By: Pavel Sima, Roivenue CEO
For long I have struggled to conceptually grasp different attribution models via mathematical equations. So, I devised simple, real-life metaphors for them. This way it shouldn’t take you more than 3 minutes to finally understand all the most frequently used attribution models in digital marketing and recall them with ease.
1) HEURESTIC (=TOTALLY USELESS) ATTRIBUTION MODELS
A) MODELS WITH ONE SOURCE OF ATTRIBUTION
All the glory goes to the striker – even if he is totally lame, everything was fought out by the rest of the team and he merely put his foot in the path of the shot.
Who is to blame for all the bad deeds in my life? My mother of course. For she stood at the dawn of my existence.
These above models are as far from reality as the burger you get at the counter is from the one you see on the billboard.
B) MODELS WITH MULTIPLE SOURCES OF ATTRIBUTION
"The most important thing is not to win but to take part!" This supposes that in a tug of war, an anorectic celeb who was put on your team to attract more viewers makes the same contribution as a 250-pound wrestler.
Sure, I am grateful to all the Chinese manufacturers who put the pieces together, the freighters who brought it across the ocean, and the retailers who sold it to me. However, most of the margin will go to Apple as they put their logo on the box.
With these models, at least all the credit doesn’t go to one person. The level of inaccuracy is still humungous, though.
Do you really think a human brain can actually decide which touchpoints should be assigned which weight? Didn’t think so.
2) DATA-DRIVEN (ALGORITHMICAL) ATTRIBUTION MODELS
They try to calculate the probability a certain channel contributed to a conversion.
SHAPLEY (ALSO KNOWN AS GAME THEORY)
We all have such a friend. When he shows up at a party it’s always a blast. Then one day, he doesn’t come and you are bored to death.
A hunch tells you that he is a really important part of a good party. But how can you be sure?
Imagine, you went to 10,000 parties a year instead of ten, both with and without your mate. By the end of the evening you would rate all of them from 1–10 on a scale of awesomeness.
Then you would easily be able to calculate the probability that he’s the most important ingredient of a good party.
The same way, you could evaluate any guest (= marketing channel) at every party (= conversion path).
Shapley’s big problem is that it doesn’t take the order of guests/channels into account. Somebody is a great party starter who nevertheless falls into a coma at 9 PM. Standing on your feet at 4AM is an entirely different discipline. You need a great mixture of all these types of people to throw a legendary party.
MARKOV (ALSO KNOWN AS MARKOV CHAINS)
How does gossip get spread in an organization? Sometimes you get them directly from the source, sometimes they travel through six nodes of people, while other times they get stuck in a loop and never reach your ears.
Michael knows about the latest scandal first and tells Monica about it over lunch. In the bathroom, Monica passes it on to Jane. Back in the office, Jane tells Michael about it again. After that, Michael goes directly to you.
If you omit Michael you are pretty much screwed and will learn nothing. On the other hand, Monica and Jane are not all that important in the chain as Michael would probably have told you anyway. That’s why Michael is the most important informant.
How do you determine who’s the most valued informant (= marketing channel) in the long run? Try to send different employees for a long vacation and observe if the level of information passed on to you dropped off or not. If it did - the guy on vacation contributes to conversions a lot.
MARKOV CHAIN 1ST ORDER
It measures the influence of the movement of the information (= nudging people closer to conversion) to the very next person (= channel).
MARKOV CHAIN 2ND ORDER
It measures the influence of the movement of the information to the next two people because if, for instance, you often talked to Jamie who is mute and couldn’t pass the gossip on to anyone else, the first-order Markov would blame you. Not fair!
WHICH ATTRIBUTION MODEL IS THE BEST?
Our opinion at Roivenue is that it is the Markov model. You don’t necessarily have to trust us, though. Just looking at what different models say about your data is often an eye-opening experience. In our experience, with some channels you can find up to a 300% (!) difference in how the different models value their contribution.
Choose for yourself which attribution model you want to trust the most. Roivenue will allow you to compare all of them. Once you decide which is the best, you can switch the whole application and order all metrics to be recalculated with one click of a button.
2017 was a very fortunate year for us here @ Roivenue.
We are now very confident we are onto something and so we've decided to double down on our attribution product.
This January, Roivenue will break out on its own. As an independent company, Roivenue will be taking most of the staff of our parent company, Cross Masters, over to a new legal entity called Roivenue.
There are lots of exciting things ahead this year, including offices opening in three new countries as we can't wait to get our amazing product into the hands of as many clients as possible!
If you’d like to be part of a growing wave of smarter optimization that’s delivering vastly improved marketing ROI, don't hesitate a second longer. See our job offerings today.
Roivenue in 2015 & Roivenue in 2017
By: Pavel Sima, Roivenue CEO
Recently, I ran the numbers and found out that, on average, clients who are with us for at least a year see a 30% increase in their revenue, while achieving a 15% increase in ROI. It means they make more and spend less.
Our goal with a current round of financing is to reach 200 clients in three years time.
Put in real terms, the increase in ROI based on the 30% increase in annual revenue would mean our clients would have an average yearly revenue of 100,000,000 EUR - with 30,000,000 EUR of newly-generated revenue from Roivenue.
And when I say Roivenue, it means nothing else than our loving, smart, hard-working, talented, and dedicated colleagues who believe there's a better way of doing digital marketing, and have the guts to put several years of their careers in to prove to the world that it really is the case.
I once read that with a young company in your hands you stand a chance of creating a little universe where everything can be just right. In my world, I believe that it is only fair that the same people I meet in the office long after the official office hours should reap the benefits if and when the company succeeds
If, a couple of years down the road, we turn out to be a major success, I do want everybody to be able to maybe buy an apartment. And, if we turn out to be only a moderate success, then I would love everybody to have the liberty to maybe take a short career break and travel around the world - or sit and read books for 6 months - without the need to go to work.
Life is short and those who work hard, should be allowed to play hard as a reward.
That's why we decided to distribute 10% of the company shares to our key personnel as well as to every single full time employee.
If you share the same values as we do, come join us. There's a lot of chairs still untaken.
By: Pavel Sima, Roivenue CEO
At its Google Marketing Next event in mid May, Google announced that it is rolling out a free, still-in-beta version of Google Attribution in its Analytics platform. Finally, even mid-size businesses who cannot afford the Google Analytics 360 suite will have access to data-driven attribution models so that they can better evaluate the performance of their marketing mix.
This is great news for companies that do most of their digital advertising inside the Google ecosystem. But, the plan has cracks for marketers with more complex portfolios.
1. It's never going to be independent
Immediately after the announcement, big agencies, clients, and bloggers alike raised the question of independence. Some even spoke about having no “separation of Church and State.“ Having Google evaluate the performance of campaigns run on their own properties (namely AdWords and DoubleClick) has always been problematic. Asking Google the question of how their properties perform relative to other marketing platforms is now even more unthinkable to pundits and marketers alike.
2. There's no such thing as free attribution
The “Free data-driven attribution for everybody” announcement came with a big caveat: Google Attribution will also have a 360 (see: paid) version. While we still don't know which features will be included in the free version, it’s safe to assume that there will be significant hurdles - similar to long-standing sampling issues in the free version of Google Analytics.
The more fundamental question is just why would Google allocate so much computational power that goes into data-driven algorithms and give it away for free. Well, advertisers are not going to pay with dollars but something much more valuable – their advertising data that will train Google’s algorithms to gain an advantage over other ad networks. And, of course, free Analytics users voluntarily stepping into a sales funnel for Google Attribution 360.
3. Want to know marketing ROI? You need your costs, not just your visits
Measuring the origins of sales to the last cent is great. But marketers ultimately need to know the ROI of what they are doing online. While Google may have information about how much a particular campaign or click cost in AdWords or DoubleClick, and where clicks come from on other ad and social channels, it cannot tell users about the impact of their spend from any non-Google platform. Facebook and other players are not going to let Google scrape that valuable business information.
According to eMarketer, Google accounts for 40.7% of US digital ad revenue and has a 12.5% share of the display market. That’s obviously significant, but it also means that 60% of ad revenue, and 88% of the display market, is outside the Google ecosystem. In Europe, the numbers are even lower for Google.
To marketers who advertise on other display networks, Facebook, affiliate networks, re-marketing, product comparison sites, aggregators or buy ads in private deals (and we've seen clients with as rich portfolios as 15 different platforms where they spend money), Google Attribution is never going to tell them ROI. This release may be of interest to analyst teams, but CMOs will still need to measure cost/benefit impacts from other platforms in another platform.
4. Ultimately, it's about profits and customer lifetime value (CLV)
If you want to combine the perfect attribution models with margins, profits, or CLV, you are still left with two options: develop a data-driven attribution tool in-house (which is costly and consumes development resources to both develop and maintain it) or enrich a third party solution with your customer data.
Paranoid or not, most companies are already feeling anxious about the breadth of data Google gathers about their business and consequently are unwilling to trust it with even more of their customer and business data. Therefore they have to seek a third party solution that offers data-driven attribution and customer data imports while at the same time being independent from ad networks.
The fact is that data-driven attribution is a piece of a much larger puzzle that marketing departments need to solve.
So what are marketers left with?
If you can get past the issues around privacy of your data, and do all your advertising inside Google ecosystem already, free Google Attribution may work for you. And, it's actually great that there is finally going to be an affordable solution that will get millions of companies out of the misery of being locked in the last-click cage.
The rest of the marketing world (see: most of the marketing world) can derive greater benefits from an all-encompassing third-party solution like Roivenue.
From the very beginning, Roivenue's mission has been to overcome all of the issues mentioned above, and to give our clients the most precise, unbiased, clear view of their entire marketing ecosystem and the ROI associated with it.
No matter how complicated your marketing ecosystem is, Roivenue gets data from every advertising platform (even if it has no API – yes, we do magic for you), takes into account all touchpoints, never samples, connects to any CRM or ERP you use, has trained account managers to guide you, is here to stay and won't ever let you down.
By: Martin Faith
I have been exploring the possibilities of Power BI over the past year and, as I look back, I’m quite happy with that education. Over my professional career to date, my reporting has been done almost exclusively in Excel. Although I progressed as an Excel user and was becoming more effective with it, I couldn’t help but think that things could be easier.
Even if you're an Excel pro and can manage to query data into your worksheet with Power Query, the final output of your work is still a clunky chart or a pivot table. Where is the tool that produces sleek, intuitive and appealing reports the user will look forward to? A report that someone will look forward to? To me, at that time (and probably to you at the moment), this seemed like a far off utopia.
Fast forward to present day… Now, I work in Power BI on a daily basis, and teach a Power BI basics class. I see clients looking forward to the reports we deliver for them. Reporting is mostly synonymous with boring, not anymore. Here are the few ways Power BI makes your data analysis and reporting life better:
One Report to Rule Them All
Prepare each report exactly ONCE. One of the (many) reasons why people get frustrated with Excel is repetition of reporting – the same reports…over and over and over again. With Power BI, those days are done.
Prepare a report, set up your query infrastructure, check that your data source is working properly, and you’re all set. All you need to do is sit back and watch live data feed into your dashboards. You may need to ask your boss what to do with all the time you have saved yourself, your colleagues, and the company as a whole.
Cross Source Analysis
Some of your data is in your CRM, some is in a Google Sheet, there’s a bit downloaded from a web application. With Power BI, you can analyze all the data in one place, and possibly even in one visual.
You might need to work a bit on setting up a relationship between the data sources, but I’ve already found a few tricks to make this work.
Create a DISTINCT value list which establishes a connection between the two tables you are looking to connect, play around a bit with the data model, and off you go. Or even better, if you are looking to analyze your data over time, create a date table with CALENDARAUTO function, connect the dates through this table, and you can compare even apples and oranges over time.
Beautiful, Branded Reports
No more quirky filters. No more cross-tab references. A properly-prepared data model in Power BI lets you create reports that give stakeholders both a high-level perspective and the ability to zoom in and focus on the details that make a difference.
In fact, I’m using this right now. We’ve been able to reduce reporting for our agency clients from dozens of Excel reports prepared manually every week, to a single Power BI report page, which is
prepared once, branded, and easy to work with.
Everyone on the Same Page
Report sharing has never been easier.
We made this one in just a few minutes with a demo account. Expand it to full screen and click around!
No More Data Surprises
The feature I am starting to use more and more often is report notifications in Power BI Service. After completion of a report and its upload to Service, Power BI enables you to define a value-based notification. If I’m looking to hit a sales goal, reach a conversion KPI, or waiting for a price to sell, Power BI will let me know when the time comes.
In Roivenue, we use this feature for data load checks. If the data does not load properly, a notification monitoring its completion will pop up in my email and I can resolve the issue before anyone sees anything may be amiss. No more manually clicking through our accounts, checking them again and again to make sure our loads are complete. So, thanks Power BI for saving me hours of time each week!
Make your life easier and become friends with Power BI. For the price of a six-pack, you can have monthly access to all of the described goodies and many many more. Wide-ranging possibilities of data sources enable you to produce meaningful analysis regardless of what you do – if you are a marketer, technical analyst, or simply a curious blogger. Just try plugging in any Wikipedia page into Power BI Desktop and you will see what I mean.
It might seem like I'm selling Power BI a bit, but let me assure you that I am not affiliated with Microsoft in any way. I just want to help people use their data more effectively.
While there is no shortage of data flowing into your business, getting the most out it can be challenging. Reliance on free or native tools gives marketers a view – but it’s often not enough. Here, Pavel Šima explains why Google Analytics is good, but comes up short as a stand-alone solution for most marketing pros.
By: Pavel Sima, Roivenue CEO
1. Naive Marketing Attribution
Google Analytics offers a range of statistics and capabilities, but only a very basic model of attribution. The most-used is “last click,” which awards all the credit for the order to the last source the customer clicked on before ordering.
Aside from last click, in Google Analytics you can find first click (all the praise goes to the first click), linear (every channel involved in the conversion path gets the same equal credit), and time decay (biggest award goes to the channels near the end of conversion path) metrics.
There are numerous issues at play, but the most-important (for now) with these attribution models is that they arbitrarily decide in advance which click has which importance (weight) for you. Attribution modeling is complicated. Because every business has different marketing mix, products, and competition, it can be difficult to get everything into one view – a view that lets marketers focis on making decisions rather than aggregating diffuse streams of data.
Another issue with Google Analytics is that it cannot evaluate the influence of the ads that were only seen. Advertisers who want to include the banners which were only seen – no clicks or other interactions) to the attribution paths are left wanting.
Consider transitioning to data-driven attribution modeling - statistically based operations that compute the real benefit of a given marketing channel relating all conversions and all of the possible combinations of conversion paths. Data-driven attribution models are offered by several premium solutions for web analytics, such as Google Analytics Premium or Adobe Analytics Premium. Their issue is that they are so-called “black-box” applications, meaning front-end users do not know the calculations that lead to the results.
Roivenue does not impose an attribution model you should choose. Instead, it shows you three typically-used models (our math is public!), lets you compare them, and helps you choose the most-relaible one. Roivenue can compute attribution based on only-seen banners, for example, for a fraction of the price of big suppliers of attribution solutions.
In this case, the solution is simple - Roivenue never works with samples. Data gets transferred to another server and analysed – all of the data. Not 2%, not 10%...all of it.
Because Roivenue is feed-connected to ordering system, discrepancies are nearly impossible. Roivenue knows, thanks to the ordering system feed, about 100% of transactions. Even if you miss 5% of transactions in Google Analytics (a common problem) you can see all of them in Roivenue.
Most businesses cannot see margins in Google Analytics and therefore have no way of optimizing strategies toward profits. This is an issue especially for businesses with several products with large differences in margins (one product has margin of 5% percent, while another might have 80%, for example). Optimizing investment without the right (or incomplete) data can be a gamble which can cause a bigger part of the campaign to be non-profitable and even loss-generating.
Missing margins are dealt with in different ways:
• Calculate mean margin from all the revenue, for example for the past year
• By arbitrarily stating a PCO (Percentage of Cost per Order)
• By dividing groups of products into margin categories and tracking the mean margin
• By not dealing with it at the marketing level at all!
There are a multitude of issues here, but the central result of all of these cases is that marketers do not know if they are creating profit or loss.
In order to be able to work with margins, you need to connect every ordered product to its acquisition costs plus delivery cost, payments, and discounts). Roivenue gathers this data from your ordering system and compares the marketing costs of the orders in relation to the gross profit they bring.
4. Returned Goods
Awesome. You have succeeded in buying cheap traffic which converts fantastically. Hold your horses!
If you look only in Google Analytics, you can see web conversions but miss any data telling you if the customer has paid for and accepted the goods. What if the goods from this cheap traffic have 3x higher returns than usual? In other words, what if three times more customers do not accept the parcel and do not pay for it?
Logistics cannot advise you because they cannot connect the returned orders with the campaigns. Google Analytics may erroneously make you think that you have found a gold mine but all you do is put more budget in to losing campaigns.
Google Analytics ends with a conversion (or order) and does not track product delivery and money transferred.
As in the previous example, the solution is to connect web and marketing analytics with your ordering system and to get that data talking and working together.
That is exactly the reason for an extra step in Roivenue between conversion and revenue. Conversion —> Delivered conversions —> Revenue.
Summing it Up
Google Analytics is great tool for beginners and for evaluating events on web pages. If your business grows quickly, however, you also encounter its limits:
The solution can consist of custom integrations and buying premium solutions for web analytics. Roivenue is an alternative tool which can solve all of the integration processes for you, with almost zero demands on your IT and for a fraction of the price of global solutions.
By: Martin Faith
Roivenue uses collaborative game theory to help marketers invest correctly. You don’t need to be a mathematician to use it, but a general understanding of the main principles certainly doesn’t hurt.
The traditional ways of looking at attribution are first or last touch. But honestly, these do not describe reality. The customer journey is more complex than ever, and the idea that the first or last ad someone saw online deserves full credit for a purchase was never really valid.
There are also linear and time-decay models, which are perhaps a bit more fair, but, again, is this how things tend to work? We don’t think so. Football (soccer) provides a useful example.
Your favorite football team just won a tournament and got 1 million euros to divide amongst them. Who should get this money?
The various models of attribution would treat the reward like this:
In our opinion, none of these models are correct. Everybody contributed to the win in some way, and it’s no easy task to assess who played the biggest role in the win. The same is true for marketing.
The reason why we simply cannot agree with these models is that they distribute the success across respective touchpoints based on a human decision. What is this decision based on? It’s impossible to tell. This is why we use the models we do – we think they are correct. Rather than valuing performance based on feelings or trying to hit the jackpot with our eyes closed, we turn to science and data.
Typical marketing and ad spends include Facebook ads, AdWords, paid search, and any number of re-targeting networks. The customer journey can, and usually does, include any number of combinations of these channels – often in separate, unique instances – and, if you want to calculate ROI accurately, you need to account for each of these instances and the role they played. There could be up to 75% difference in reported results among those common approaches.
This is why we use the Shapley value, which is, interestingly, actually rooted in football. The Shapley value tells us about the impact of a channel on the overall performance of conversion path. In other words, does the whole marketing mix benefit from activity on this channel or not?
The Shapley Value erases several flaws in previously-mentioned attribution models. It is fair, efficient, and since it is data driven, it is consistently remarkably accurate.
The other two models Roivenue uses are based on Markov chains. These tell us the direct impact of a channel based on the following touchpoint in the conversion path. A channel with high attribution value according to a 1st order Markov model has a positive impact on conversion rate of the following model in the sequence. In the case of Markov 2nd order models, the impact is on the channel two steps ahead in the sequence.
It sounds complicated, we know. But, the fact is that it is complicated. This is exactly why we built Roivenue – to turn your increasingly complex marketing data into simple, actionable, and accurate insights.
Data has become the foundation of modern business, but business knowledge of how to use it to its full potential has not kept pace. We held an event with our colleagues from AT Internet in an effort to help marketers understand that marketing ROI is both achievable and provable.
The digital marketing and advertising environment grows more complicated and disparate almost daily, with data flowing in to businesses from dozens or even hundreds of sources.
With better data, comes better decisions. Here, Roivenue’s Pavel Sima discussed how Roivenue’s data-driven attribution modelling allows marketers to invest intelligently. The days of last-click and first-touch attribution are gone, watch the presentation to find out why.
Measuring performance means measuring performance everywhere. Consumers are mobile, but mobile is but a part of a larger, multi-device ecosystem. Marketers and advertisers need to know how their mobile products are performing and, even more importantly, that data needs to be part of their larger analytical framework. Adrien Guenther from AT Internet discusses more here:
Roivenue helps you measure, analyze, and improve your marketing ROI. Get in touch to start making your marketing investment do more for your business.
To learn more about AT Internet’s analytics solutions, get in touch with us here in Prague and we’ll get you started.
Tomáš Mikletič manages 14 e-commerce sites in 7 Markets, with a mix of specialists and agencies. Roivenue sits at the heart of that process. We hosted Tomáš and a group made up of marketers, brand managers, advertisers, and analysts to discuss how the right tools can free up time for getting real, meaningful work done.
Watch the entire presentation and discussion (in Slovak/Czech) here:
Did You Know?
Google recently announced a free attribution tool, but it really won't cut it for marketers with complex investments. Our CEO Pavel Šíma explains in-depth.
Everyone loves conversion rates, but without greater context, it's not nearly as important as some marketers think.
Conversion rate is an important metric, but it’s not the be all end all of KPIs. The fact is that growing your site will likely decrease conversion rates – and that’s not necessarily a bad thing.
Despite its limitations, conversion rate is still a powerful metric that works very well for measuring specific tasks – things like building individual landing pages around conversion and email marketing.
Here’s an example, with two alternative tables of numbers for a site bringing in £565k in revenue over a given period:
We can see that most channels convert between 1% and 3%, but direct visits to the site and visits via email are far more likely to result in a sale (25% chance and 14% chance, respectively).
On its face, the answer seems obvious: invest more in the high-converting channels. But, that misses the point, and is exactly why focusing solely on conversion rate can be detrimental. Channels like search and social certainly support conversion through awareness and visibility, even if users aren’t directly converting through those channels.
For argument’s sake, let’s cut investment in low-conversion channels:
The overall conversion rate has more than doubled, revenue is the same, and we've spent less on advertising.
That all looks fantastic at first glance. But, we've turned off most of the growth channels of the site. Look at the second table again and ask, "In a year's time, will we still be able to squeeze out new sales from our same old email list?" Or, "Will we be able to win back the customers our competitors have grabbed from us through their PPC and affiliate campaigns?"
Measure Conversion Rates Effectively
Break your conversion rate down by channel:
Break conversion rate down by visitor type:
Break it down by task:
Focus on microconversions:
Of course, the best way to make conversion rate meaningful is to measure it relative to more KPIs and tie it to an overall marketing ROI assessment. For more on exactly how to do that, get in touch.