• Tina Mani

Building a marketing tech platform powered by Artificial Intelligence

In the last post of this tech series, we introduced the YFret platform. YFret is a virtual assistant to marketer powered by deep learning and artificial intelligence. It is a mobile companion that suggests the best products to promote, figures out the relevant audience on the best channels, and launches the campaign- all in less than 3 clicks. If you are old school, call us “omnichannel marketing automation platform”, if you will. Grow up. We are more than that.

In this post, we want to share the experience of building a marketing technology platform from the ground up. The focus is mainly on the aspects of product creation during early days. The right audience would be a marketing technology enthusiast, anyone trying to build a marketing platform or part of a corporate team that works like a start-up.

This post is part advisory and part knowledge sharing. Following are my rules of thumb when building a marketing platform from the ground-up.

Save the marketer’s time

A marketer will try something that saves his time.  Clearly quantify and articulate how your product saves time and if it really does, how much it saves, say in a week.

Do not assume that the marketer will use only one tool and that’s going to be yours. If you believe so, get out of the cave fast. On an average, marketer handles about 20 different tools. It is highly likely that some of your functionality exist in the present tools. That brings me to the next item.

Give them strong reasons to ditch the existing tool(s)

If you are building a martech platform in the years 20**, even before you get going with your MVP, do a  thorough market study. Talk to real marketers, understand their daily lives and the pain points. It is very likely that existing players already solve the same problem in some fashion. If you believe you are going to disrupt in a profound way [often you believe but may not achieve it], go ahead. No stopping you. After all, without such grit, Google wouldn’t have been born.

Marketers do not like to learn how to use new tools every other week. Their primary goal is to do their job and show results. There are inquisitive and curious folks out there experimenting with tools all the time. Even they feel bombarded with new tools every day. Go Google “top tools marketers should know” and you will figure out zillions of expert advice on “top X tools”.

Keep the technology aside

When you are starting out, keep the technology piece aside. You might have been awarded a  best CTO award or a been a serial hackathon winner, it does not matter at all. Your technical credentials will help at some point but tame the elephantine ego for now. Building complex software is easy. Your customer will not care about your technology stack as long as it solves her problems today, tomorrow and the week after next.

Go with what you know, pick up the right language, database, infrastructure appropriate for the problem you are solving. Design for 2x performance, test for 1.5x performance and ship. Plan early on for scaling when you see the sign of growth. Do not delay your MVP assuming you are going to hit Google scale on days 1 and 2.

Show the value on the face

This should sound simple. Whatever value you promised to your customers during the sales pitch should surface up in your tool. The value/improvements should be easily measurable, quantifiable and comparable over a period of time. Ensure the reports sent to the customer reflect the same. Don’t expect her to visit your tool to measure the performance.

80-20 rule

This one is almost proverbial now. Easy to preach but hard to practice. If you have a list of 100 items, you will have to figure out the 20 that bring unique and maximum value. For that matter, when you are starting up, if you have 100 highly important items, it’s a sure sign of danger.

And if there are 100 features available in your system, your typical marketer will most likely use about 5 out of them. Inquisitive ones will touch about 20 of them. Big is not better when you are starting off.

Come what may, practice this religiously. There are times you would want to murder your colleagues as you fight them out of which 20 are important. Part of the game. Play it hard.

Make it work out of the box

You are on the right path if your product works out of the box. Some of the marketing tools expect code snippets to be added to the system. Some need augmenting web pages. Others may need integration with the customers third party vendors. If your product has any such demands, those may come in your way. How badly you are affected depends on the complexity of the integration and the maturity level of the customer.

If your customer demands a trial period and you cannot prove your worth without integration, your sales cycle is definitely long. It will affect your month on month growth and consequently your traction.

Getting started should be super easy

Do not ever underestimate the simplicity part of the story. I repeat ‘Building complex user experience is easy; Building a usable product is hard’.

  1. Go with a default configuration for everything. Customizing the initial configuration should be a piece of cake and a natural part of the flow.Normalize the user experience for similar tasks, dashboard and workflows

  2. Inline help is more important than dedicated help.manuals. If using your tool requires loads of documentation, videos and training academies, you are old school. Sorry – Hubspot & Marketo!

Declutter the visualization and dashboard

Trumpet your achievements loud and clear. Do not over do it. There are important statistics and then there are rest. If the customer is calling you to understand the visualization, you have failed badly. Bombarding the customer with visuals that she barely understands will force her to take extreme steps. So watch out.

I know start-ups whose sole aim in their lives is to simplify dashboard of other Gorilla products e.g Google Analytics.

Normalize the user experience

Your panel may have functions, experiences and workflow that are quite similar in nature. Ensure that the experience is uniform, the workflow follows the same pattern, the UI elements look and feel are the same. The graphs, charts, tables, pagination flow are normalized too. Note that there is some learning involved in every software. The objective is to shorten the learning time and making sure the customer feels at home the next time around.

The customer takes you for granted

Every customer, without a single exception, wants your software to fulfill the promises you made. Ensure it is bug-free today, tomorrow and next year too. Everyone likes hearing fancy ideas and seeing shiny products. But if the product is buggy and does not work when most needed, your reputation is at stake. How many products have you continued to use when they had glitches or did not work as specified?

Trust me, your next customer will knock on your door if your existing customers are happy. A marketer’s world is closely knit. Testimonials and word-of-mouth are priceless.

Your customer remembers every single bug in your system [that came in her way to perform essential tasks].

Be your own customer success manager for a while

Until you reach sufficient scale, you should be the account manager for your customers. Understanding how hard it is to do day to day operations and get the results. In our case, most of the features we improved or removed were because we learnt this working very closely with our customers. Otherwise we would have just assumed everything was great and lived in a happy shell.

Produce weekly, monthly, quarterly reports on conversion improvement

Reporting is considered to be one of the mundane tasks by developers – but the most useful. User experience should include even your reporting. Your most important metrics should reach out to your customer by email or notifications. You can confuse or bore your customer [or both] if you do not focus on the reporting experience.

Managing customer specific customization

Your first few customers are going to be key not because they are going to turn you into a millionaire [assuming you are not already] but because they are your guinea pig. Alright, let me refine it. You are going to learn from your first few customers. So it’s natural to be humble and be a keen learner. The catch here is, most of them would ask for simple to expensive customization. You will have to decide which one will improve your core offering [in that case it should be in your backlog]. Be gutsy to say no when it does not add value and yes when it helps improve your product. We have been distracted many a times due to noncore customization. Time is money.


If your product offers UI components that the customer can choose and customize, offer her few common out of the box layouts/themes. Enrich the components by adding every fulfilled request for customization.

Make the most commonly needed workflows available out of the box.


Payment gateway

Chasing customers to get your bills is not a happy ride. Don’t be surprised if some of your customers de-prioritize your payments over other items. Secure your payment in advance. Provide an option to top up if the fee is based on usage and if the customer exhausted her credits. Implementing payment integration into the panel [on day one] seems like extra work considering the size of your backlog. This is not exciting work but very much necessary. You have to create a workflow for low balance reminders, reactivating the account and top up flow as well.

What’s next

In the next post, we will talk about what goes into building the technology stack of a marketing platform from the ground up.

If you are a marketing technology enthusiast and would like a demo of the product we have built, reach out to us.

YFret – The Ecommerce Marketing Assistant


Bring your digital interactions to life with conversations 

Site designed by Brandhero  © All Rights Reserved