Quantitative Marketing Investments
Whether they are called Growth, Demand Generation, or the more granular fields of performance marketing, ROI-focused, performance-based, programmatic, digital, or ML-powered marketing, these types of investment must meet a high level of precision and clarity - the two most important principles Maplerivertree adheres to and advises our clients to follow. In our minds, these functions (and the teams running them) can all be sorted under the umbrella of quantitative marketing investments. It is often used to drive short-term pains in sales attainment, design lasting and high-return marketing channels, support product-led business cycles, or improve brand awareness, thought leadership, or credibility.
Before committing even one dollar into building or funding these types of business functions, your organization must clarify and have an internal alignment on the mathematics: how does one dollar invested in a proposed marketing channel convert into EBITDA, what’s the financial guardrail, or tolerated efficient frontier? How does it reflect customer life-time value, first-time buyer biases, and return buyer biases? What’s the (omni-)channel attribution model? What algorithms, machine-learning framework, or closed-loop system are deployed behind the investment decisions, and how frequently is one decision made (often hundreds per minute through automation in enterprise B2C businesses)? In this initiative stage of planning, data-driven analytics (or the ‘science’ side of business and marketing) must outweigh the ‘art’ of marketing. The best leader or manager to instigate this function in your business should be good with numbers, fluent in both statistics and corporate finances, and have an appetite in adopting scientific methods, especially machine learning related disciplines, into marketing machines. A special note: when it comes to B2B investment in demand generation marketing, utilitarian lead-stage pipeline and a clear collaboration process between the marketing team and the sales team need to be assessed and put in place first. At the end of this page, we added an addendum to go over the basics for lead-stage definitions.
For established and well-managed marketing channels, the next phase of investments generally focus on five areas: One, improve return on investment (ROI, or sometimes ROAS) with a stepped budget plan; two, scale in a short-term (“what can we do with 10x the money, or what’s our entitlement in this space?”); three, refine the proportion of acquisitions or sales by buyer types (“let’s shift more budget and emphasis this quarter to drive younger buyers.”); four, expand channels or marketing levers to other business needs (“can we duplicate existing acquisition efforts to remedy declining renewals, or field-test the Canadian market?”); five, experiments for pure offsite UX improvements or compliance. More often than not, in our experience, a business will set goals to achieve multiple of these areas simultaneously.
In the next section, we will provide an example analysis back in 2017, in which Maplerivertree helped a business chart the course for its quantitative marketing investments.
Please note, while we work with clients at various business stages, when it comes to quantitative marketing investments, we have the most experience working with companies at four general budget levels (annual; USD): $300,000-$800,000, $3-5 millions, $7-12 millions, $23-28 millions. A typical engagement starts with data. We work with your data scientists, data warehousing, tech-stacks, ad-engineers, and marketing teams to investigate growth pains or rooted inefficiencies. To learn more, please contact us.
The company in this example had successfully established a profitable e-comm business and allocated nine millions dollars a year to programmatic marketing channels, at the time when Maplerivertree engaged. The company was positioning itself to be acquired by a larger tech firm. The majority of its marketing revenue came from four programmatic ad products: DSA, PLA, ETA, and GDN. The macro environment in digital saw double-digit revenue growth with +17% and +13% in 2016/17 respectively. Providing this context sets a benchmark for performance evaluation, in much the same way an equity investor discounts inflation rates in its rate of return.
Select visualizations used in the analysis are shown below. Due to confidentiality concerns, the initial chart-by-chart commentary shown below has been removed, and the GMS (gross merchandise sales) figures have been adjusted. ↓↓↓ Horizontal Scroll (Chart 20/20).
The top-level findings from the analysis were:
Exhaustion of core audience. Like many high-growth e-comm businesses at the time (and today), this company had adopted increasingly sophisticated bidding algorithms and DSP (Demand-side platform) environments. Coupled with common in-house P13N projects and audience curation systems, its programmatic marketing arm was already able to bucket audiences by expected ROIs, and repeatedly target them. A gradual erosion thus started to happen behind the scene (while ironically, this phenomenon cannot be readily captured in their robust internal reporting); only after months did the business start to realize a drastic decline in the effectiveness of their channels. Please note, ‘effectiveness’ is measured differently from ROI; a channel can maintain the same rate of return but requires increasingly more complex and frequent algorithm updates and bidding adjustments. Sometimes, similar business might not believe there was an issue with their advanced tools, because in their minds, the tools and tactics were working amazingly (precisely because the sophistication generally produced near-term ROI gains.) and all analytics affirmed the observation; they typically ended up concluding and reporting up to management that the decline was due to seasonality or forces unexpected. In reality, however, the primary underlying reasons here were simple: one, the top percentage of their best audience groups had been fatigued and thus started to yield less and less traffic (data) and profits, and two, budget increasingly had been shifted by the automatic system to high-ROI segments and failed to incubate prospects (because they appeared to have much lower ROI at the onset).
Maplerivertree flagged a high-risk area in the recommendation that the existing investment strategies (nine million per year) was ROI-positive at the expense of exhaustion of core audiences, and it was not sustainable.
Reliance on saturated markets; cannibalization with potential buyers; and untapped opportunity. The company’s e-comm platform mainly traded six types of products, and among them, Alpha and Delta took up over half of their revenue. Product Alpha was at the tailend of its product cycle and in decline; the overcapacity of same-category offerings in the market that was already apparent during the period of maturity was predicted to soon become endemic. In the operational research, Maplerivertree found the marketing team (incl. Ad Engineers and Ops) dedicated over two-thirds of their time maintaining the current revenue from advertising these products. The operational overhead was growing steadily over the last 18 months, and it was clear that given the dwindling customer appetite in Alpha, increased ROI and GMS was not in sight. The team was over-indexed pursuing diminishing returns. Meanwhile, product Delta was the core business that multiple potential buyers of this company were also competing for. It was for this competitiveness in the online auction space, Maplerivertree found the biggest swing in performance in advertising this product, and the most opportunities lost due to insufficient budget. Lastly, an overlooked product category, denoted as ‘Rare’ in the visualization, was neither in the radar of other competitors/buyers nor adequately optimized. ‘Rare’ served a niche market and demographics (a separate analysis pinpointed the age-groups and US Zipcode hot zones for them), and this company had all the tools and inventory to monopolize this market, should the team be willing to shift resources. With all these above findings, Maplerivertree conducted one more trial with the team, in which it was found that a 40% reduction of operational focus on Alpha (lower frequency of bid optimization) only led to an 8% expected drop in revenue.
Maplerivertree recommended as a priority to gradually shift investments, both budgetary and ops, to Delta and Rare. The former put more pressure on potential buyers as the company was positioning itself to be acquired, and the latter opened a new revenue stream to sustain growth.
Mid/long-term buyer value and mobile. (removed)
To support multiple down-stream projects to support the above goals from the ground-level, Maplerivertree subsequently devised three main frameworks for this company. They are:
— Curation of high-ROI audience. This is the bread-and-butter project undertaken by most technology companies that cater to online traffic, subscriptions, or PLM (product-led marketing), and as the cliche goes, when it comes to the audience (or funnel) design, the devil is in the details. In this specific example, a series of auxiliary analyses were conducted, including customer life-time value, total addressable market (TAM), and pricing elasticity, in order to accurately determine the irritation criteria.
— Site traffic closed-loop system design. Closed-loop in this context refers to an automatic loop where results (e.g. the capital Y in your cost function) feeds into the decision-making machines. The most advanced version of this system, at the time of the writing, is the application of reinforcement learning models (link) in business. Be aware of the significant operational overhead before funding this type of project. Maplerivertree typically does not advise building them unless all your low-hanging fruits have been maximized.
— Applying Machine learning model to increase PLA efficiency in advertising select inventory types. This project lasted multiple months, when the Maplerivertree worked with its engineering team to iterate multiple ML models with closed-loop data, remove under-fitting issues (link), and apply (V1, a batch gradient descent model) to real-life bidding. (Details removed)
In Maplerivertree’s experience, quantitative marketing can often be undervalued, especially by early-stage, or high-technical companies. Some wear tinted pair glasses and regard any paid marketing channels as basic, ‘transactional’ pagers. In reality, however, it is one (if not the most) marketing discipline with immense depth, technicality, and not only requires massive determination from the management team to commit to (as it requires large up-front funding and specialized talents), but also can propel a business from within and inform its product roadmap, market strategy, and brand positioning. To learn about how Maplerivertree can support your specific needs, please contact us. ■
Addendum - Lead Stage Definition
Lead stage definition is one of the most overlooked but critical decisions in managing B2B sales, often chosen and adopted because “it is always done this way” or “I know of an exited startup that defined it so.” What Maplerivertree focuses on is the definition of early stages, typically one to five stages, of sales leads, commonly referred to as pre-sales leads or marketing leads, or SDR leads. And we found many companies use ill-defined lead stages and continue to use them year after year, or never consider reassessing whether their lead stages are still compatible with revenue growth, shifting industry trends, or marketing channels. Arbitrary and ineffective lead stages cause inefficacy for three functional organizations: the marketing teams who nurture lead sources, such as tradeshows, online channels, form-fills, the sales development teams who qualify, re-visit, and conserve leads, and the account management teams who close deals from the qualified leads.
Example of detailed lead stage definitions (above) and definition by verticals or market segments (below).
A repeating theme is one where the marketing organization of a B2B business thrives to deliver what many call actionable or prosecutable leads. These leads are cleansed through exclusion, based on specific criteria, such as particular verticals (e.g., public section, when one does not provide government-compliant technologies), or demographics (e.g., students, or low-income zip codes).
These sieved leads, however, are not qualified –they do meet exclusion criteria, but seldom meet inclusion criteria.
In a company with effective product-market fit and positioning, the sales stages should forcefully articulate inclusion metrics, as granular as the size of target accounts, vertical industries, and select public venues. The problem with only instituting exclusion rules is that the marketing team is essentially sending a high volume of unqualified leads to the Sales Development Representatives. This leads to a high cost from manual qualifying and is not difficult to mitigate, since common Marketing automation platforms, such as Marketo, are often made to automate such qualification of leads so that one does not invest expensive manual labor to that qualification step. This nonetheless is not the real problem.
Example of Leads workflow for US (above) and International (below) sales and marketing organizations.
The real problem stems from the fact that, if the business’ goal is to deliver qualified leads to sales, the volume of actionable leads does not tell us whether marketing is performing. This is because the marketing team needs to wait until the SDR team qualifies the delivered leads to determine whether one or many channels drive quality leads. But there are many variables between the marketing efforts and an SDR qualifying a lead that can impact whether that lead becomes qualified. For instance, any time-off or turnover of sales development representatives can influence the gap of time between the marketing activities and qualification and thus leads to misinterpretation of the effectiveness of the original lead source. It becomes even harder if marketing is looking for campaign-specific feedback. Thus the volume of leads from various channels entails inaccurate insight, and bad feedback prevents marketing from running insightful experiments.
Maplerivertree takes up challenges of this
kind and solves them by examining the unique needs and situations of the business.
It typically takes four to five weeks for our consultants to work with your operations team to propose data-backed adjustments to the lead stages definition, as well as the path to success framework.
A proposal includes detailed inbound lead definitions and workflow, which explains the various lead types, their definitions, action owners, stage activities, stage-gate, how one knows if a stage-gate is met, what happens after a stage-gate is met, what happens if a stage-gate is not met, should a lead is sent back to nurture then where is nurture, name of subsequent and the owner of the nurture, and if existing nurtures exist. Furthermore, this definition is further specified for different target B2B customer segments, such as Enterprise, financial services, or paramedical. This workflow also effectively defines the point when the sales process begins – unless leads are later deemed unqualified, and at which point it comes back to the marketing pipeline.
Examples of lead status definitions, comprehensive list of unqualified reasons, and Lifecycle stages.
In addition to the lead definitions and workflow, the proposal will include a detailed lead status definition, as well as Lifecycle stages. The lead status definition clarifies sub-stages in between lead stages, such as an open lead, a re-opened lead, an attempting contact, a contacted, a qualified, an unqualified lead, and an unresponsive lead. For each of the unqualified leads, the proposal lists out all possible unqualified reasons, route-back owners, and actions. ■