Long sales cycles, the #1 enemy of SEO attribution
Attribution models are the most common way to quantify the returns on marketing investments, but not every company can use them successfully. Long sales cycles (+90 days) are the biggest enemy of attribution models, often to the detriment of channels like SEO that play an important role in early touches.
Companies with long sales cycles often either sell B2B enterprise software (SaaS), high-priced products (e-commerce) or long-term commitments (consumer). The function of Marketing is to drive customers (if they can complete the buyer journey without talking to a person) or leads who connect with a sales rep. The core challenge for long sales cycle companies is giving channels, and touch points the right amount of credit to optimize them.
In 2023, companies must shift from growth to efficiency and can’t throw dollars against the wall to see which marketing spaghetti sticks. Marketing budgets and teams are getting smaller and need to measure every dollar twice.
The problem with faulty attribution is a wrong sense of reality, which can lead to an underinvestment in critical Growth channels. In highly competitive markets, starving important channels opens a window for competitors and can be the difference between winning and losing.
How should companies with long sales cycles approach SEO attribution?
I offer 4 workarounds for the long sales-cycle problem of attribution:
1/ Find the 2nd best model through testing. If you can't see reality, look for a proxy. Time decay models over-weight channels that assisted the conversion toward the end of the buyer journey, but u-shaped, w-shaped and custom attribution models can give sufficient credit to the first touch they measured. Compare them with each other and shift your spend for a short duration of time (1-3 months) to the model showing better reutrns to see if you net out more conversions.
2/ For SEO, test a budget increase for 6 months. Commit to a grace period no matter what happens. 6 months is a good time frame because it allows you to hit at least one Core Algorithm Update and see if it gives you a boost based on the work you did with more budget. If you're not seeing results after 6 months, you either don't have the right people, or SEO is not a growth channel for you, which is not unrealistic for a company with long sales cycles.
3/ Give self-reported attribution more weight in your decisions. At the end of the day, attribution models should lead us to better decisions. If customers say they found a product through SEO, but attribution doesn't reflect that, it's more likely something is wrong with the data than the customers. Self-reported attribution won't help you quantify and project returns but you'll know when you're on the right track.
4/ Use a customer journey tracking tool like Knotch or Amplitude that doesn't have a limited attribution window. For organic channels, it's worth questioning why there should be an attribution window at all. If a customer takes two years to convert, why not try to track their touch points?
The limitations or attribution models
Let's take a step back to understand why these four workarounds are necessary.
Long sales cycles take 180-360 days, during which potential customers have many touch points - sometimes over 30 - with the website and brand. But Google Analytics is capped at 90 days, and there is no way to extend the lookback window.
CPA (cost per acquisition) is easy to see in Google Analytics, but for organic traffic and other organic channels, it's zero because there is no direct spend. But, no channel is really free except for Word of Mouth. SEO costs time, people and assets (content, links, etc). Since cost is not directly visible for SEO, many leaders shy away from it.
Attribution is not a mathematical formula you solve once. It's the foundation for hypothesis testing and iterating based on feedback signals. However, since SEO has such a long feedback loop, results take more time than many are willing to wait for.
Companies (with long sales cycles) have 4 categories of marketing attribution models to compare:
- Multi-touch (MTA)
- Linear touch (LTA)
- Data-driven (DDA)
- No attribution
A growing number of companies have given up on the idea of measuring attribution. I can’t blame them. Measuring touch points across devices, offline vs. online, and interactions (impressions vs. clicks) is close to impossible. So why bother? You might as well operate on good judgment, but few leaders have the trained gut it takes to succeed. They would make better decisions based on data because, without good judgment, you make decisions based on conviction, which is sensitive to bias. Maybe they heard something a consultant whispered in their ear (👋) or a headline they saw in the news.
Out of the three options, most tools default to linear touch (LTA). It’s stunning that Google Analytics never made it beyond default last click in almost 20 years of existence! Maybe the reason is its owner makes so much money with the channel most suitable to last click: ads. To be fair, most marketers are nifty enough to know linear touch can be quite misleading.
DDA and custom attribution models are the best options so far, and Google pushes marketers aggressively toward them, but not everything about that move smells like roses.
Google's shift to data-driven attribution might make the problem worse
Google announced to get rid of four attribution models in place of data-driven attribution [link]:
- first click
- time decay
Based on Tweets by Ginny Marvin, Google's Ads liaison, fewer than 3% use rules-based conversion actions. But the change could make SEO attribution even harder for companies with long sales cycles:
First, why isn't last click isn't on the list of models to remove? LTA favors paid channels tremendously and devalues SEO the most.
Second, DDA seems to make many good decisions and is able to understand data much better than any human could. But should we really trust attribution models we can’t understand?
Third, one of the benefits of attribution models is they integrate easily with ad platforms like Google Ads. But Google Ads already forces users to pick ML models for keyword bidding. Fewer insights from attribution models on top take away critical transparency. Marketers won't understand when their Dollars are wasted.
The time is right for more transparent alternatives to attribution.