..And how it can catapult your verticalised SaaS conversion rates.

The era of adding generative AI features into software is now in full force.

Established product teams for SaaS business models are scrambling to be the first to implement new AI functionality that will differentiate their offering for their market.

The question now isn’t so much how to add AI  (as it’s becoming increasingly more accessible) but what should be done that’s truly going to add high value utility and shift the needle (acquisition, retention, etc).

Some of the well established players have already seen success- abet with more obvious implementations (notion with note cleanup, Canva with generative image features), but for more verticalised SaaS it’s not as clear.

Adding AI image detection or AI generated text to a ‘sometimes’ or ‘5% usage’ feature is cool, but likely won’t return the big results you’re hoping for.

If you or your team are looking for that market defining implementation using the latest LLM or generative technology, here’s a couple of pointers we’ve used to successfully release some first-to-market, high-value tools for our vertical.

1. Find ways to reduce your customers time/ effort while also keeping them in control.

Similar to typical software innovation in many ways, by saving your customers time (and in-turn money) with AI, you’ll be looking at genuine problems to solve.

What’s an approach to this? Map out the key customer journey steps and establish where there is direct (or indirect) significant time invested. (Extra points for this time-suck to be prevalent earlier in the customer journey- see point #2 for more on this).

Then, ask yourself- how could AI technology help here?

Couple of examples…

Could it auto setup a profile/ project/ portal based on a handful of responses from onboarding?

Could it auto-schedule a calendar of bookings for staff based on a single CSV input?

What other inputs are you capturing that can be utilized for subsequent generation?

One of the bigger user experience challenges here is that there’s still a somewhat lack of trust with AI (eg. Hallucinations, pulling incorrect or made-up results). To overcome this, provide users with the ability to review prior to proceeding- so they stay in control.

2. Implement the feature as early in the journey as possible

Especially for more niche verticals, AI that’s actually delivering what’s promised is still a wow-factor for prospects.

Everyone knows it’s the future, and they’re all looking at ways to implement it into their business. If you uncover a truly valuable AI solution, you want to highlight it to new trial users early to maximize that conversion.

Not only that, but by having the feature early in the journey you’ll be able to test it with a larger set of users to learn and iterate quicker.

Highlight it by unashamedly calling out the AI feature, throwing on the ‘AI stars’ icon (you know the ones) to maximize those clicks. This maybe is a little cheesy, but the phase won’t last, so while the novelty is still fresh it’s worth capitalizing on it.

How can you showcase it early in the journey? Here’s some ideas:

  • Present a version of it on your landing page (yes, before sign-up) that gives prospects a taste before signing up and using their data.
  • Add the feature into the onboarding workflow. Show them how the inputs result in a high-quality output (that ‘value realization’ moment).
  • Highlight the feature on the dashboard. Great for both new and returning users, you want the feature to be just a click away if it’s delivering great utility.

3. Use a dataset that isn’t just GPT.

The advantage of LLM’s is that the data outputs are so accessible even my dad can get seemingly great outputs with a simple prompt across an increasing number of models (ChatGPT, Claude, Gemini, etc).

The problem is it’s not always great for verticalised SaaS as the data required is sometimes very specific to get accurate, usable results.

The solution to this is utilizing your LLM of choice accompanied with your own data. This could be sourced from your product, your customers data, 3rd party providers, etc. perhaps even use multiple sources of data as an input.

You’ve heard the question that become prevalent with this AI era- ‘what’s the moat?’ well, this is it. It’s where you use AI to get accurate, differentiated outputs from your AI generation. Accompany these results with an interface that seamlessly integrates into your product workflow and you’ll have something your customer really wants.

When creating truly differentiated AI features, It’s important to not forget the age-old  (sometimes cringe-worthy) question ‘what problem are we solving?’

Start by addressing the known problems your customers already face with the added super power of AI.

Deliver these high value features that still allow customers to be in control, while being highlighted earlier in the customer journey and utilize proprietary datasets.

By highlighting ways to think about some of these known AI hurdles, I’m hoping it unlocks a new way of approaching AI in your SaaS so you can get to market with an innovative, new, conversion crushing solution.

‍