What You Need to Know about Paid Advertising Attribution
Paid advertising is one of the largest costs for high growth companies. Therefore, robust attribution of paid advertising is a must. There are a myriad of attribution solutions including that claimed by ad platforms; however, while these are helpful inputs, the most reliable attribution across all contexts uses source of truth system data coupled with paid advertising test design. Build a solid technological foundation with dashboards and visualizations to help operationalize the attribution and decision making. With these steps taken, paid advertising will be deployed with eyes wide open to deliver the best results.
Background
After working with dozens of talented high growth startup founders and CEOs across industries who used paid advertising to accelerate their business growth, we have helped navigate some of the most important, fundamental questions:
“Is my 5/6/7 figure advertising budget actually working?”
“Is paid advertising driving results I wouldn’t have otherwise gotten aka incremental?”
“Is ad spend deployed to deliver the maximum incremental results?”
Before working with some clients, they turned off all paid ads to see if it’s truly incremental - but the shortcoming is that there are different time horizons to realize results from different media. In almost all cases, building a paid advertising attribution system addresses each of these questions methodically until there is full confidence in the effectiveness of the paid advertising.
The Three Categories of Attribution
Ad Platforms
Ad platforms like Meta and Google have the best dataset from which to provide attribution for their channel. When deployed correctly, it can be a powerful input into attribution; however, ad platforms are incentivized to sell more media; they provide results that are just “good enough” to maintain and grow ad spend. Ad platforms may deploy teams of reps that highlight metrics in their reporting UI and custom reports using data that isn’t publicly available. While the conversation is about business outcomes, their annual performance reviews and compensation are based on ad spend from their book of business (aka client portfolio). So their advice, attribution, and reporting which are provided at no additional cost should be viewed from that perspective.
To get the most out of ad platform attribution, the desired results must be communicated accurately. This typically requires integration with front end website pixel(s) and backend server(s) which have a lot of capabilities required to maximize recognized conversions. Until this is completed, ad platforms (or more specifically the ad servers they configure) cannot even represent its clients in achieving “good enough”.
When dealing with fewer sales of high-price items, companies may need to use proxy conversions for ad server automated optimization to achieve “good enough”, with accompanying manual optimizations based on the true desired result conversion.
When dealing with attribution of multiple ad platforms, conversions inevitably add up to more than 100% of actual conversions according to “source of truth” systems which tie most closely to revenue and profit. Even then, default settings are usually generous with view through conversions i.e. viewing an ad prior to the conversion coupled with a lookback window of 1/7/30/90 days.
Transactions are usually the result of a journey comprised of many influential touchpoints versus the all-or-nothing methodologies such as last touch in these ad platforms; even when trying to use “multitouch”, it’s limited to their platform versus the many sources of those influential touchpoints that go way beyond a single platform.
Not only will attribution add up to more than 100%; but it can often distort the true impact on the source of truth results. We have seen companies where ad platform attribution makes results look fantastic, but don’t see nearly as good a story when looking at the source of truth reporting. Conversely, we have seen instances where nearly pausing entire spend on a single ad platform can have negligible impact on the source of truth reporting.
Ad platform attribution is a powerful directional input into optimization analysis but by no means tells the accurate or complete story.
External Platforms
External platforms like Google Analytics 4 (“GA4”), Triple Whale, and AppsFlyer are better aligned with advertisers but their data access is limited and/or involved. Click tracking is a challenge. Going beyond the impression is a bigger challenge. And then modeling across channels is the biggest challenge. Plus, the tooling required continues to get more expensive.
External platforms typically rely on click tracking which is “free”; however, getting a clean dataset comes at a heavy cost. Standardizing tracking nomenclature and deployment of “listening” technology across digital properties is a significant effort. Training team members to setup URLs that follow the convention and then governing its proper use takes time and even then is rarely adhered to perfectly. And if a company maintains multiple digital properties, then the same process is required for engineers in deploying listeners and data layers in a standardized way.
Even if companies invest to get accurate click tracking, there will be major limitations due to ad blockers, private browsing, and by-design large technology company incompatibility. Ad blocker users continue to grow every quarter with over 400mm globally (Statista). Approximately 5%-15% of internet users browse in private or incognito mode (Statista, DuckDuckGo). And large technology companies design incompatibility into their products such as social media platform browsers which obscure external tracking similar to private browsing mode.
The most powerful advertising may not result in an immediate, attributable click; however, it will serve one of the most important objectives in advertising - creating new demand. Just because it doesn’t generate a click, doesn’t mean it shouldn’t be tracked. However, impression tracking is both challenging and requires a lot of assumptions.
Some platforms will ingest offline media logs to do time-based attribution. This is a very helpful input into attribution; however, it’s built on a myriad of assumptions. Most will model statistically significant differences between usual volumes within a certain time interval like 10 minutes, adjusting for changes in spend within other channels. Similar to the paid search comment above, it can often distort the true impact on the source of truth results. We have seen companies where time-based attribution makes results look fantastic, but don’t see nearly as good a story when looking at the source of truth reporting.
Some advertisers will invest in impression tracking software for digital media but face key challenges. Impression tracking software like like Google Campaign Manager (Formerly DoubleClick Campaign Manager (“DCM”)) are expensive in both cost and effort; furthermore, ad platforms such as social media “walled gardens” guard access to their datasets especially at the individual level; so even with impression tracking in place, it will be blocked when targeting specific types of audiences.
More advanced functionality with multi-touch attribution and marketing mix modeling requires years of stability in marketing channels and even then requires assumptions; the best outcomes are often expensive, accompanied by teams of data scientists to comb through and model the data.
Google Analytics’ Universal Analytics used to be free, and the defacto standard; when they migrated literally almost the entire world’s digital properties to GA4, they’re experiencing backlash as the new platform is less user friendly; more user friendly and/or advanced alternatives have significant associated costs.
External platform attribution is a powerful directional input into optimization analysis with perspective into results across channels but by no means tells the accurate or complete story.
Source of Truth Platforms
No matter which attribution is chosen, the ultimate determining factor in success is profit and revenue which is calculated in backend databases like Shopify and Hubspot.
Attribution that isn’t from a source of truth system can be misleading. If your attribution in Meta and Google looks like return on ad spend (“ROAS”) is excellent but you’re not seeing the corresponding lift in sales revenue, then your attribution is wrong. If your attribution in an external platform like GA4 looks like affiliate and paid search drive a massive % of sales, then you’re not getting sufficient insight into what’s creating demand before it’s influenced and captured closer to the sale.
Some sources of truth like Shopify and Hubspot can be configured as external attribution systems, consuming UTMs and other referral data; however, the most powerful attribution is from old fashioned test design in conjunction with looking deeply at source of truth system data. If you’re spending at steady levels across a variety of channels except for one which is increasing, and your source of truth show sales revenue increasing dramatically, then as long as there wasn’t a major change in price, offer, or seasonality, the new channel is responsible. If spending in all channels is steady, and your source of truth shows sales revenue increasing dramatically, then price, offer, or seasonality is responsible. Analyzing prior period data supports these analyses. Drawing on a history of running marketing channels and this type of analysis helps build intuition to sharpen these analyses.
Consistent analysis of source of truth data in conjunction with disciplined test design gives you the clearest path forward. When performance warrants larger changes, this becomes even more important as the assumptions made will be based on the best data available. Ad platform and external attribution supports source of truth attribution, as a great input into shaping assumptions.
"Organic" Conversions
One of the hardest but most important areas of attribution is organic conversions.
Lumped into organic conversions are usually direct URL entry, search engine optimization (“SEO”), offline channels (TV (linear cable, streaming/OTT), Audio (radio, podcast), out of home (billboard, bus shelter, transit poster), and digital channels with early exposure (influencer, YouTube).
These sources of “organic” can be excellent at creating new demand and accordingly come with a high price tag. Attributing these correctly within the “organic” bucket is high stakes and another reason why source of truth attribution with test design is so helpful to marketers
Other techniques can help with these harder to measure channels. Geographic targeting in channels where that doesn’t command a pricing premium can help reduce the spend exposure to learn as well as help with attribution. This enables analysis of results at this geographic level relative to control markets or the overall average. There is also analysis of branded vs non-branded search characteristics; increases in the latter can indicate SEO success.
Dashboards & Visualization
Once you go through all of the work above to develop your framework for attribution and put it into practice, build a dashboard to systematize its review with key stakeholders. Investing in the business intelligence (“BI”) visualization attribution input data helps accelerate marketing insights and efficiency. However, don’t confuse BI Visualization with attribution; it presents the findings of all your attribution methods in dashboards for consumption but won’t help without a solid attribution framework.
Tech Stack
Attribution relies on a complex technology stack that when configured correctly is very helpful; when it isn’t, there will be lots of problems. A lot of vendors have been grouping together different elements of the tech stack such as consumption of ad platform APIs to surface important data like ad spend in a standardized way to be viewed alongside source of truth data. But remember that attribution goes well beyond any single piece of software, even though the advances in software have been incredible.
Conclusion
To recap, paid advertising is one of the largest costs for growing companies. Robust attribution of paid advertising is therefore a must. The most reliable attribution is based on the source of truth with test design, supported by ad and external platform inputs. Visualize your attribution and ensure the tech stack is solid to operationalize this. Don’t get overwhelmed; be empowered by this framework; with each step forward in attribution, there is usually a corresponding step forward in business outcomes.
