ANALYTICS

Human led.
Technology enabled.

We are data-wrangling experts. Our technology-agnostic solutions work within your current tech stack to optimize your end-to-end customer engagement journeys. We help fill the top of the funnel, retain and grow customers, engage and drive LTV, and filter analysis by customer preference, motivation, and intent. You could say it’s in our DNA.

Human Led

Allant's Data Scientists are expert statisticians, but marketing analytics is much more than math and algorithms. We operate under the premise that desired outcomes are driven by the perfect combination of Science + Art. Numbers don't lie, but misinterpreting metrics and measures at any step can derail a model. We leverage your industry expertise with our proven experience and quickly become brand experts. We will help you make decisions with your head instead of your gut.

Technology Enabled

Allant utilizes the most sophisticated platforms available today: statistical analysis and predictive analytics machine-learning/AI (PAML). Our tools integrate seamlessly with R and Python and can simultaneously develop hundreds of iterations leveraging various modeling methodologies. In the same amount of time it takes some analysts to build a single model, we are able to analyze the output of hundreds of candidate models to hone in on the best solution.

We will help brands create a centralized marketing database that is accessible and intuitively explorable, with the ability to activate the data across all engagement channels to drive meaningful, actionable, and scalable insights throughout the customer lifecycle.

ACQUISITION

Predictive Modeling is more than just an audience targeting mechanism: it produces a set of expected results across the entire prospect universe. Understanding which prospect groupings are expected to engage ahead of time allows the model to function as an ongoing optimization tool.

Profiling

Do you genuinely know your customers? Allant provides audience targeting with a profiling approach to drive real customer and business intelligence ROI by 100%. This is when compared to random targeting or even univariate selection criteria based on common ranges that describe most of a brand’s customers. For example: Descriptive profiling is knowing that 85% of your customers have household income between $75 - $175K income (and are married with children, etc). To turn the Descriptive profile into Predictive and Prescriptive actionable insight, brands need to understand the differentiating factors between their customers and the general population. They need to learn that the percentage of their customers with an income range of $125 - $200K is 3X the percentage that this income tier accounts for in their general marketing universe.

Predictive Modeling

Allant helps brands acquire customers who look just like their current customers with the highest customer lifetime value (CLTV) and who fall in the highest recency | frequency | monetary (RFM) tiers. Since a particular demographic may not always be the most responsive and communication preferences for specific engagement channels may vary from segment to segment, Allant will implement a holistic predictive modeling approach to track at a more granular level. It’s prudent to also implement a multi-pronged approach to focus on segments with lower value but higher response rates.

Customer-Clone Lookalike Models

Allant’s customer-clone lookalike modeling identifies the characteristics that differentiate a brand’s customers from the general marketable universe – or a brand’s predefined subset that is their marketable universe. Unlike selection criteria, which is exclusive by nature (e.g., must fall within a certain age range, income range, or are married or not), a predictive model all-inclusively examines how close someone resembles a brand’s ideal customer across all characteristics simultaneously. So, for example, someone who is not in the ideal range for one of the predictive variables will still score high if they resemble the brand’s customers across the other 10-15 predictive variables.

The OUTPUT is a list of the entire prospect universe ranked in order of: 'who most resembles the brand’s customers' -to- 'who looks least like their customers.’

The OUTCOME – Allant’s customer-clone lookalike modeling increases acquisition rates over selection criteria by 2-3X+. Typically, we capture 70% of the responders from targeting just 30% of the universe.

Response Models

There are always differentiating characteristics between 'those who look most like a brand’s customers' vs. 'those who look like their customers AND are most likely to respond to acquisition communications.' Allant response modeling builds on the customer-clone lookalike model to focus on recent new acquisitions to identify those nuanced characteristics that predict the likelihood of response.

The OUTPUT is a list of the entire prospect universe ranked in order of: 'who is most likely to respond ' -to- 'who is least likely to respond.’

The OUTCOME – Allant’s response modeling increases acquisition rates over customer-clone modeling by 2-3X+. Typically, we can capture 40% of the responders by targeting just 10% of the universe.

Beyond simply evaluating the acquisition effort based on opt-ins, conversions, and ROI, backend results analysis and subsequent optimizations involve asking: Were the hypotheses proven or disproven? Did results align with expectations? Were there unexpected results? Every acquisition effort should produce better results than the previous. We strike gold when we discover a previously unidentified new segment and expand our marketable universe.

Marketing Performance Management Frameworks

ACTIVATION

Early and personalized engagement between a new customer and a brand drives repeat purchases and increases the average transactional spend. When a customer opens the line of communication, relevant and personalized messaging is the doorway to building a relationship.

New Customer Onboarding

Early engagement while a brand is top-of-mind of their customer is the key. A recent client analysis that tracked new customers over three years discovered that 99% of those who opted into email and a loyalty program did so within four weeks of their first purchase. Additionally, those who opted in were 3X more likely to have a repeat purchase within the first month and their average spend per transaction was 4X more. Was it the chicken or the egg? Does it matter if they loved your brand from the onset or if the subsequent marketing drove this behavior? What's important is that when a customer opens the line of communication, it is a brand’s responsibility to seize the opportunity to send personalized, relevant messaging to grow the relationship.

Communication Across Desired Channels

Retention / Loyalty Analytics

Basket Analysis

Sequence Analysis

RETENTION

Understanding a brand’s customers’ transactional behavior, preferred communication channels and propensity for product preferences are vital elements to expanding your brand value. Retaining and growing customers relies on a matrix of customer categorization via segmentation, scoring, testing, and optimization.

Recency Frequency Models

Classifying customers’ transactional behavior is the key to understanding them. A Recency Frequency Monetary model (RFM) can help define behaviors. Even with no demographic detail, Allant can create RFM models to drive audience targeting and maximize revenue growth. This segmentation method groups customers with similar behaviors and allows for ongoing tracking by tiers to focus on those most likely to increase spending.

Organic Segmentation

Customer Lifetime Value

Engagement Scoring

Multi Touch Attribution

Media Mix Modeling

Creative / Offer Optimization

In addition to A/B/C/D testing, there are analytic approaches like Experimental Design, aka Fractional Factorial Analysis. This is a way to test 1,000+ message variations simultaneously. Based on most campaign types, there are six factors brands ideally want to test:

For example, if we test every combination of the above, respectively 4 x 5 x 4 x 3 x 3 x 2, there would be 1,440 possible combinations. Allant’s modeling process selects 10-15 specific combinations to test in-market that statistically represent all 1,440 combinations. After the results are in, the model extrapolates results and ranks all 1,440 combinations in order of most likely to drive the desired outcome.

Attrition / Churn Modeling

In many cases, the highest drop-off occurs after the first transaction or two. Driving the next transaction during the early phases (within a specific period) is most indicative of future customer survival. This Hazard and Survival analysis combination sets the foundation for ongoing churn prevention. Allant’s attrition model identifies those who are most at risk so that focused efforts with strong offers minimize lost customers.

Cross-sell / Up-sell

Many methods exist to entice customers to try other products, the key is identifying the right audience and sending relevant communications at the right time. Allant helps brands leverage the following either individually or in combination for complete optimization:

Audience Targeting: segmentation, RFM, predictive modeling
Next Likely Product: sequence analysis, market basket, seasonality
Timing: RFM analysis, next expected purchase date modeling, seasonality
Offer: vary the strength of the offer based on recency of purchase, next expected purchase date, and redemption history

Predictive Analytics and Machine Learning

LOYALTY

Looking at a customer's interactions and behaviors over time will identify actionable insights and optimize the customer journey and lifetime value.

Longitudinal Data Analysis

Approaches can be expansive and span the customer journey from pre-acquisition to an unlimited amount of time for a complete end-to-end view or can focus in on a single phase. For example, by optimizing the early-phase new customer journey. Allant helps brands understand the pillars, identify gaps, prioritize the most impactful initiatives, and discover any overlapping facets for economies of scale. The process can clearly detail objectives, analytic tools, segments, communications, and channels. Additional layers can detail communication flows with response and engagement-driven cadence variations, business rules, and creative examples.

Next Expected Purchase Rates

Engagement Scoring

Reactivation Winback