Unlocking the power of B2B SaaS sales insights with AI

Over 90% of the features that B2B SaaS teams build, simply don’t matter. They don’t matter to customers and the growth of the business, they might however, matter to the c-level that pushed for it or the sales rep that claimed it would “get the deal through”.?

“[Most] features are not must-haves at all; in fact you’ll do yourself a great service if you won’t develop and launch them.”

• Itamar Gilad, Must have features

Without solid insights, it's easy for strongly held opinions to find their way through, but this is not how successful product management works. Counterbalancing these opinions with customer and market research is crucial to building a roadmap that is outcome-focused.

Despite the best efforts of product teams to stay close to the market, relying on a few prospect anecdotes as proxies for the larger market and customer base is not enough. B2B SaaS companies fail to build high-impact products when they lack a deep understanding of their customers' and market needs. Without this understanding, companies build products that do not effectively solve their customers' problems, resulting in poor win rates, low attach rates, and flat ARPA.

Sales product feedback is underutilized

Customer feedback is biased in a good way because your current customers understand the product and its challenges, while sales feedback is raw, unbiased and is a better proxy for the overall market segment. Too often, sales feedback hits the product team when it serves the sales representative, that is, to close a deal.

“The prospect really wants to understand when we’ll deliver the [any feature] they’re after, if we committed to this on our roadmap, we’ll get the contract signed”

• Every B2B SaaS Sales Rep

It’s easier for product teams focus on customer feedback over sales and prospect feedback due to it’s accessibility. Sales feedback is underutilized because it is difficult to access and is stuck in CRMs, sales reps' heads, and emails.

At best, product teams rely on traditional sales insights methods that rely heavily on manual data analysis, which can be time-consuming and prone to errors. Moreover, these methods are limited in their ability to identify patterns and trends in large datasets, making it difficult for product teams to extract meaningful insights from their sales data.

This is where AI comes in. By leveraging AI, B2B SaaS companies can transform their sales feedback into the fuel they need to scale their products. AI can analyze large amounts of data from prospect touch points and provide valuable insights into needs and market trends by various dimensions. With this information, companies can make informed decisions and build products that the market actually want and need.

AI-informed insights are already redefining spaces

AI is already transforming how Customer Success teams leverage feedback, utilizing Zendesk’s AI for instance, to identify common customer questions and problems so teams can generate solutions and improve customer retention. Gong's AI-powered platform enables sales teams to identify patterns and trends in customer interactions, which helps them optimize their sales strategies.

Product Teams are just beginning to learn how AI can improve their processes, such as aiding product ideation, competitive intelligence summaries, and drafting survey questions. See Aatir Abdul Rauf’s post on the 45 ChatGPT Use Cases for Product Managers.

How to extract value from sales with AI

On any given week, a B2B SaaS organization is logging hundreds of relevant prospect conversations via calls, emails and meeting notes. Simply reading through this content is a daunting task, let alone pulling signal from noise and connecting the dots across conversations.

Let’s look at an example post-meeting sales note for a fictitious social media publishing tool -

Met with the new CMO (Linda) at Acme, Inc. and need to get her buy in. We have a call scheduled with her tomorrow and setup this call before the call to game plan and discuss what can we show Linda that will be most impactful. Key pain points of their current solution (Publishing.io): lack of visibility into the social calendar. “it’s hard to slice the social media calendar by different dimensions: author, channel, campaign, etc.”. Ultimately, Linda wants a social calendar that can scale and wants to bring together all marketing, there’s a need to see things more holistically. They are currently going through an assessment of channels and capabilities; and developing in a lot of different areas but not in an integrated fashion. Their team is reactive; and they need to be able to plan simultaneously. Key requirements: 1. integrations with top social networks, primarily Facebook. 2. Ability to export their social calendar as a PDF for distribution. They’re looking at 2 competitors, Buffer and Percolate. They also need help crafting content with AI, and love Buffer’s new AI assistant - that would save them a lot of time.?

It's easy enough to quickly understand the needs, current solutions, alternatives, and future opportunities for a single prospect. However, doing this across hundreds of different conversations in various formats is untenable for most teams.

But rather than parse this out and classify it manually, let’s leverage ChatGPT do to the heavy lifting. Let’s create a prompt to put this example feedback into a basic taxonomy: Problem, Current Solutions, Needs and Questions.

Prompt:

I’m going to share sales feedback for a B2B SaaS social media calendar solution. Create a table and parse the sales notes into individual points. Add a column to the table called “Topic” and associate the different Points to the following values “Problem, Current Solution, Needs, Questions”        

Output:

This is already far more consumable but for better classifications we can apply themes to the feedback - yet another dimensions we’ll be able to pivot on down the road.?

Prompt:

Great. Now add another column to the table called “Theme” and identify the theme of each point into keywords or short phrases. Points can have more than 1 theme per row.        

Output:

The output needs some help refining the taxonomy but over time the model can learn about your product, features, use cases, etc., and improve the classifications.

Moving this output into a DB, Google Sheets or Airtable, you can begin to aggregate other sales notes. Note, this is manual process today but can be automated end-to-end to scale to thousands of CRM inputs.

You can further append other attributes to this data set from the CRM or your existing product taxonomy, such as -

• Prospect name
• $ Opportunity Size
• Market Segment
• Product Line Name
• Won/Loss
• Use case

Extend this across more sales inputs and you’re starting to build a rich database of insights from the market that can be used to answer critical questions, identify trends, and normalize the conversation about product priorities around real, structured data from the market.

Here you can see an example output in Airtable, appended with other sales attributes.

You can explore the raw data or build views to answer questions such as -

What are the [PROBLEMS] of prospects related to [TIME CONSUMING]?

What’s the [CURRENT SOLUTION] for prospects with the problem of [SILOED TEAMS]?

What’s are the top [NEEDS] for [ENTERPRISE] prospects?

Leveraging ChatGPT to extract better insights for your product team is still a manual process, yet it’s still more efficient, and far more effective, than parsing and classifying sales feedback manually. The more sales insights added over-time, the more valuable and accurate this data set becomes and the more clear your product opportunities become.

If you’d like to discuss how to enable your product organization to build high-impact products with better sales insights, let’s chat.