I’ve been documenting my attribution modelling journey….and as I often say; good is better than perfect. If you’re startup or a smaller brands looking to move beyond lack-touch attribution. Then this blog post series is for you.
This series looks over the last 8 months and the work I’ve been focused on more and more as I struggle and research what’s the best path forward. Like any marketing, this is an orgaism and it’s change and grow over time. Some things will get added and other will get cut away and that’s ok.
The key point is you’re looking beyond last touch and growing your business in the right way.
…as I try to figure out a model that works for startups. It’s not perfect, though it’s a starting point and good is certainly better than perfect so we can launch something and iterate as we go.
If you’re a startup and resources are scarce and you can’t get money to use an attribution vendor (yet), than we’re in the same boat and this post is for you. So today I want to share the framework I’ve started to work on and what shape it has taken. To build this framework I started by answering the 3 attribution questions in my last post.
Defining Success and our Attribution Window
Customer signing up for the service is our main KPI. However, I also look at people subscribing to the blog, downloading an ebook or whitepaper. The last two are tracked because maybe a campaign you thought would be for driving customer acquisition, is actually more suited for blog subscribers or getting people to download an ebook. If your company offered webinars or product demos and you’ve those as goals (or events) in Google Analytics, than you can add that to the spreadsheet below.
I picked a 30 day attribution window for the cookie that sits on a customer’s computer. However, when I think about the effect of a campaign and how long to look at it after it launches. I picked 90 days because we don’t generally have long sales cycles and if someone takes longer than 90 days to convert, it becomes a mess to track. If I had more time I’d look at 180 days after a campaign launches to see its effect on the business. Maybe that will be round two for next year.
The spreadsheet below is what I created when I put everything together. I added in a threshold of $500 for campaigns we track in this birds eye view because anything under that limit doesn’t tend to have enough data to make a proper informed decision that doesn’t risk getting a false positive for you campaign and data. All data is fake below and just to give you ideas of what your could look like.
As I wrote last month, I’m starting to think about attribution at work. I’m doing this because we’re trying to figure out how to judge the success of a campaign we launch. I’ve already written about the 5 different multi-channel attribution models and what makes each one different. As I’ve started down this journey though, I’ve came across one more model that’s interesting and promising: Last Non-Direct Click
This takes all your conversions that show up under “Direct” traffic in Google Analytics (GA) and looks at the second last click before that conversion happened and resigns it to that channel in GA. Many customers will come to your site through a campaign or a landing page and then bounce off your site and or look at other pages on your site. When that happens, those customers will then become a “Direct” channel customer because going from a sub-domain to your main site domain will cause them to get reclassified in Google Analytics.
Attribution has been all the rage at the office the last few weeks. From looking at what would happen if we turned off paid completely, to how we can look at multi-channel attribution but not actually look at multi-channel attribution to judge the success of campaigns we’re working on. To judge the success of a campaign, beyond getting a customer, I’ve started to look at longer attribution windows and look at different goals within our organization.
I’m starting to only hit the tip of the iceberg and though my model is rough and could use a lot of work. It’s a starting point and you’ve to start somewhere…or you’ll never get anywhere.
The last few weeks have reminded me of a good post a few months back looking at 3 top UK retailers and how they have grown up and handled going from last click to multi-channel attribution. A line from the article I like is…
Normally I tell clients, bosses and people I talk with at conferences that I spend 80% of my time focused on 80% of my paid search business because that’s where I can have the biggest impact. I can see the largest return for my efforts in terms of new customers coming in and increasing profitable revenue.
It’s a common knowledge that I always advise clients to look at the top 10 countries that produce 80% of their revenue and make sure each country has their own AdWords account. Your company could have 15 countries that make up 80% of your revenue but I’d not go beyond that amount as it becomes a lot of work for one person to manage.
Organization of information is cleaner across your accounts by country
Learning what keywords & ad text is working for each country. One size doesn’t fit all
Spotting trends in data is a lot easier by country in AdWords & Google Analytics (GA)
Growth isn’t an issue as you’ve room to grow each country like its own mini-business
A bonus is when you hire a person or two to join your team, you can give each person a set of countries to manage and not worry about having to break out each county at that point. If you did, you’d lose all that account history (and quality score potentially) you’ve build up because you didn’t plan for the future.
Now what I’m about to say goes against everything I just said above because you’re not dealing with 80% of your business, you’re dealing with the other 20%. Unless money is unlimited where you work. You need to consolidate resources and make a Global Adwords Account targeting non-top 10 countries.
The Problem 2016. There are more startups than money, talent or consumers can support or even VC money can fund. We’re reaching a point where the pendulum is swinging back in the other direction for […]