Release Calendar & Notes for July 2024
| Features | Current Env | Go Live |
|---|---|---|
| AI Twin | Test | 16-July-2024 |
| Copilot | Test | 16-July-2024 |
| Insights: New Interface | Test | 16-July-2024 |
| Improvements | Test | 16-July-2024 |
We have seen the benefit of being able to query multiple reports of different types in conversations. This allows for a more flexible way to interact and truly get insights out in a simple manner rather than going through a ton of charts and graphs.
At the same time, we are often challenged by the survey/primary research companies and panels, that our data is very “quant” and theirs seems more “personal”. To close this gap by leveraging our advantage of deterministic & observed data, we invented the AI Twin. It’s a embodiment of the customer you want to conquer. It represents who your customer is, what they think and what they say/feel.
Speak to it like you would to a real person.
Understand nuggets like favorite music, movies etc.
Get to know the reasoning/rationale behind consumer decisions
Figure out gaps in consumer perception
Identify product feature improvements
Co-create marketing ideas/plans
Important Things to consider while making the Twin
1) The selection of the “Stage” of the Twin is crucial, as this will determine what type of an interview you will conduct and how the AI Twin will respond. A potential customer will be very different from a churned customer for example.
2) The reports that you select are equally important. Some examples:
a) If you are selecting a potential customer, then a 3rd party interest or a website visitor audience is a good choice for the persona (behavior report). For loyal customers you might want to take customers that were acquired a few years ago or have bought 2+ products.
b) Similarly, for keywords in intent/mentions reports. category keywords coupled with just brand might be good for potential, “review” might be a good suffix for intent reports in “About to Purchase” stage.
There is no specific instruction here, but it’s important to be aware of what type of reports you are generating for the twin you have in mind.
In conversations, you can now, pick if you want to speak with Copilot or with the AI Twin. So what is co-pilot and how is it different from existing conversations? It’s quite similar and some might not even notice the change, but under hood it’s a lot more intelligent and has a few newer workflows that will benefit users.
Different Modes of Copilot
It’s important that we first understand the different modes of co-pilot to know what it has evolved and what is capable of.
1) Analyze: In this mode, the AI will answer questions related to the “assets in use” exclusively. This is the default mode when you choose assets from the history table and start a conversation. In this mode, if a question is asked that cannot be answered from the existing assets, the user will be suggested to switch to another mode. The advantage of this mode is that its super-fast and is 100% factual with very little chance for AI hallucinations.
2) Investigate: In this mode, the AI will first look up the “assets in use” if it cannot find an answer within it, it will do a best effort search across all reports in the last 30 days in the current portfolio. If it finds appropriate assets that can answer the question it will, else it will prompt the user to switch to Explore mode. In this mode, since there are additional searches, the response time will be slightly slower than in Analyst mode. Hallucinations are possible, since it might think a report is relevant, but it might turn out to be exactly what was implied by the user. This is the default mode when you start a question directly from omnibox.
3) Explore (Experimental – Live only in test and wont be pushed to app until stable): This is the mode powerful and intelligent mode in conversations. This mode combines the analyst and investigate capabilities, along with the ability to generate “custom reports”. These are not your typical behavior, intent or mentions reports. Rather they are a combination of cards that are specifically created to answer the user’s question. This is an agentic flow. Which means that the AI creates a plan, selects the exact tools that are required to answer the question and then orchestrates it all to produce a final answer. Its highly recommended to see the demo video to understand how this works better. Needless to say, this is the most expensive mode. In the sense, that its computationally heavy, commercially more expensive and time consuming.
Example questions for each mode
1) Analyze:
a) “provide an insightful summary for my boss who is the creative head.”
b) “help me understand our customer in a 360 degree view.”
c) “what is the difference between these two audiences? Give me a rich answer that is intuitive to understand.”
2) Investigate:
a) “how does this audience compare to our loyal customers?”
b) “in addition to these reports, what other things are people talking that might affect this analysis?”
3) Explore:
a) Who would our closest competitor be and how should we address the gaps between how people perceive us and them?
b) Are nike lovers younger than those that like adidas?
c) Are our customers more affluent than people who love to ski?
d) How do people perceive the latest apple airpods vs google pixelbuds?
Insights mode gets a facelift. You should feel this difference in a pretty obvious manner. This is the columniation of 3 months of hard work on the UI/UX front. Countless simulations, user journey mapping & heated debates have led to this final flow being designed and developed.
Polish: While we have always had an intuitive and approachable interface, it is rough around the edges. This release is a big step to ensure our platform looks and feels far more polished.
Incorporate newer journeys: To better integrate conversations and support newer forms of conversation like AI Twin and Copilot it’s a must
Consistency: The different parts of the product should feel the same and consistent. This helps to reduce user friction and smoother sessions.
Positive Feedback: We want the user to know that they are doing the right things and that positive stuff like “reports have been correctly generated”, “Your AI Twin has been built”. This is something we are still very nascent at and expect a lot more improvements in this area in the coming releases.
Improved keyword funnel accuracy by ~12% taking it to 89.9%
Standard mentions report time reduced by ~40 seconds
Advanced mentions report time reduced by ~110 seconds
Standard intent report time reduced by ~30 seconds
Advanced intent report time reduced by ~60 seconds
Asset table renamed to “History”. It now contains assets (reports, documents) and conversations
Several bug fixes/workflow enhancements to the “Improve” section to better align the multi-platform expectations
Massively improved the memory requirements of creating word clouds (applicable for intent, mentions insight reports & contextual plans)
Google Custom Segment API Push – Along with download CSV, users have the ability to push custom segments directly into the ad account.
Added a bypass feature: If a user does not have access to the Meta BM, but we would like the user to be able to do 1st party runs, the admin can enable this. This feature will help us reduce friction for new “non-media” users and use cases.
Overall, these improvements not only make our outputs way faster, but also enable our system to handle more parallel requests, thereby increasing reliability and scalability without adding additional infrastructure.
consumr.ai is a consumer intelligence platform that analyses behavioral data from tik tok, meta, google, snapchat and pinterest to tell you what your audience thinks, searches, buys and content they consume. Unlock the power of first party intelligence strengthened by real time behavioral signals from digital marketing platforms.
consumr.ai is a consumer intelligence platform that analyses behavioral data from tik tok, meta, google, snapchat and pinterest to tell you what your audience thinks, searches, buys and content they consume. Unlock the power of first party intelligence strengthened by real time behavioral signals from digital marketing platforms.