The power of artificial intelligence in the modeling of promotional scenarios for predictive ROI and marketing performance is already becoming a “Must have” for TPx. But where else do we look to AI to improve the entire demand chain where trade promotion plays the central role?
By: Rob Hand
CEO, Hand Promotion Management, LLC.
May 22, 2023
We’ve all now seen and heard the recent avalanche of fears expressed by the media about the potential devastation AI can cause in our lives, and how the runaway advancement of AI technology could even end the human race!
Even Elon Musk warns us, and if there is anyone in future tech we need to consider as an expert, it’s him.
Lately, we have been inundated with visions of AI-driven programs that can take a current news story and create AI versions of real video that replaces faces and actually looks like someone is taking an action they would never take or worse, that something we all knew as science fiction appears to be science FACT.
An example of that was a fake news story about a military invasion with video of tanks rolling and missiles firing, buildings being hit, massive exodus of refugees and huge explosions everywhere—all AI-generated and so real that many news outlets actually led with it in their daily updates.
And even though that IS happening in Ukraine, this one was in a place you would not expect it to be, which made the unbelievable all the more believable. It was a pure AI-driven video.
OK, so THAT’s scary, and we should fear the potential of out-of-control technology that is designed to fool us into thinking or acting in a certain way that would be otherwise detrimental to life. But before we begin to look for terminator robots, let’s back up and consider the good AI can do in the future.
And let’s think about trade promotion.
Promotion Funding
Digging deep into the individual promotional tactics to determine the impact they individually have on the value of a particular promotion or cost performance over time is rarely done in CPG—that is usually done at a much higher level. That is not to say P&L measurement is not done, rather going into the detail of the cost of each tactic relevant to the resulting sales out versus the cost of the sell-in is a KPI rarely measured. In fact, determining what the KPI is and creating the right metrics for it is often impossible because of the lack of detail in the data.
Recently, I spoke with the vice president of finance for a growing natural foods company who had the luxury of establishing the trade promotion data in the beginning of the company’s lifecycle with a new TPx vendor with the capacity to add more fields of data than are typically found in TPM solutions. He knew the importance of detailed historical promotional data.
In addition to the standard field of tactic type, new fields were created to capture the specific cost of each tactic for each promotion and establish a table of ranges that were populated as the planners added new “costs” for each tactic in future promotions. This gave the RGM teams the ability to measure specific costs per tactic against actual promotional sales results. In addition, the costs were compared with future sell-in deals where tactics were created at different rates.
Using AI, the finance teams were able to rapidly determine the most effective ranges of costs in the promotional performance against the cost of the promotion configured at the sell-in. Using this data, the AI-driven machine learning was able to deliver a predicted result to the promotion planner that enabled a more cost-efficient use of the funds.
In the use case we created, there was a 7% reduction in total cost of the promotions across the last two quarters of 2022. This meant reduced trade spend as well as more promotions!
Think about how much that is, when you are looking at a $800 million revenue line and a 22% trade spend.
Trade Promotion Optimization
Today, AI has a proven history of success in what is commonly known as Promotion Optimization. AI is used to generate realistic predictions of trade promotion performance through continually improving the machine learning modeling of scenarios based on both historical data as well as external causal factors and market conditions.
While we are clearly in the early lifecycle stages of this technology, enough validation has been experienced by consumer products companies to cement the technology as not only proven but mandated for future trade promotion planning. TPx vendors are continually refining and improving their TPO solutions, and data providers are accelerating their own technology to ensure more accurate and trusted data delivery.
The real value of promotion optimization is not so much the ability to look at past trade promotion performance and project a future outcome, but rather the ability to take more data from outside the historical files. Converting the huge buckets of point-of-sale (POS) data, weather, social media sentiment and content, logistics, instore performance and consumer marketing data into more accurate and realistic predictions will refine the real value of AI-driven promotion optimization.
Inventory Management
Out-of-stock conditions continue to plague the consumer products industry across all categories and market segments. Next to the common cold, this must be the most difficult malady to cure in the world!
We already implement AI-driven systems in our warehouses and logistics, so how is AI going to solve the stock-outs?
In a recent discussion with Retail Velocity, the proof factor of AI-driven inventory and stock management became clear in the use cases presented to control the visibility to the flow of stock from the factory to the store shelf.
According to Chris Spallino, senior director of marketing communications at Retail Velocity, AI is the “secret sauce” to configure business outcomes, but the most important factors are granularity and time.
“Being able to track the performance of promotional items, new products, and merchandising changes where it matters most—at the store-level—is critical,” says Spallino. Using AI with this granular data, companies now slice and dice POS data at any level (store, distribution center, chain, or across all customers), which has already proven helpful in new product rollouts and tests.”
“The pace of the supply chain is now hyper speed,” says the vice president of sales of a global soft drink company. “We can’t operate with just facts—we have to use AI to model the incredible volume and detail of the data we receive, primarily from direct POS data, into predictive outcomes that help us see the future stock flow and inventory levels.”
Spallino points to AI as the potential “end all” of out-of-stocks.
“AI is proven to increase retail execution optimization. A CPG company revealed that a large mass retailer’s stores had variable inventory levels— some carried only two weeks of supply (WOS) while others were overstocked with 15 WOS. After reviewing the data and via WOS exception reports, the retailer and supplier collaboratively identified items at specific stores which were OOS or had low WOS, and store-specific orders were sent the same day to improve in-stock levels. Using only one AI-based dashboard resulted in 116% incremental sales growth for that week, which represented over a million dollar increase.”
When you have the lowest level of detail in POS data, there is no doubt that the volume is huge and certainly a challenge for legacy technology. Their approach is to use AI to more efficiently and effectively extract, ingest, clean, and harmonize the data to accommodate the AI-driven prescriptive outputs smart CPGs use to solve out-of-stocks, expand and increase the precision and trust of historical promotional data and project instore inventory scenarios that prevent wasted promotional dollars.
Consumer Demand
We’re not talking about the demand plan here.
No, we are referring to getting what the consumer needs, wants and is willing to buy in the hands of the promotion planners BEFORE the shopping trip commences. In previous blogs and podcasts, we’ve dealt with the issue of getting to the heart of what the consumer needs and building promotions that deliver to those needs. This is not happening today.
In my book, “The Invisible Economy of Consumer Engagement,” I devoted an entire chapter to how we need to understand the consumer chain, or the need-to-action process a consumer goes through to get to the point of purchase. Today’s sales promotions rarely, if ever, consider anything more than what the key account manager needs to move in product volume, or what the retailer needs to shore up profitability.
Well, no, that’s not fair, the retailers do indeed work hard to identify the consumer demand in their product assortment and positioning, of course. But the hard truth is that, historically, trade deals are done to move volume and generate profit, with the metrics of outcome of the actual promotion often less of a key performance indicator.
Is all that about to change?
It had better.
What we are seeing with innovative consumer products leaders is a realization that AI is the answer to a lot of questions and a salve for a lot of financial ROI pain.
For hundreds of years, the officers’ toast aboard warships and headquarters tents before battle was a wish to “confuse the enemy.” While that is clearly a solid military strategy, it is not a good idea for strategic consumer engagement. With the heavily manual processes of planning and the lack of collaboration between sales and marketing, the end result is often overlapping, missed, and even conflicting offers in promotions. To confuse the consumer is tantamount to not only a failed promotion, but potentially a failed product.
What AI can do is take the broad swath of data around consumer research, advertising reaction, coupon fulfillment, and competitive performance to model realistic scenarios based on real historical performance of the product across all channels and project and predict a reasonable expectation of what WILL happen.
Going one step further is leveraging the machine learning to integrate that data into a promotion that not only drives volume, generates revenue and profit, but also has the right mix of tactics, timing and product to engage the consumer.
There is a lot of nebulous chatting going on about how cool AI is and where it can benefit, but those hundred thousand feet discussions, though always thought provoking, often fail to get to the “nitty-gritty” details of how to make it work.
In the recent Analytics Unite 2023 conference held by Consumer Goods Technology, Retail Information Systems and EnsembleIQ, Mark Edmonson, chief marketing officer of Materne North America talked about how they used $2-$3 off temporary price reductions on GoGo squeeZ pre-Covid only to see the sales volume drop post-Covid. “We spent so much money buying those shoppers versus identifying those consumers who are truly the right people for our brand,” he stated. “Investment in the right technology to get the right consumers to opt into our brand is more important than spending on advertising.”
Ah, so does AI help fix his problem? You bet.
How about blending that great consumer research that is done to determine and identify the “right people” to buy the brand with proven successful historical promotion performance and use AI-driven relationships between the two data sets to target markets and demographics that would deliver those persons to the store shelf or ecommerce site? Adding this level of intelligence to the database for which the machine learning algorithms can be written and executed expands the value of the promotion, does it not?
We have talked now for a long time about how important is it to accelerate the collaboration between corporate marketing who owns the consumer intelligence and sales who owns the trade promotions. This is exactly the kind of blended data that would actually deliver a powerful combination of intelligence that would evolve modern trade promotions into generating real consumer demand.
Siloed Thinking
We continually wring hands over the problems of multiple siloed databases and analytics tools across the corporate organization, making it difficult to consolidate the intelligence we need to execute the business plan. It is a pandemic of its own on a global scale for consumer products companies.
But, siloed thinking also creates barriers against the effective use of costly AI and advanced analytics technology. The longstanding business philosophy of “managing your own organization as a business” has led us to these independent decisions that resulted in the acquisition of multiple systems and now, unfortunately, makes it difficult to consolidate the needs across business organizations to ensure consistency and continuity.
But the good news is that this is changing. The emergence of revenue growth management as a rapidly evolving and elevated function within the corporate organization structure is helping to consolidate key business areas and functions like pricing, trade promotion, product assortment and even brand and category management—all pointed toward the successful execution of the corporate business plan for revenue growth.
While the rest of the world is beginning to issue dire warnings of runaway AI, robotic takeover, and the death of the human race, the practical application of AI within trade promotion funding, management, planning, execution, settlement and analytics is already showing signs of reversing the age-old failure rate of promotional ROI.
That’s a great thing for the human race!
