The Avaus methodology
A systematic path to becoming genuinely data-driven.
Most companies set out to become data-driven. Few get there in a way that produces lasting commercial impact. This is the framework Avaus has refined across 15+ years: value drivers - the business outcomes that matter most to your growth - the Data-Algo-Action unit of work, target automation level, capabilities, and the operating model that holds it all together.
~10x
PRODUCTIVITY INCREASE IN MARKETING AND SALES PROCESSES AT TARGET AUTOMATION LEVEL
3-7%
Annual revenue uplift over a 3 year horizon at target automation level
Quantify your value drivers.
Before choosing a technology or hiring a data scientist, you need to know where the money is. Map your value drivers explicitly, then size each one based on use cases identified and a baseline assessment. Without this anchor, AI initiatives drift into cost centres.
Rule of thumb
Virtually every operational process in your organisation can be improved by at least 5% through the systematic application of data, AI and automation. Use that as your floor when sizing potential.
Improve customer acquisition
Support up- and cross-selling
Loyalize / prevent churn
Reduce waste
Reduce cost to serve
Subtotal
Define use cases that combine data, intelligence, and action.
A use case is the smallest unit that produces a measurable business outcome. Avaus defines each one through three ingredients: the data that fuels it, the algorithms that decide, and the actions they trigger. Use cases without all three either lack inputs, lack decisioning, or never reach the customer.
The leap that matters
Going from rules-based to algorithmic decisioning is what creates the largest efficiency gains. Adding self-learning to the model means it gets better the more it runs.
The Avaus best-practice methodology
Refined across 15+ years of commercial AI work, we define every use case through three building blocks - [Data] is the fuel, [Algo] is the decisioning, [Action] is what reaches the customer or process.
01
[ Data ]
The fuel
All companies have more data available for data-driven action than they think - and need to collect more to stay competitive. A modern data strategy covers structured and unstructured signals alike.
02
[ Algo ]
The decisioning
Algorithms define what we want the data to do - from simple rules to self-learning models. The leap from rules-based to algorithmic decisioning is where the largest efficiency gains live.
03
[ Action ]
The output
The processes and customer-facing activities you automate. Low-hanging fruit sits in marketing and sales, but the model extends to pricing, supply, inventory and service.
Only by combining all three do you turn data into results. Read more on avaus.com/data-algo-action.
Define your target automation level.
Decide which value drivers and channels you want to automate 3-5 years out. The target automation level (TAL) becomes the north star: it determines the size of the portfolio, the pace of delivery, and the resources required. Without a TAL, prioritisation becomes a yearly debate instead of a multi-year programme.
Three sizing questions
How complex is your customer journey and service portfolio? How many channels will you activate data in? How many business lines or markets do you serve? Multiplying these three gives you the rough size of the portfolio.
Typical year-one rollout
Q1
Establish clear view
- Business case
- Use case list
- Annual targets
Q2
Initiate program
- First use cases
- Critical capabilities
- Establish program
Q3
First results
- Scaling use cases
- Reporting
- Change mgmt
Q4
Functioning op model
- Op model health
- Agentic processes
- Optimisation
Build the capabilities your program needs.
Once you know what you want to do and at what pace, the resource picture becomes concrete. Translate ambition into a realistic operational plan: which capabilities do you build internally, which do you buy, and what data foundation supports the portfolio you have committed to.
Where teams underestimate
Most organisations underestimate the data work and overestimate the modelling work. Getting the right data in the right place at the right quality is consistently the longest lead-time item. And remember - everything is data in the world of generative AI.
Four capability pillars
Data foundations
The pipelines, quality and governance that make trustworthy data available where decisions are made.
Technology
The platforms, models and infrastructure that turn data into automated decisions and actions at scale.
People & skills
The mix of business, data, engineering and AI talent needed to design, build and operate the portfolio.
Ways of working
The processes, rituals and decision rights that let teams ship use cases together, quarter after quarter.
Build momentum with each use case you deliver.
Value does not arrive in one big bang. You build it by shipping use case after use case into production, each one stacking on top of the last. Impact compounds: every quarter adds to the base, so a steady cadence of new use cases quickly turns into a portfolio generating measurable business value.
Why cadence beats size
One large use case is fragile. A growing portfolio of smaller ones is resilient: every quarter adds to the base, and by Q4 of year one the cumulative impact is already material.
Q1
Q2
Q3
Q4
Each use case added compounds the value of all previous ones. By Q4 of year one, the portfolio is already generating measurable incremental revenue.
Design principle behind the methodology
"The hardest thing in commercial AI isn't the technology. It's consistency. 2% better every month compounds to 2x in three years - simple math, but surprisingly rare in practice."
Build the operating model that makes it stick.
Based on 15+ years of practice, Avaus has codified a best-practice operating model framework for data-driven growth: 40 elements across 8 dimensions, from targets and strategies through to change management. Technology alone does not make companies data-driven. The operating model does.
The core Avaus finding
The primary blocker to AI-driven growth is almost never the technology. It is the operating model: unclear ownership, misaligned incentives, and processes that were not designed with data and automation in mind.
Why the operating model matters
Operating
model
Compounding
value
Use case
factory
Use cases ship value once. The operating model makes that value compound, quarter after quarter.
The framework in full
40 elements. 8 dimensions. One operating model.
The full Avaus best-practice framework for data-driven growth - what needs to be in place to scale data and AI ways of working over time.
Operating model framework
A Targets Explicit statements of what to accomplish | Ambition | Revenue | Profitability | CX | Automation |
B Strategies Key choices that guide development | Channels | Relevance | Use cases | Data | Scaling |
C Structures How to collaborate and make decisions | Forums | Cadence | Roles | Decision rights | Organization |
D Implementation Deployment and development processes | Development | QA | Documentation | Scaling | Maintenance |
E Resourcing and investments Enablers for use case implementation | Infra | Data | AI / ML | Content | Agentic AI |
F Governance Key areas in need of steering | Program | Partners | Data | AI | Security |
G Measurement What to monitor to ensure value creation | Automation | Maturity | UC health | Performance | Risks |
H Change mgmt Enablers of people & organisational change | Sponsorship | Plan | Communication | Incentives | Upskilling |
40 elements across 8 dimensions. The operating model is the difference between use cases that ship once and use cases that compound.
The final kicker
Then hand the work off to agents.
The reason you invest in a structured, well-defined operating model is not just rigour for its own sake. It is what makes the next step possible: as your processes become explicit and repeatable, you can increasingly hand off work to agents and AI-powered workflows.
The same commercial output starts to require significantly less human effort, and the agents belong to you. That is what further accelerates value capture, year after year.
Y1
Y2
Y3
Y4
Next step
Ready to find your value drivers and plan your commercial AI program?
The methodology works best when it starts with a conversation about your situation: which value drivers matter most, where you currently sit, and what a realistic 12-24 month ambition looks like. That is where every Avaus engagement begins.