You have recently launched a new product, want to develop a new sales motion, or are planning to conquer new markets. For this, you need insights on the personas and companies that would be a good target. No worries, here is the step-by-step guide to help you identify your new Ideal Customer Profile!
Create an audience
What is an audience?
- An audience is a subset of records (leads, contacts, accounts) on which you want to focus on to better understand their demographics traits.
What should my audience look like?
See what already exists!
If you don't have a clear idea yet of the audience you would like to analyze, you can first review the audiences that have been previously created. Just navigate to the app, click on Mapping > Audience Mapping.
Think about your objectives
If none of the existing audiences suits your needs, think about what you are trying to achieve.
Let's say you want to understand the Ideal Customer Profile of your new Self-Serve motion. How would you define this Self-Serve audience, i.e the population of leads, contacts or accounts that are buying your product on their own? It could be from a leadsource 'Product Signup' or from a Segment event that fires each time someone buys a product. Now you just have to define this audience through the UI mode of the app.
Let's say you want to understand the typology of customers coming from the APAC region, a new territory that you invested in recently. How do you track customers coming from this territory? If you have a field 'Territory' available at the person/account level, you can directly select it and input the value 'APAC'. If you don't have this field, you can also use 'Country' or 'Region' fields from your own CRM that will however require you to input multiple values to reach the APAC coverage (ex: Australia, Indonesia, New Zealand,...)
Other examples of Audiences
- Leads coming from the leadsource = 'Webinar'
- Leads with the title 'VP', 'CEO', 'Founder' or 'President'
- Leads with more than X product events
You can also combine different criteria to reach a more specific audience but remember: your audience should contain enough records to be able to deduce any statistical conclusion. We typically try to have at least 1000 records in an audience.
Step 1 Use the UI to create the filter defining your audience
Navigate to the MadKudu app > Mapping > Audience Mapping, and create an audience in the MadKudu with the specific target segment criteria. This requires:
- Choosing a name for the audience
- Selecting the CRM to use
- Selecting the object
- Selecting the field in the object to create the filter and the possible values
Step 2 Publish the audience
Click publish once you are satisfied with the audience, and wait for the success email on the audience mapping being successfully created.
You'll then need for the map process to run. This happens on average every 4h to 6h: we recommend you wait 6h before proceeding to the next step.
Prepare your ICP analysis on the Data Science Studio
Step 1: Duplicate the live customer fit model
Navigate to the Data Science Studio homepage where all your models are displayed.
TIP: How to find Springbok from the app/from a link.
If you don't remember how to access the Data Science Studio, you have two ways to find the precious link!
- If you go to your MadKudu app, you will notice a number in the URL link. This number is your tenant. To access the Data Science Studio, just replace [tenant-number] with your actual tenant number in the link below.
Example: if the link to your MadKudu app is https://app.madkudu.com/org/1234/homepage then the link to your data science studio will be
- Navigate to your MadKudu app and go to the Predictions tab. Click on 'View Model' to access the performance page of your customer fit model. On the top right of the performance page, click on the 'Diagnostics' button, and then 'Go to the Data Science Studio' in the new page that was opened. You will arrive on the model page of your live customer fit model. Just click on the Springbok button in the top navigation bar to go back to the Data Science Studio homepage.
On the customer fit model that is currently live (you can see that with the orange label 'Live'), click on 'Duplicate' and name your duplicated model (ex: lCP Analysis Product Signups).
Click 'Open' > 'Customer Profile' on the duplicated model.
Step 2: Load a dataset with the audience you created
Click 'Change datasets': you'll be redirected to a page with multiple options to create a dataset.
As you created a 'MadKudu audience' from the app before, you can click on 'Build from MadKudu audiences'.
Two sets of options are displayed:
- Training dataset options
- Validation dataset options
As ICP analysis is only based on the training dataset options you configure, you only need to set them up, while leaving as they are the validation dataset options.
- Dataset start date: start date of the dataset
- Dataset end date: end date of the dataset
- Audience: name of the audience you configured in MadKudu app
- Remove any past converted account: this option enables to remove in the dataset all the accounts that had converted before the start date of the training dataset
- Conversion model: name of the conversion against which the ICP will be assessed. Pay attention to the case here, you'll need to have the exact same name that in the Conversion Mapping page in MadKudu app.
Once the options are set up, click on 'Build and load the dataset' and wait for the success mail informing you of the successful upload of the dataset.
Recommendations to build your training dataset
- Length: we typically recommend to include 3-6 months of data in the training dataset, depending on the number of conversions you have.
- Recency: choose recent months with recent data. However, you should at least leave your average deal length between the dataset end date and the current date. Indeed, you need to leave the time for opportunities to have actually converted. If your dataset ends in August 2021, today's date is 1st October 2021 but your average deal length is 4 months, it means that lof of the conversions will be left out.
- To reach statistical meaning and for you to be able to draw conclusions from the ICP analysis, aim to have at least 100 conversions in the dataset. It means that the number of conversions coming from your audience should be at least 100. You can review this number once the dataset is loaded by going to Data Science Studio > Process > Data and reviewing the number of conversions in the training dataset.
If you want more information on what the different advanced options mean, please refer to this article: Customer Fit Training and Validation datasets.
Evaluate the ICP of the new audience
On the page of the model you created, click on 'Data Science Studio' > 'Process'. Wait for the training dataset to be loaded in the Data Science Studio.
To review the ICP of your new audience, click on 'Univariate Analysis'.
You've now finished building your ICP analysis leveraging the Univariate Analysis feature of the Data Science Studio, congrats! The next and final step is to get insights out of it. See you at the next article!
To Go Further
- How to read and get insights from a Univariate Analysis
- My field does not appear in the audience: have you granted access to these fields to MadKudu? If yes, the reason could be that we are not pulling them yet > How to pull additional fields to use in the MadKudu platform
- Do I need to update my scoring model if I need to go after a new target segment?