As marketers and growth managers, we all want to be more “data-driven”. But what does that really mean? And how do we do it in practice?
Data-driven marketing, even though it’s listed as a skill on everyone’s Linkedin profile, isn’t as ubiquitous as you might think.
In reality, only 25% of businesses are actually using their data in a meaningful way. So why is this?
In truth, it’s not overly complicated. You just need to know where to begin. As more and more of our marketing efforts are becoming measurable and quantifiable, we’re presented with an enormous opportunity to bring new discipline and new structure into the way we conduct “marketing”.
To help you get your footing we are going to give you a step by step guide for beginners that outlines what you need to know to build your first “data-driven” marketing strategy.
One that doesn’t just operate off guesses and instead makes calculated, informed decisions that will fuel and accelerate the growth of your business.
Step 1: Mapping
Take stock. Pause and reflect. Start with the end in mind.
This step involves assessing your business and asking “what do we want to happen?” and building from there.
So let’s begin with a marketing plan. One that maps out your current strategy and how you plan to grow your business.
When starting a marketing strategy, we have a preference for using both customer personas and jobs to be done frameworks. Getting the basics of these nailed should help you to better understand your customer and encompass their wants, behaviours, and motivations.
To give you an idea of what you should be aiming for Buffer did a simple yet thorough guide to creating personas for your business.
This persona template outlines key aspects of your customer – demographics, job, lifestyle etc. These put together, all give a clear, well-rounded picture of who a potential user is for your product.
Importantly, this template not only allows a business to outline who their customer is but also be able to understand what problems you are aiming to solve for your users. It’s the nitty-gritty information of both who and why.
To make it more concrete, let’s look at a detailed example of a customer persona from UX Magazine. They outlined a possible customer persona for music streaming service Spotify.
It’s got some basic yet interesting facts that they believe encapsulates a Spotify user as well as logical reasons why this person would actually want to use Spotify as a tool.
Armed with your customer personas in hand, it’s time to think about how you will actually reach your customer. What marketing channels are you going to use in your growth strategy?
You’ve got to fish where the fish are, so where do these personas swim?
Let’s imagine we’re Spotify’s first growth team. We have our personas and “jobs to be done” and now we need to pick some marketing activities to grow our user base.
Here’s some ideas we might put on the whiteboard:
- Sponsor music concerts: Spend $100,000 on sponsoring high profile music events
- Social Media Integrations: Allow users to share and play music on Facebook
- Top 50 Tech Sites: PR to the biggest tech sites
Good job growth team, those are some awesome, fool-proof ideas! Or are they?…..
Step 2: Scrutinize
Although you may have a clear picture of who your ideal user is and what they want, the truth is you’ve likely made quite a few assumptions.
This target market and the growth strategies you have chosen in reality are just your best guess. There is rarely complete data behind every assertion made.
Which is fine, we can’t have perfect data from day one so we have to do some guess work. So it’s time to start scrutinizing.
Take a magnifying glass to your mapping and list out the assumptions you have made. And try to justify your reasoning.
If we look at Spotify, we listed 3 marketing strategies. For your team, this might be 5 or 10 great ideas for your next growth activities.
After building your list of ideas, it’s time to start unpacking the assumptions within them. Every activity has inherent assumptions – that this pond is where the fish are, that this is the right bait, that these are the right fish.
So let’s take the 3 activities for Spotify and unpack some of the key assumptions we made when we pitched them.
- Sponsor Music Concerts
Assumption: Our customer persona attends music concerts
Assumption: Sponsorships can convert to downloads
- Social Media Integration
Assumption: Our target user likes to share music
Assumption: Users who see shared music on Facebook will click and install
- Top 50 Tech Sites
Assumption: Our target persona regularly visits these sites
Assumption: Reading about Spotify on a tech site will convert readers to users
As you can see, some assumptions feel like safe bets. “Users like to share music” seems like something you can take for granted. But the assumption that a sponsorship will convert to app downloads feels less certain if you’ve never run a sponsorship for your app before.
So we need some way to quantify and prioritize that uncertainty before we can start to eliminate it.
Step 3: Prioritize
To help tackle these assumptions in a systematic way, we use our own framework we call “Data vs. Danger”.
These are two semi-quantifiable factors you can use to rank your assumptions.
“Data” is simply asking “how much data do we have to support this assumption?” It’s a measure of your certainty that this marketing tactic will get results.
“I have a hunch” gets a sad face in the data column.
“We have good market research” and “Competitors have done similar” – that’s much better.
“We’ve conducted user surveys” and “We’ve micro tested the campaign” – that’s the best.
“Danger” asks the question “how big of a risk is it if we’re wrong?” It’s an attempt to quantify the riskiness of leaving an assumption unproven.
“We’ll invest 2-3 hours and $100” is low on the danger scale
“We’ll waste 3 months and $100,000” is high on the danger scale
If you have very little Data, but the Danger is very low, it might be worth a punt anyway. If the danger is high and the data is low you need a lot more research before implementing the plan.
Here is a look at this in action with Spotify.
The “50 Tech sites” outreach ranks pretty highly here. You already know the readership demographics match your persona and what’s the cost of running some PR and guest posts? A few hours, a day at most? Not a huge loss if it proves unsuccessful.
Sponsoring music concerts looks risky though. You know it’s the right audience but you don’t know that a sponsorship will drive downloads. If I see a Spotify logo beside Lady Gaga on stage do I get the downloads I need to make the ROI positive?
We’ve picked 3 examples for Spotify here, but most companies will have 5 – 10 ideas with more than 2 assumptions in each.
At this point 1 or 2 probably seem med-high data and low risk. “No brainers” to get started right away. So you can skip them to “Step 5: Execute”
But the remainder are less certain. Not enough data and too much danger. So it’s time to make them prove their worth.
Step 4: Get Aggressive with Data
The goal of this step isn’t to prove that your growth strategies will work. In fact, it’s the opposite.
We usually have more plans on the list than we have time or money to execute, so you need to put them through vigorous testing to see if it stands up to scrutiny.
Some common steps companies take at this stage include market research, customer surveys, micro-tests and experiments.
Spending a few hours or days to save spending a few weeks or months.
Assess similar companies. Discover what growth strategies worked or didn’t work for them and what measurable results they got from doing them.
Your own current users should also be a fountain of information. Email and survey them in an effort to assess what they prioritise and what their needs and goals are.
With Spotify, we suggested social media integrations, so research can help prove or disprove an assumption that their target customer actually uses social media.
For the sponsorship opportunity, before you go sponsoring Coachella why not try sponsor a battle of the bands contest at a local university?
As with any good test you can build a simple hypothesis with a prediction statement and a conversion goal.
Prediction statement: “A high profile sponsorship of a music event in this genre will generate app downloads”
Conversion goal: “5% of all attendees will download the app”
This conversion goal might be based on the metrics you know need to work for the wider opportunity. You know Coachella costs $25k, it has an attendance of 500k. You can afford to pay $1 per download so you need a 5% conversion rate to make it worthwhile (don’t worry, I did the math!)
For Spotify this would be: If we create Facebook ads it will increase downloads by 10%.
By running an ad on Facebook and having a goal, it will enable Spotify to collect more data and strengthen the argument for the marketing strategy. Simply if we reach the goal of 10% downloads, there is a strong case that an integration is an advisable move.
Here’s a big caveat. Probably the most important takeaway from this post.
Most marketers try get data to prove themselves right. But anybody can do that. Show me the assumption and I can find you some research or data, or phrase a survey in a certain way to prove that – would you believe – the assumption is correct!
Real scientists always try to disprove their hypothesis.
This isn’t always possible in business, but the spirit of it should drive you. These ideas aren’t your babies. Most of them are your enemies. They’ll suck your time and your money and distract you from doing the 20% of activities that will fuel 80% of your growth.
You have a list of 10 activities but only the resources and time to do 2 or 3, so you need to be aggressive with your data gathering and testing to strike as many of them off your list as possible, so that only the plans with the highest chance of success remain.
Step 5: Execute, Measure, Rinse, Repeat
Of course, this testing is only the beginning, but that’s why this post is about building your plan, rather than how to execute it.
With a little luck, you’ve managed to whittle down your list. Striking off some ideas and getting even more certainty as you go.
You don’t have to do it all chronologically either. You can be executing on the high certainty ones while you’re gathering data and (dis)proving the others.
Executing each part of the growth strategy could merit it’s own post, so we won’t dig into that here.
You now need to frequently measure your results and update your underlying plans and assumptions.
The odds are only a small handful of techniques will give you the result you want and 9 out of 10 will fall by the wayside. But you will be able to identify which ones they are and end the lower performing ones before you pour too much time and effort into them.
Building a data-driven plan
The best laid plans rarely survive contact with the customer.
The assumption in this, the final step, is that for almost every business a small (20%) of marketing tactics drive the vast majority (80%)
So continue to measure, test, seek-to-disprove and gradually you will find the channels that are providing the highest value customers at the lowest acquisition costs.
The best teams identify and double down on them and grow faster and more confidently.
The key with data driven marketing is that it doesn’t stop. You should continuously be introducing new channels or marketing techniques to your strategy and using this process to validate them and discover what works best –execute, measure, rinse, repeat.