that drove the most revenue and highest propensity pipeline. <\/p>\n\n\n\n Doesn\u2019t just apply to large enterprises, but the startups that sell to them as well. <\/p>\n\n\n\n For example, I\u2019m working with an early stage data quality startup with a narrow TAM and a very specific salve for their data operations problems. To fulfill a successful organic program, we drill down to the center of the bullseye and write about pain points of data leaders that relate to the product. <\/p>\n\n\n\n However, there are only like 10 of those exact match phrases. To get those to rank, as well as to drive some relevant demand on related topics, and then push them further down the funnel, educating them on data contracts, we need to expand our sphere of topics. Sometimes, this means we\u2019ll hit audiences that don\u2019t actually have the propensity to buy.<\/p>\n\n\n\n But we don\u2019t know what we don\u2019t know until we know it.<\/p>\n\n\n\n So we analyze, as leading indicators, relevant organic traffic and website conversions. But not all conversions are created equal. We\u2019re discovering, to our surprise, that certain topics we thought weren\u2019t as relevant (i.e. they target a slightly more junior ICP) have a higher hit rate when it comes to generating qualified pipeline. <\/p>\n\n\n\n Thus, our \u201cattribution model,\u201d while still at the GA4 layer, measures proximate first click touch points, always goes deeper when it comes to making decisions about where to alter our organic strategy, as we base that on the deeper funnel insights we get from the CRM. <\/p>\n\n\n\n Attribution should be a means to an end\u2014a tool to guide decisions, not a scoreboard for internal politics.<\/p>\n\n\n\n B2B organizations are silo factories, despite one million keynote presentations from 1994 until today on how to align sales and marketing. Sales. Marketing. Product. Each subset within those teams. Even the hybrid teams, like \u201cGrowth.\u201d <\/p>\n\n\n\n Each has its own metrics, its own priorities, and its own version of success.<\/p>\n\n\n\n Attribution models often reinforce these silos, focusing investments on narrow slices of the buyer\u2019s journey. For example, marketing might pour money into sourcing leads that are already in-market because that\u2019s where attribution says the ROI is.<\/p>\n\n\n\n But what about the 99% of your TAM that\u2019s not actively buying? Who\u2019s warming them up? Who\u2019s building trust and awareness for the long game?<\/p>\n\n\n\n We take our own medicine over here at Omniscient. Why do you think we\u2019re investing in a physical magazine? In our podcast? In IRL events and all the other branded stuff we do? <\/p>\n\n\n\n It\u2019s because we know that it\u2019s not very often that people are in the market for an organic growth agency, but when they are, we\u2019ll be top of mind and possibly ahead of the pack because we\u2019ve already proven trustworthy and hopefully somewhat smart. <\/p>\n\n\n\n But we\u2019re also measuring the f*ck out of our core lead generation channels, including SEO and LLMs (which we invest in significantly), LinkedIn, referral partnerships, this newsletter, paid ads, and outbound. <\/p>\n\n\n\n It\u2019s a good old fashioned barbell,<\/a> some of which we apply to our (last click) attribution model, and the rest we look at as an investment in the 99% of our TAM who will eventually hire us (but not today). <\/p>\n\n\n\n When attribution drives investments into the wrong areas, it\u2019s like obsessing over the rudder while ignoring the iceberg dead ahead.<\/p>\n\n\n\n It\u2019s time to flip the script. Instead of using attribution to argue over credit, we should:<\/p>\n\n\n\n In short, attribution isn\u2019t about the past (who did what, clap clap, good job). It\u2019s about the future\u2014how to make better decisions to drive unified outcomes.<\/p>\n\n\n\n Actually, let me rephrase the summary of this essay: stop midwitting this stuff. <\/p>\n\n\n\n Ask yourself, \u201cshould we measure some components of our marketing investments?\u201d Yes, probably. \u201cAll of them?\u201d Probably not, at least not under the same measurement model. <\/p>\n\n\n\n \u201cHow should we think about this channel\u2019s contribution to our overall north star metric or incremental revenue?\u201d Well, it feasibly would contribute by\u2026.. <\/p>\n\n\n\n Okay, now figure out a model by which you can analyze whether or not that is happening and use it to make better decisions and investment allocations, and then just start shipping good work and talking to your customers more. <\/p>\n\n\n\n Aye aye aye!<\/p>\n\n\n\n <\/p>\n\n\n\n Want more insights like this? Subscribe to our <\/a>Field <\/a>Notes<\/a>.<\/p>\n","protected":false},"excerpt":{"rendered":" Attribution models. If you\u2019re in B2B, you\u2019ve heard of them,…<\/p>\n","protected":false},"author":3,"featured_media":6660,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"content-type":"","footnotes":""},"categories":[184,25],"tags":[],"wf_post_folders":[],"class_list":{"0":"post-5347","1":"post","2":"type-post","3":"status-publish","4":"format-standard","5":"has-post-thumbnail","7":"category-field-notes","8":"category-seo"},"yoast_head":"\nThe Blame Game: Why Attribution Feels Broken<\/strong><\/h2>\n\n\n\n
Using Attribution to Solve Business Problems (Not Create Them)<\/strong><\/h2>\n\n\n\n
\n
The Bigger Picture: Aligning Teams Around Shared Goals<\/strong><\/h2>\n\n\n\n
The Path Forward: Rethinking Attribution<\/strong><\/h2>\n\n\n\n
\n