The $100M Problem No One's Talking About in Enterprise Sales

👋 Henry here, welcome to B2B Growth Insights, where I take my learnings from being at Pricepoint Partners to showcase how the best B2B SaaS companies do pricing, growth, and retention. Insights from today’s article:

• 🗺️ How to make AI agents actually work

Today I wanted to share something a little different, and show a behind-the-scenes look at a recent project I worked on.

I’m hoping that sharing actionable insights from cutting edge experiments and custom built growth tools will be something helpful for this audience.

Context

The platform generates revenue through marketplace commissions and advertising packages. Account managers, despite full calendars, needed to stay informed about each partner's initiatives to propose relevant solutions.

What’s the problem?

Most enterprise sales reps will know this too well. It’s unbelievably difficult to stay well informed of their customer’s initiatives, and it’s made even harder if they are handling a brand across multiple regions and business units.

I recently tackled this at a software marketplace where account managers were handling 12-50 enterprise partnerships each.

These weren’t small accounts - most partners operated in 12+ geographies, many with multiple business units, making it incredibly challenging for reps to identify shifts in their priorities & marketing strategies to be able to sell solution packages that aligned with the individual goals.

What was the solution?

We took a crack at a custom built GTM software solution, that would autonomously scrape the internet - trade publications, corporate blog announcements, LinkedIn posts, and even leaked/rumoured product announcements available on social media to arm reps with the insights they needed to sell based on the customer’s strategic objectives.

After scraping the internet every week, it would analyse the text and then do some analysis and provide a suggested upsell package the rep could pitch if it identified a change in marketing strategy or variances by geography.

The idea would be that sales reps could go into the platform, enter the brands they were interested in, and then receive AI generated analysis of what the news meant for upsell opportunities.

How did it actually work?

AI agents can’t complete multiple tasks reliably yet so the best way to ensure accurate and reliable performance is breaking down your workflow into chunks.

If anyone’s curious on implementing this don’t try and one-shot a prompt. It provides way too much work for an LLM and confuses it. I got way better responses when breaking down the workflow into specific tasks, with a specialised AI agent at each step:

1. ScrapeGraph AI to gather partner news from trade publications, corporate blogs etc.

2. GPT-4o mini for summarizing article content

3. Claude 3.5 for analyzing strategic implications

4. A final agent for crafting tailored package recommendations

So what?

(1) This allowed reps to understand far more accounts in much greater detail, and propose packages that are more personalised

(2) By tracking shifting priorities in multiple regions this would allow us to find budget from different departments within partner orgs

(3) Allow us to better sell packages to unmanaged accounts without the need for a dedicated AM.

Results

The output is surprisingly good and recommend anyone else in enterprise sales build this if upsells are a major revenue driver. After a couple of hours tinkering with the workflow we got solution pitches based on:

• A brand shifting from promotional discounts to brand building

• A corporate blog post announcing a partnership that suggested untapped marketing budgets in the US

• A brand had too much capacity after overhiring which enabled us to pitch a partnership at a generous discount

Hope this was helpful!