Driving revenue through streamlining the content generation process

Inception & augmenting tasks with AI

Our customers faced inefficiency, spending up to 2 hours daily on repetitive tasks like writing copy, which 40% of users identified as tedious. This inefficiency detracted from their productivity, highlighting a pain point for them. Recognizing this, we saw an opportunity to enhance customer satisfaction and productivity by automating these tasks using the latest advancements in AI. We began by identifying text generation as a prime candidate through brainstorming sessions and user feedback analysis. After exploring various approaches to the problem, we developed and released an AI text generation feature. The need for this feature was validated when 10% of our users engaged with it weekly. This initial validation demonstrated the potential of AI to address user problems effectively while aligning the goals of improving efficiency and customer satisfaction. It also paved the way for further AI-driven augmentation, such as automating page creation, to continue enhancing our customers' experience.

Initial steps & ideation

First step was to define the best approach for generating a specific page. I started out with a blue-sky exploration to envision how might we approach generating the page in an ideal world without any technical constraints or limitations. The exploration included the ability to generate custom layout and content based on a single prompt while having the ability to further customise layout variations or colour schemes in a wizard setting. This artefact sparked further conversations with engineering and product over what's feasible and realistic for our timelines.

Scaling down the proposal

As a result of the conversations with engineering and product, what we could realistically ship was generating the page's copy and combine this dynamic output with variations of static page structures and layouts. I recreated the vision based on this outline and after putting this proposal in front of customers, while it generated significant interest, customers mentioned that they would still want full customisation of layout and the static approach of selecting from pre-made layout variations would soon become too repetitive for their needs. This resulted in us taking a different route later on to deliver value by creating custom blog content as a starting point.

A different direction

Through previous research, we uncovered that the usefulness of creating custom content relied in it being highly personalised to each specific use-case. As part of the next phase of the project, we incorporated the creation of custom blog content based on SEO recommendations to increase traffic, being well-positioned to fix some of the most difficult parts of SEO strategy: generating high-quality content to improve a store’s ranking.

Goals & Starting points

We wanted to make it easy for Shogun's merchants to easily generate new blog content to increase their SEO ranking. This would result in generating value through increased traffic and therefore conversion. We set as success metric to increase the adoption and publishing of blog content.
I set out to create a smoke test to validate the need for the feature. In collaboration with the product manager, we conducted discovery calls with the customers who signalled interest to better understand how they currently improve their SEO ranking. Through this, we were able to validate our hypothesis from the get-go: customers wanted help with the creation of quality content to improve their SEO ranking, either because they lacked the time or resources to focus on it.

Solution

The functionality released consisted in the ability to automatically create blog content and edit the output using the AI wizard. The first step consisted in giving a brief description of the blog's main topic. Afterwards, the user had the ability to tweak the content and order of the main paragraphs, embed products from their store inside the content skeleton as well as preview the final text output before generating the entire page.
The flow for generating AI blogs got positive feedback from users in customer calls.

Conclusion & next steps

The project received positive feedback and hit its usage targets since the first two months of its release: users continue to show interest and upgrade to the paid plan containing the feature and dozens of customers have published one or more blog pages.
As for next steps, the plan is to cover more page types beyond only blog pages and to be able to tailor the creation of content to product pages as well.