When you are at the beginning of your journey with AI, you may be wondering how much it will cost to implement it or how long it may take. Artificial intelligence (AI) has its – quite justified – fame of something that is expensive, time-consuming, and risky. Studies show that as many as 8 out of 10 AI implementations fall short of expectations and never deliver the value that was promised!

So is it really millions of dollars and years of development? Or a few thousand bucks and a few weeks of work? To be honest… it depends. Reliable estimates are only available when details on the project are known. There are some factors, however, that can impact the final cost of implementing AI technology in an FMCG company, and in this article, we will try to list them and show you how to calculate the price range that you can expect for this kind of project.

Mind that in this article, we focus specifically on the FMCG industry. It doesn’t mean that the insights we share cannot be applied to other industries. They can and they very often do. If you are, however, looking for some more general insights, go to this article and learn how much AI projects cost.

What factors can impact the price of AI?

1. Custom vs. off-the-shelf solution

The first question you may need to ask yourself is whether you need a custom solution or something off-the-shelf. If there is an app available in the market which does precisely what you need, there is no point in developing a new solution doing exactly the same. To be honest, it’s a rare scenario that applies rather to smaller organizations.

In most cases it will be just a mix of two: custom solution using off-the-shelf components. It is rare, and in many cases completely pointless, to build technology from scratch. Most new products use an off-the-shelf component at one stage or another and it applies to Machine Learning projects as well. And while relying 100% on the off-the-shelf solution may not be sufficient in case of most of the FMCG companies, using off-the-shelf components to build a custom solution may reduce the cost (and the time) of its building.

To give you a brief overview of the differences between choosing these two options, we estimated the cost of running 10 000 predictions using text detection, objects detection and localization (e.g. for the inventory):

  • Using Google Vision AI, we would pay $13.5 for text detection and $20.25 for objects detection and localization = $33.75 in total
  • Using AWS Rekognition, we would pay $10 for each option = $20 in total
  • Running similar custom predictions on Google Compute Engine using the standard virtual machine (N1) with the graphics card costs 250$-300$ (with no limits regarding number of predictions).

Of course, sometimes it is simply not possible to use an off-the-shelf component as it won’t provide you with the functionality you need. As long as it does, and this is definitely something that your AI development tech partner should be able to tell you, there is no point of reinventing the wheel.

How much does it cost to implement AI in an FMCG company

2. Requirements and team composition

Depending on your requirements, the team composition in an AI project may vary and this will definitely impact the total cost of AI adoption.

So what does the team in an AI project look like? (Let’s focus only on the core team on the agency side.)

Data scientist(s) / AI Engineer(s)

An absolute must in any data science project! Depending on the project scope, you may need more than one data scientist and their daily rates may vary from around $550 to $1100 – accordingly to their seniority levels.

Software engineer(s)

Do you need your AI models to integrate with some system you use? Or maybe it’s even necessary to build a new system because the old one is not able to support AI-powered solutions? Software development team will build your dashboards, interfaces, or whole apps to allow for easy use of the model. Again, depending on their seniority level, you may expect the cost of $400-700 per day for a software engineer.

Scrum Master

A person who makes sure the team’s work goes as smoothly as possible and facilitates communication between the dev team and your team. Depending on the team size, it’s the additional cost of $1200-4600 per month.

How to build successful products with AI?
Watch a free webinar

3. Inhouse or with an agency?

Do you want to build an in-house data science team or establish a tech partnership with an external agency? Both approaches have their pros and cons – you can read more about that in our article about building your data science team. For now, let’s not analyze those and focus only on the prices.

If you want to build an in-house team, you have to account for recruitment and training costs – which, reportedly, can be about 15 thousand dollars. Then, there is a salary which, according to Glassdoor, is on average 120K dollars yearly.

Agencies charge around $500-600 daily for an AI engineer / data scientist, and $1000-$1200 daily for a senior AI engineer and the price is impacted by the skills, experience, location, etc. Even though it may sound expensive, it may be a smart choice if you’re just beginning with AI and you are not sure whether it will work or not – and simply don’t want to hire people in these uncertain conditions. 

4. Data

If you have sufficient amounts of data that is clean and well-structured, it will definitely decrease the price of AI development. And what if you don’t? Data cleaning and preparation will take extra time of data scientists.

In some cases it is possible to use data available in the existing databases from the internet (open source or paid) to train the model and then apply this model on your data.

5. Scope

The larger AI project we think of, the more impossible it gets to provide an accurate estimate. Simply put, there are too many factors that may have an impact on the project. What does the project consist of? What’s the goal? What are the success and failure criteria? What data are you going to use? 

To estimate the project scope, you may use the help of an AI advisor or a data science partner who will also help you prioritize work and estimate given tasks in hours to be able to better evaluate how much a project will cost.

If you want to know, however, what price range you can expect for different types of project, you can safely assume that we’re talking about $10-20K for the Proof of Concept, which will help you validate the idea of using AI to solve the specific business problem, and $50-500K for the full implementation.

6. Other costs

Other costs related to implementing AI in an FMCG company (or any other industry, to be honest) include:

  • infrastructure

The models need to be fed with data and that data needs to be stored somewhere. The infrastructure costs may include, for example, cloud and data storage. 

  • integration (e.g. API development, documentation)

Even though the models are the core of this project, it won’t be usable if the results aren’t presented somewhere. Integrating them with the system that your team uses is a crucial part of the project.

  • maintenance

The AI models can’t be left unsupervised. They need continuous support, the new data has to be cleaned and annotated, and even if much of the recurring work is automated, it simply can’t be left all by itself.

7. Next steps

Building AI-powered models is rarely the end of the AI adoption process. In order to utilize them to their full potential, you will probably need to integrate them with your CRM (if we’re talking about using AI in sales), ERP or other systems.

The problem starts when it turns out that the system you are using is not compatible with AI. If that’s your case, you may need to think about building a new one.

But does it mean that if you want to use some other AI-based solution in 3 years, to deal with some other use case, you will need to build a new system again? No. If the new system is meant to be AI-ready, we don’t focus only on the temporary needs. During the Discovery Session, we try to identify all the places that have potential to be AI-powered in the future – regardless of the data you have right now or your current business processes, needs and priorities. Thanks to that, we are able to plan the whole architecture of that new system in a way that will make it possible to integrate other artificial intelligence models in the future. It doesn’t affect the functionality of the system itself but it makes it more “future-proof”.

Pricing an AI project in the FMCG

Ok. So we’ve spent almost 7 minutes taking a look at different factors that may impact an AI project’s cost and… it still comes to the most popular answer of all times: it depends. 

The costs of AI projects vary from a few thousand dollars for a Proof of Concept to a few hundred thousand dollars for the full implementation – and the reliable estimation needs to start with having a closer look at your needs, requirements, and possibilities to assess the complexity of the solution and the work that it requires.

The good thing is that you are able to start small, even if the estimated cost of the full implementation is half a million dollars, it doesn’t mean that you need to go for it “on spec”. That’s exactly what the Proof of Concept is for.

Starting with the PoC, you validate the idea of using AI for a specific business case. Project like that takes around one month, starts with the workshop which aims to identify where AI can be used, has the highest potential to succeed and should bring the highest ROI, continues with cleaning the data and using it with training a set of different models. 

That way, you can verify the feasibility of AI adoption in your company using real business cases and real data before you decide to commit to a long-term and costly project. After such a “test project” it is also easier to estimate the cost of full implementation as you are already aware of the problems with data that may arise.


We’ve started this article with a bit of scary statistics saying that as many as 8 out of 10 AI implementations fall short of expectations and never deliver the value that was promised.

As the study wasn’t fake, we need to remember that at the same time, 9 out of 10 senior executives and managers still see AI as a business opportunity.

9 out of 10 senior executives and managers from 29 industries in 97 countries see AI as a business opportunity
Source: ROI of AI by MIT & BCG

Even though it is quite a high-risk investment, you can mitigate this risk by approaching AI in a smart way. In other words: start small and validate your ideas before wading into deep water.

Keeping in mind that the full AI implementation in an FMCG company may take roughly between $50K and $500K, don’t put all that money on bet.

Or at least check what success probability that bet has.

Reducing churn by more than 20% with 10x ROI
Read case study