Freckle CEO: Every single online KPI goes away with better (offline) attribution
In a wide-ranging interview, Freckle founder and CEO Neil Sweeney offers his provocative assessment of location data, beacons, programmatic, multi-touch attribution and where it's all going.
It goes without saying that smartphones have radically changed the marketing landscape. One of those changes is the capacity to determine whether and how digital ads are affecting foot traffic into stores. Location data collected from mobile devices also offer a range of other benefits: audience and operational insights, competitive benchmarking and more.
Yet, despite these capabilities, location data and offline attribution are still not widely utilized. I recently talked with Neil Sweeney, founder and CEO at Freckle IOT about the current state of location intelligence, offline attribution and where things are headed in the near future.
Greg Sterling: How widely adopted is offline attribution today?
Neil Sweeney: I think it is maybe 5 percent [of brands and retailers] at most. Those that are adopting offline attribution, if they are, are doing so using a standalone channel (i.e., Google) and not holistically across the board.
GS: What are some of the other current use cases you’re seeing for location data, beyond offline attribution?
NS: Audits are probably the second largest. If you are a retailer and want to know who is in your location — irrespective of media — on Monday afternoon, in a specific store, how do you measure this?
In short, you don’t. Retailers are in hand-to-hand combat right now and are in desperate need of this real-time data so they can make more intelligent decisions.
Another thing I am seeing is location data being used as a baseline in the creation of segments, which is a completely intellectually bankrupt model. Segments are the equivalent of branding, a feel-good tactic with no science behind it. They won’t succeed due to their inability to determine their overall effectiveness in driving a sale (that is what attribution does).
GS: Which marketing channels are currently most aligned with offline attribution?
NS: Digital by far the most: desktop and mobile. This makes sense, as those in this vertical are the most data- and measurement-savvy and have experience working with DMPs and third-party measurement firms, such as viewability vendors, etc.
As you get down the tunnel of traditional media, your implementation tends to fall off. Outdoor is interesting. The industry has been buying a non-real-time, self-reported number for way too long, and this is ripe for disruption.
Search is also starting to take form due to the overall spend levels going into search, but its challenge is the dominance of one firm.
Terrestrial radio and TV are the hardest, but both of these mediums are moving to an OS model where that AM signal in your car will be replaced by an app that will allow for attribution to take form. The same goes for TV.
In short: All channels are moving toward offline attribution but at different speeds. They will all end up in the same spot. When they all embrace attribution down each channel, is when it gets sexy; that is when the data science really needs to kick in. This is the part that I am excited about.
GS: How do you see real-world attribution impacting online KPIs and metrics (e.g., clicks, engagement, etc.)? Do they go away? Do they exist side-by-side?
NS: First of all, I am biased. I think every single one of them goes away. Both online and offline attribution are, in my opinion, the base level metric from which all media will be based in the future. The click is just dumb, viewability has no basis as a proxy for action, and fraud is a concept that can be solved if you got away from clicks and actually focused on attribution.
The industry is perpetuating its own issues by embracing the wrong measurement metrics while spending way too much time on silly concepts that provide no incremental value (e.g., header tag).
As a brand, you care about one thing only: selling product. That product is bought either online or in a store. The market will increasingly move towards a performance model based on attribution. If you can’t get my customer to buy a product, online or offline, then I am not paying for it.
GS: What are the current limitations of offline attribution as you see them?
NS: For sure, it is knowledge as a starting point — people are really just being exposed to it and are now having to navigate new waters. This will rectify itself over time. But what is and what is not possible is what brands are dealing with.
Matching in digital is an issue for sure. When you match cookie to ID, you will lose some fidelity and this is for sure a challenge. It is getting better but isn’t perfect. Scale is a bit of an issue as well, but this also speaks to the knowledge concept. When looking at attribution, you must understand that it is really a combination of scale + fidelity + matching.
GS: Do you see problems with focusing on a single channel (re attribution)? Does that obscure some larger “reality” of the consumer journey?
NS: Yes. It is not fair to say it is a “problem,” rather I would say it is a starting point. It is not fair to suggest that measuring just mobile impressions represents a proxy for attribution, especially when this only represents <10 percent of your total impressions; but it is definitely better than zero percent.
I encourage brands and agencies to start in whatever channel is the easiest and then integrate horizontally. Every additional input and channel you add makes your model better. In time you will eventually get all inputs and all channels.
Congratulations, you now are in multi-touch and this is 100 percent where the entire industry is going. The agencies and brands that embrace this will be so far ahead of everyone else, it will be ridiculous.
GS: Do you foresee a time when offline attribution simply becomes part of a larger multi-touch attribution scenario? If so, what will it take to get there and over what time frame?
NS: Yes, offline attribution is a precursor and a necessary step for MTA.
It will take 16 months. In 2018, every major platform will have a “solution,” the majority of which will not be home-baked. The DMPs will move into this space, as it is a massive gap in their toolkit. It will be used as a weapon against the walled gardens, which will try to sell their own solutions.
The comparable would be viewability divided by 2 (Moore’s law). If it took viewability five years to roll out, it will take half the time for attribution. You’re are not learning a new “habit,” rather a new methodology.
GS: How do machine learning and AI play into offline attribution in the near term?
NS: If you are not doing this now, you are dead. Anyone whose business is location is the equal to someone who is saying their business is flour. Flour you can get in any grocery store; it’s a necessary ingredient to make something else. ML/AI use “flour” to make their cakes, cookies and tarts; without it you just have flour. Good luck with that.
GS: Location data in the programmatic universe is often criticized as being sloppy, fraudulent or inaccurate. Do you see that changing? What will change it?
NS: In the short term, it won’t really change. Buying location data in a DSP today is equivalent to driving a convertible down Las Vegas Boulevard with $150K in loose cash in the back seat: [throwing] away your money. What is going to change is a move away from the banner, not the move to better location. It’s way too hard.
The way apps surface location is so different by app, it would be impossible to police perfectly. This isn’t to suggest it is fraud (some is, but not all of it) but programmatic has bigger issues around outright fraud and domain spoofing; better location is way down the list of things to fix.
GS: What do you see as the future for beacons and other in-store positioning and location technologies?
NS: I think I was doing beacons really before anyone else (hence the SDK at Freckle). There is some value there, but there are also some limitations — namely the reliance on an SDK.
The SDK is not a sustainable technology moving forward. So, as a consequence, that will have an impact on beacons. In-store positioning is tricky, as it is siloed. Retailers want to know what is going on in their stores, but they also want to know what is going on in competitors’ stores. You can’t solve that with beacons.
Where do I think it is going? Various other systems including your car, smart home devices, etc., will eventually connect to the CPU in your pocket (and your identity) to power everything around you — including what is happening in store.