# AI Priorities for the New Year: The Questions Every Executive Team Should Settle in January

Author: Marin Niehues, Niehues Consulting · Published: 2023-12-25 · AI Strategy & Business Case
Source: https://www.niehuesconsulting.de/en/blog/ai-priorities-2024/

An executive team sets its AI priorities by settling which business outcome should change, which initiatives start for it, who owns them and how success is measured. Everything else follows from these decisions, the budget included.

This article appears between the holidays and is meant for the first working day in January. It poses eight questions that an executive team and department heads can answer in half a day, arranges the answers in a quarterly grid and shows the criteria by which you decide between buying and building. This article provides an overview and does not constitute legal advice.

**AI priorities are decisions about business outcomes, about owners and about capacity. The choice of tools follows from those decisions.**

## Where do companies stand at the turn of the year?

At the turn of the year around one in eight German companies uses AI, and the distribution is uneven. The Federal Statistical Office reported in November 2023 a share of 12 per cent among companies with ten or more employees, with 10 per cent among small and 35 per cent among large companies with 250 or more employees. Companies without AI named a lack of know-how as the most frequent obstacle.

The documented effect concentrates on single work steps. In a field study of 5,179 customer support agents, the number of issues resolved per hour rose on average by 14 per cent with an AI assistant, and by 34 per cent for less experienced staff. An experiment with 453 professionals, published in Science, found writing tasks completed 40 per cent faster. How these findings should be read is set out in the [AI review of 2023](/en/blog/ai-review-2023/).

For your planning this gives a starting position with two sides. The benefit of single applications is documented, and pressure from shareholders and competitors is growing. At the same time many companies lack the ability to bring applications into operation. How to handle the expectations of owners is described in the article [AI pressure from shareholders](/en/blog/shareholder-pressure-ai/).

## Which questions should the executive team settle in January?

The executive team should settle eight questions in January that together make up the year's AI plan. Each question has a criterion by which you can tell whether it has been answered.

**Eight questions for AI priorities**
Question | Answered when | Without an answer you risk | With Niehues Consulting
1 Which business outcome should change? | A metric with baseline and target exists | Tools without a goal | Baseline and success measure fixed beforehand
2 Which initiatives start first? | One to three initiatives are named and ranked | Many pilots without operation | Prioritisation by value and feasibility
3 Who owns each initiative? | An owner in the business unit is named | Initiatives without a decision-maker | Decision-makers and delivery owners clarified
4 Are data and knowledge usable? | Source and quality have been checked | A prototype on sample data | Data check before the prototype
5 Who runs the solution? | Responsibility for support and upkeep exists | Standstill after launch | Operations owned from day one
6 Buy, configure or build? | Decision documented against criteria | Duplicate development or dependency | Selection and sequencing of applications
7 Which rules apply to employees? | Usage rules for AI tools are published | Use without any steering | Roles and decision paths in the workflow
8 What may it cost, and when does it show effect? | Budget and the metric after 90 days are fixed | One-off projects without proof of impact | Business case confirmed with real values

The first three questions come first, because they fix goal, sequence and accountability. Questions four to seven clarify whether the applications can arrive in daily work. Question eight links the plan to the company budget. If three or more answers are missing, the plan is not yet ready for a budget release.

## How many AI initiatives should a company start in its first year?

A company should start one to three initiatives in its first year and begin the next only when the first runs in steady-state operation. Each initiative ties up attention in the business unit and capacity among the few data and development specialists that a mid-sized company has. The Federal Statistical Office survey names lack of know-how, at 72 per cent, as the most frequent reason for non-use, which confirms the bottleneck at this point.

How the pressure to do a lot now can be turned into a robust plan is described in the article [AI activism](/en/blog/ai-activism-board-pressure-into-plan/). The capacity calculation behind it is simple: count the available hours of specialists and business units per quarter and deduct daily business. What remains determines how many initiatives are possible at the same time.

## What does a quarterly grid for AI look like?

A quarterly grid places each initiative in four stages with one executive decision each. The grid follows the phases of understanding, building and rolling out and can be adapted to the size of your company.

**Quarterly grid for an AI initiative**
Stage | Result | Executive decision
First quarter: understand | Metric, baseline, use cases assessed, owner named | Which initiatives start?
Second quarter: build | Prototype in the real workflow with real data, business case with real values | Does the initiative go into operation or end here?
Third quarter: roll out | Operations owned, users enabled, processes adapted | Is the solution extended to further areas?
Fourth quarter: impact | Metric measured against baseline | Which initiatives follow next year?

The grid treats operation as a step of its own with its own decision. An initiative that ends in the second quarter has served the purpose of the grid, because it tested and ruled out the economics before the rollout.

## Buy, configure or build?

For most use cases, buying or configuring a market solution is enough in the first year. Building pays off only when the application creates a competitive advantage or no market solution meets the requirement. The availability of AI functions in existing software, for instance with Microsoft 365 Copilot since November 2023, widens the options for buying. You must still plan rollout and benefit. The model race is secondary for this decision, as the article [Gemini versus GPT-4](/en/blog/gemini-vs-gpt-4/) shows.

Three criteria help with the choice. The first is uniqueness: the more an application sets your business apart from competitors, the more building is worth considering. The second is the data position: the more sensitive and proprietary the data, the more closely you check where it flows. The third is operability: without a team for upkeep and further development, a self-built application becomes a burden.

## Which rules do employees need?

Employees need a short, understandable usage rule that sets out which tools are permitted and which information does not belong in public AI services. Such a rule makes usage visible and lets the executive team know where AI is used. Legal questions in individual cases are for qualified lawyers.

## AI priorities with Niehues Consulting

[Niehues Consulting](/en/) is a consultancy for AI management and organisational design from a delivery perspective, based in Düsseldorf. Founder Marin Niehues brings AI into operations, in projects from VC-funded scale-ups to DAX 40 groups and in more than 20 countries on six continents. In the service area of AI management, the work runs from business analysis and requirements through the business case to the selection and sequencing of applications and the steering of the portfolio.

The approach follows the Niehues Delivery Model with its three phases. Three test questions matter for the priorities of the year: Which business outcome should change measurably? Who decides, who delivers, who uses? Who is accountable for operations from day one? Scope and terms of an engagement are discussed in the free initial consultation, which you can arrange in 30 minutes through the [contact form](/en/#kontakt).

**Conclusion**
AI priorities for the new year consist of a few decisions: one business outcome, one to three initiatives, one owner per initiative and a metric for the first 90 days. Settle the eight questions in January, place the initiatives in the quarterly grid and release the budget once at least six answers are in hand.

**Sources**
- Statistisches Bundesamt (Federal Statistical Office): Etwa jedes achte Unternehmen nutzt künstliche Intelligenz, press release, 27 November 2023
- Brynjolfsson, Li, Raymond: Generative AI at Work, NBER Working Paper 31161, 2023
- Noy, Zhang: Experimental evidence on the productivity effects of generative artificial intelligence, Science, 2023

## Frequently asked questions

### What AI priorities should a company set?

A company should set priorities on applications that change a measurable metric and have an owner in the business unit. Start with one to three initiatives and rate them by value and feasibility. Everything else follows once the first initiative runs in steady-state operation.

### How do I plan AI in a company for next year?

Plan with a quarterly grid: understand in the first quarter and build in the second, roll out in the third and measure impact in the fourth. Each stage ends with an executive decision. The eight questions in this article provide the basis.

### How much budget does AI need in a company?

The budget follows from the number of initiatives, the capacity required and the running costs. For each initiative, count licences or development together with internal business-unit time and ongoing operation. Scope and terms of a consulting engagement are discussed in the free initial consultation.

### Should a company buy AI or build it?

For most use cases, buying or configuring a market solution is enough at first. Building pays off when the application creates a competitive advantage and a team can take over operations. The data position weighs in as well.

### How many AI projects should a mid-sized company run at once?

A mid-sized company should run one to three AI projects at once. Specialists for data and development, and the attention of business units, are scarce. More projects spread that capacity without any of them reaching steady-state operation.

### Who on the executive team should own AI?

Responsibility for AI should sit with one member of the executive team who decides goals and budget. Below that, each initiative needs an owner in the business unit. Responsibility without decision authority leads to delays.

### How do I know whether we are ready for AI?

You are ready when at least six of the eight questions in this article are answered. Goal, owner and operation matter most. If they are missing, begin with those questions, and the choice of tool follows afterwards.

### Which metrics show that an AI initiative works?

The metrics that work are those that measure a business outcome, for example lead time, cost per transaction or quality. Fix the baseline before the start and measure after 90 days in steady-state operation. A metric without a baseline proves no change.