What’s the exact problem you’re trying to solve and whom for?
Untrite was founded 6 years ago, initially serving as a tech consultancy and a custom software company. After seeing a pattern of problems related to knowledge and data access our clients were facing, in early 2019 we decided to start developing a product, an engine with aim of removing data silos, and focus our operations on that.
Being proud geeks as we are, we loved how AI landscape has been evolving, how many frameworks and open source projects have been popping up like mushrooms and decided to take advantage of what’s currently possible and do something both useful and cool. We were particularly excited how Natural Language Processing (NLP), a type of AI dealing with understanding text, became a bridge between computers understanding humans who converse in a natural language, not a binary one.
It took us a while to understand a specific problem we wanted to fix and a type of clients we wanted to serve. We had an expertise and experience in advising on and building complex technology products but initially made a mistake of addressing too of a generic problem — finding the right information among all hidden and known data within a company.
The tech we built was an engine, a platform that connected data through APIs or custom built connectors or dump rules (in case of legacy systems) to different systems; pulled that textual, unstructured data and through use of NLP made sense of it. It understood the context, a meaning behind the words and phrases, and was automatically finding connections between different sources and concepts. It seemed like a complete game changer, as you no longer needed to look for information by knowing exact keywords or places it is stored or people who may know about it. Instead, you could just ask a question in a plain, human language.
But nobody cared.
And I don’t blame them.
Don’t ask them to figure it out˙
We didn’t make it easy for our prospects to explain an exact use case; which KPIs can be attached and that’s an expected ROI. We didn’t know it either. We considered so many potential applications that we felt it’s best to run these ideas through companies and see what picks the most. However, just like Rob Fitzpatrick, the author of “The Mom Test” book suggests, you shouldn’t ask your mom whether your business is a good idea, because she loves you and will lie to you. Equally, you shouldn’t ask anyone whom you’re trying to sell to if your project is a good idea. You need to figure it out yourself. You need to sound and feel like you know their problem and you’re the right person to fix it. Especially, that with most of these cold prospects you only get one shot to make them excited.
Sure, sometimes you may find equally excited tech nerd who will go in lengths to find a use case for their company, but that’s rather an exception.
Crafting your message˙
There you go, I will reveal the biggest mystery of its all: whatever you communicate, keep your message simple. There is this really cool book on that — Building a StoryBrand: Clarify Your Message So Customers Will Listen by Donald Miller.
Donald argues there, that by trying hard to sound smart, you’re actually diminishing your chances to make a sale. He advocates for simplifying your message to maximum. He brings an example of a painting company which had three hard-to-find-common-denominator products so they by going too much in detail, they were actually confusing potential clients and losing lots of sales. Donald writes:
“If we wiped the website clean and simply featured an image of a guy in a white lab coat painting something next to text that read, “We Paint All Kinds of S#*%,” accompanied by a button in the middle of the page that said, “Get a Quote.”
You should have that in mind *especially* at the stage where your brand is unknown and in early stages. Both your messaging and product will change million times before you find your product market fit.
Donald argues that all successful storytelling follows more or less the same guideline; there is a villain, a hero and a guide who helps the hero win.
You’re not the hero.
Your customer is, and your task is to help the hero overcome the villain by guiding them. All books, movies and everything in between follow the same template; Batman vs Penguin (where Robin is a guide), Neo in Matrix (where Oracle and Morpheus are the guides) and so on…
Donald gives an example of how a financial advisor should draft his story:
Villain: Financial companies that don’t listen to their customers
External: I need investment help.
Internal: I’m confused about how to do this (especially with all tech driven resources out there).
Philosophical: If I’m going to invest my money, I deserve an advisor who thoughtfully explain things in person
When you define something your customer wants, the customer is invited to alter their story in your direction. If they see your brand as a trustworthy and reliable guide, they will likely engage.
So in our case, Untrite’s story template looks more or less like this:
Villain: Too much dispersed information in silos
External: I want to get only relevant information to my task, fast.
Internal: I’m confused about how to do this (especially with all available systems out there).
Philosophical: Computers should understand what I’m trying to achieve/ It shouldn’t feel a burden to work with information.
When in Rome do as the Romans do˙
This one is closely related to the above. When you’re talking with a software engineer, his interest will be more on the technical side. When you’re talking with a business-oriented person such as head of innovation and such, he will most likely want to know which business division would benefit from your product? Which KPIs could they attach? Who would use it and what would be the return on investment?
We’ve learned many painful lessons that showing cool tech and waiting for the customer to figure out use cases is the worst idea. We knew the rule of thumb for startups well — to not be a solution looking for a problem — yet we were falling into the same trap over and over again because we haven’t tailored our message to a person we’ve been speaking with.
As much as you, the startup, may hate a corporate pace and jargon, you must play the rules if you want to win the contract.
Show me what you’ve got˙
Although not impossible, it’s hard to convey the value without showing a demo or a visual of some kind. The decision makers you reach out to are bombarded with dozens of pitches every day so when you get your shot, you need to “wow” them.
So to iterate, before you land your first demo, you need to figure out where do you make a mistake (because you surely do, otherwise you’d be too busy selling rather than reading this post):
Five stages of hell˙
- Nobody responds to your cold emails:
Hint: K.I.S.S — keep it simple stupid. Your message is probably too long. Get straight to the point; state who you are, what do you offer and why should they care. Don’t promise things that you can’t deliver because usually your prospect can smell bs from afar. - “Sorry, we’re not at this stage yet / We’re in the middle of digital transformation / ERP/sales*
*here place any other excuse”
Hint: Better than point 1, but still causing confusion. Your value proposition is not clear enough or you’re talking with the wrong person — it’s not their problem so they don’t care. - “I can totally see a use case but we’re too small to get a good ROI.
Hint: your value proposition is not still not clear enough or they’re really really too small. But don’t give up here just as yet. We’ve had companies saying that because they’re too small as they don’t have enough data for our AI to work well.
Often, with novelty products you need to take your customer ‘through a journey’, explain that the investment will be returned many times fold and that they don’t need to have gazillion amounts of data points for your solution to bring them value (if that’s really the case). If you can, get some numbers out of the prospect, run them and make calculations of potential ROI — it’s always about either saving the money or making it for the client. - “We want it and would like to run a PoC.”
Hint: Well done on arriving that far. Now, set clear goals and boundaries— what works for you and for them. PoC is usually run at cost so don’t promise the Earth you can’t later deliver / that will cost you much more than it’s worth. - “Where do I sign?”
While in all 4 above cases it was typically you initiating the contact, once you prove your value with a fairly known name, the word of mouth will kick in. Other companies will experience FOMO — they come to you doors/windows wanting what your other client got. Now you call the shots and can choose whom to work with.
I also highly recommend checking Bob’s reasons why some startup-corporation collaborations won’t get off the ground. (It’s obviously for B2B collaborations).
PoCs: Make it easy for them to start˙
By now you can see that the General theme of a successful collaboration with another company is “Don’t make me think”. It doesn’t stop at the PoC stage. We design and run our PoCs the easiest way possible to our customers.
While, big players such as IBM or Microsoft can afford giving upfront credit (usually for enterprise size contracts it’s 200k GBP and doing a PoC stage for free*) you shouldn’t. This is how we do it at a PoC stage:
- We avoid paperwork and bureaucracy by asking for a damp real company data rather than for API permissions and dealing with legal and compliance.
- We find a self-contained, narrow use case, may it be a particular set of machines (in case of a manufacturer) or a business team.
- Because of the above, client investment is relatively low — both in time and money. This means less risk for the client but enough involvement to actually test the relationship. If it works for both parties, we expand into a fully-fledged pilot.
*The little big joke of the industry: As there is no such thing as a free lunch, large companies usually ‘give away’ free credit to a prospect, only to recoup the costs at licensing stage later.
Focusing on your ideal customer and rejecting everything else˙
It’s painful at times because different companies will see a different application for your solution. If so, it will pull you in different directions, leaving you tired and still. Business, like life is a game where you need to learn the rules to play and win, ideally, with a minimal effort. Here, your quest is to decide on your ideal customer asap and reject putting effort in everything else (at least in the beginning).
After many brainstorming sessions and meetings (special thanks to all of early clients who gave our product a commercial chance and have been helping shape it!) we found an inconvenient problem for type of clients we were especially excited to work with.
There are many products that focus on high volume low complexity tickets such as retail or utility call centres e.g. I’d like to know how to return an item X?, What’s your opening hours? etc.
We decided to go the harder, but much more rewarding route:
- We serve clients with specific knowledge / taxonomy so data-heavy organisations ranging from manufacturers, real estate corporations to public entities,
- Their information landscape is complex, enabling people to get information they need in mission and time critical situations.
- We’re at intersection of service intelligence and data management. We focus on customer support. But not the bot/tickets kind of customer support
That means that we needed to put aside everything else, such as solving problems which our tech is an equally good fit but are too far away from serving the main customer:
- HR — such tech could help with faster onboarding
- Marketing/Sales — it could help with pre-quoting since your employees would be able to search through similar projects that have been quoted/done in the past and build on top of that.
- M&A — unifying taxonomy
A common denominator˙
Right now we work mainly with manufacturing clients AND also the UK Police. It may seem unrelated at first, but if you think about it, in case of the latter, the citizen is a *customer* who needs help in a time sensitive situation, while officers need help navigating through a complex, interconnected landscape of constantly changing information.
As a result, we’re constantly addressing the following challenges:
- Compliant data — fake news/ bots — whom to trust?
- Legacy systems — it’s impossible to replace them as they are seen as strategic investment. And rightly so, since they define and make your company different from your competitors. We need to make sure clients have a way of utilising and connecting information from such systems as inter-connected information is knowledge, unconnected information is a dead information.
- Too much data (cleaning, hiding irrelevant one? Defining priorities for different type of users)
- Collective wisdom — probably problem x has been solved already; how can you tap into that resource without getting confused where and whom to ask?
- Language barrier — some materials are available and made by teams in different languages or countries — but how can you know that? Language is a silo by itself.
All in all, technology is the least important piece of a puzzle. Tech is just a tool that allows us humans to get better of us.
Technology should serve the people˙ Not the other way around˙
We’re on the mission to giving control back to people over how they interact with data and the knowledge unlocked in it. Our grand vision is to provide a one place for holistic knowledge within a company — think of it as your best colleague who always knows where a relevant information is or who may know / have worked on something similar to what you’re working on now. Right now we apply it to service performance but we have an appetite to become an interconnected platform with building blocks, modules which you pick and match as you need.
What I learned over these years is that technology isn’t the biggest challenge. Culture is.
(T. Fountaine, B. McCarthy, and T. Saleh)
We’re working on simplifying our messaging even more but I am aware that is a never ending process. Onwards and upwards!
For those who survived such long post and haven’t opened trillion tabs in the meantime — I salute you.
Liked it? Show some love — share your thoughts (be kind!).
