Working on a new idea, and then discarding it

Over the last month, I’ve been exploring a new idea in the cold outbound sales space. The idea is to generate personalized cold emails at scale using AI. Currently, there is a trade-off between quantity and quality when it comes to sending cold emails: Either you spend lots of time researching a prospect and crafting a personalized email, or you send generic emails in bulk to a large group of people. Naturally, the response rate for personalized emails is much higher than the generic ones, so I’ve been looking into how to do this at scale.

What I like about the idea is that it’s very easy for companies to justify spending money on solutions in this space and it became clear quickly during my research calls that they’re willing to spend more in the space. You can also quickly see if a product like this increases results and hence increases revenue.

I developed two prototypes exploring a potential solution. The initial versions were focusing on generating an opening line for the cold emails. It would do this by searching for any podcasts, blogs or articles that feature the prospect that is being contacted. If nothing was found, it would then search for any news about the company they’re working for. If still nothing was found, it would search the company website and try to find something interesting to say about the company.

Below are videos of the prototypes:

After working on this together with a small group of potential customers, it became clear that the main problem would be that for most people and most companies there is not much content about them online. Hence the generated opening lines ended up being fairly generic anyways. The other problem I found was that even if there was content online, it’s often difficult to tie this into something relevant to the product the seller is contacting them about. This is a significantly harder problem to solve since each seller is looking for different signals that indicate a potential customer may need their product. These signals differ wildly between companies.

It’s not clear to me how to proceed from here and hence I’m going to stop working on this idea.

Noticing another opportunity

I did notice another opportunity in the space, though. For the companies that spend hours researching prospects to personalize their reach out, there is an opportunity to save them time. These companies tend to target accounts with larger deal sizes and hence are willing to take the time to research things like annual statements, company announcements, LinkedIn posts etc. I suspect you may be able to save them time by gathering all the information for them, summarising it and extracting the signal that indicates they would be interested in the seller’s product.

I’m not sure yet if this is technically possible though, and it may run into the same issue as with the previous idea in that most companies do not have much relevant content online. This risk could be smaller though since we’d be researching larger and more established companies and so there may be a greater chance that they have relevant content online.

I’m not sure yet if I will explore this idea. It feels like a good time to pause and think about if there are other compelling ideas that I could be working on instead of this one. I’m going to take some time to evaluate this and other ideas before deciding which direction to proceed in!