When we first started talking about improving win rates at Bizzabo, the problem wasn’t access to information.
In fact, we had plenty of information:
- Salesforce fields
- Deal notes
- Call recordings
- Strong opinions from people who were close to the work
What we didn’t have was one incredibly important perspective:
Our buyers’.
Our understanding of why buyers chose us or walked away was incomplete. We knew we needed to talk to buyers directly.
But, we didn’t have extra headcount, a dedicated win/loss tool, or a researcher to run this full-time.
So we built a bridge and started a win/loss program using the tools we already had.
The challenge enterprise PMMs quietly work around
Our product is an event management platform. It’s enterprise-grade and very robust.
Different buyers evaluate completely different parts of the product depending on their role, their event strategy, and where they are in the business.
At the same time:
- Our product marketing team is small
- Budgets were already locked
- Leadership wanted specific insights, grounded in data
Even with these constraints, there was no reason to wait for the ideal setup.
We started with the tools we already had. Our tech stack included:
- Gmail and LinkedIn InMail (outreach)
- Google Meet (interview recording and transcription)
- Copy.ai (interview analysis)
- Google Sheets (raw outputs)
- NotebookLM (data analysis and explanations)
- Google Slides (executive summaries and results sharing)
What I kept manual on purpose
When we started the program, I owned buyer outreach myself. I basically became a BDR for a while. Cold emails, follow-ups, scheduling, and conducting the interviews themselves.
It took a lot of time.
But keeping this part manual did a few important things:
First, it kept the process human. Buyers were more open when the outreach felt personal, and response rates were higher than I expected.
Second, those conversations had nuance I didn’t want flattened too early. Tone, hesitation, emphasis, the way someone framed a concern before backing off of it.
Leading those interviews shaped how I built our AI workflows. It clarified what actually mattered in a buying decision, and what I needed AI to help me see at scale later.
My decision driver framework
This is where AI came in. After each interview, I ran the transcript through a Copy.ai workflow. Every conversation got tagged the same way, so we could start to see trends.
Since buying decisions rarely come down to a single thing, I mapped every interview to:
- Three decision drivers
- Sentiment, either positive or negative
- 1-2 sentence explanation of why
Here is my full list of decision drivers for you to steal:
Marketing
- Brand strength and reputation
- Case studies and testimonials
- Thought leadership
Sales process
- Trust and professionalism
- Methodology and approach
- Decision-maker buy-in
Pricing and value
- Perceived ROI
- Cost and budget considerations
- Pricing clarity
Product
- Features and integrations
- UX/UI
- Pace of innovation
- Security and reliability
Support and service
- Implementation and onboarding
- Ongoing support
- Self-service resources
Buyer factors
- Timing and urgency
- Switching costs
- Org or leadership changes
- Product fit
Alongside those decision drivers, I used AI to tag whether the buyer was ICP or non-ICP, identify evaluation must-haves, and note which competitors we were up against.
Everything was fed into a spreadsheet, which in turn fed into a NotebookLM that we could query to surface trends.
Where AI actually earned its place
Once interviews were complete, AI became genuinely useful.
Instead of manually combing through transcripts and notes, I could:
- Apply the same decision driver logic every time
- Reduce my own pattern-hunting bias
- Work with a dataset that actually scaled
Eventually, we shared the NotebookLM organization-wide. That changed how people interacted with win/loss entirely.
Leadership didn’t have to wait on a follow-up deck or ask me to re-run analysis. Everyone could explore what they cared about, when they cared about it.
That alone increased trust and adoption.
What patterns we could finally see
Somewhere around the ninth or tenth interview, patterns definitely started to emerge.
One of the clearest showed up around pricing.
At a glance, it seemed like we were losing a number of deals on price. If we stopped there, the conclusion would have been obvious: maybe we needed to revisit our pricing structure.
But when we segmented the data, the story shifted.
Those price-related losses were concentrated almost entirely in non-ICP deals.
For our ICP buyers, pricing was (of course) part of the conversation, but rarely the deciding factor. Martech integrations, scalability, and data cleanliness mattered more.
That distinction changed the conversation internally. Instead of launching a broad pricing project, we refocused on how we targeted, positioned, and sold to the buyers we were actually built for.
That insight came from seeing the same pattern repeat, once we finally had the structure to recognize it.
If you’re building this without ideal resources
You don’t need a perfect tech stack to start.
You do need:
- Alignment on what you’re trying to learn
- Willingness to keep some parts human
- A way to recognize patterns in the data.
Audit what tools you already have, write a project brief, and decide what’s worth doing manually for a while.
The goal isn’t perfection at the start. It’s learning fast enough to know what’s worth tackling next.
Are you working on win/loss or competitive intelligence at your company? I’d love to talk more! Find me on LinkedIn.
