Automation can be a force multiplier for localization departments and language supply companies. By automating a range of tasks to a particular “degree of automation” you can realize significant gains and enable your work processes and output product to scale. Nowadays you may already have automated systems in place upon which you want to run more automation (e.g. you have a TMS and a BMS and want to connect them) or your teams are being influenced by fast changing technology and AI platforms (such as Generative AI). However the fundamentals still apply to making the decision to automate and to what degree, we recommend running a return on investment (ROI) calculation every time.
When building your ROI case, we suggest there are three decisions to make:
- Can automation be useful for you in this specific scenario?
- Does the scope of the project mean it is a good automation candidate?
- When should the ROI analysis be completed?
Are you making these three decisions in each step of your automation journey? Read on and find out more how you can make the right and smart decisions around automation every time.
Decision 1: Is automation useful?
Before you start a journey with automation, you need to answer whether it is useful for the organization and whether your organization is ready to embrace it? This might seem like an odd question to ask, however as you will see it is important to make that assessment, and do it early on.
Automation can span from something that you can configure and test in 10 minutes (e.g. in your CRM, send an email when a new deal is opened), to something that really requires an investment of tens of thousands of dollars. ROI is calculated the same way in both cases, but over a different period of time, and with different certainty. The simplest automations are always offered as automation capabilities in applications you are already using – a checkbox or a project template capability in a TMS, a configuration in a business management system, or an automation tool in your CRM system. Anything beyond that requires hours or days of time to look at.
In the translation and localization industry, both at Language Service Providers (LSPs) and end customers, automation works best if you are able to influence customers (or colleagues) to be consistent. Very often, however, you need to start small, then show the results to the customers, and then you do more. We recommend introducing a first step of checking, which decides whether a process can be automated or not. For example, in a CRM, if you want to create a new deal (open a new deal card) when you receive an email with the word “Order” in its subject, if your customer does not put it there, or puts it in the body of the email, the deal will not be created. How can you then ensure you capture the fact that the order is actually being placed? Introducing a verification step for each email would take more time from you than not creating the deal automatically. You have some options (which broadly describe the types of automation we normally see):
- Create a way to trigger deal creation from your email client (partial automation),
- Manually check your mails from the customers with whom you have this agreement on a weekly or monthly basis (supervision of the process),
- You can go on doing this for eternity without notifying the client, and then the client will start forgetting to put the “order” word in the subject,
- Or you can notify the client each time you spot they did not comply with the request.
- Create an alternative way of ordering for the client, for example a website which creates the deal or sends an email that corresponds to a template (new process),
- Request from the customer that they use an email template, and have an auto-responder answer if the email template is sent but not filled out (auto-checking of process compliance).
By running through the options for full or partial automation, you can make the first decision and determine if automation would be useful in this scenario. Now you can look at the project scope and decision 2.
Decision 2: Does the scope of the project mean it is a good automation candidate?
Start small? Dream big?
Return on investment is always calculated over a period of time. Automation is best seen as opportunity cost.
Let’s compare two scenarios:
Scenario 1:
You created the deal creation automation and it helps you with some emails. You still review the emails, everything goes as before, but you saved some extra clicks. Maybe you spent 30 minutes setting this up, and it saves you 60 minutes every month. Every month, though, you need to spend 5 minutes to check if it still works – automation, especially simple automation, might break without you noticing. This will be taken into account in the example below.
ROI is calculated by subtracting the cost of the investment from its value delivered, then dividing this new number by the cost of the investment, and finally, multiplying it by 100.
This formula can be calculated at any moment in time, and the results can differ accordingly.
| Timeframe | Investment | Investment up to now | Return | Return up to now | ROI |
|---|---|---|---|---|---|
Month 1 |
30 minutes |
30 minutes |
60 minutes |
60 minutes |
100% |
Month 2 |
5 minutes for checking and tweaking |
35 minutes |
60 minutes |
120 minutes |
242% |
Month 3 |
5 minutes |
40 minutes |
60 minutes |
180 minutes |
350% |
Year 1 |
5 minutes each month |
85 minutes (30 + 11 x 5) |
60 minutes each month |
720 minutes |
747% |
This investment produces a return in one month. It removes frustration and repetitive work for one person. However, its benefits are hardly visible for a large organization. There is one important aspect though: how does this affect the work of other people? If you are a bottleneck, with this automation you still remain a bottleneck. If you are on vacation and nobody monitors your email, order delivery may be delayed, resulting in less deals from the customer. That can indeed affect the organization.
Scenario 2:
You are running a team of five people who are exclusively dealing with the intake of orders. Each of them has a monthly salary of 3,000 dollars. You invest 100,000 dollars and create a system that does everything automatically and only needs 5 hours of supervision monthly.
What’s the ROI? First, you only need one person to manage the automation. But what do you do with the other four? Financially speaking, on the company level the only return that is visible in your payroll would be if you disposed of those four people. However, from a departmental perspective, if you free up four people and transfer them to do another job, the order intake department saves the cost of four people and another department increases spending by taking onboard these four people.
Let’s say the system is operational after month 2, and you pay for it in month 1.
First, you can make two decisions:
- If it takes two months to deliver this, does the investment include the cost of people (5 x 3,000 dollars) for these two months? If you don’t, your ROI figure is going to be higher, so sometimes this is ignored to be able to demonstrate higher ROI.
- After the system is operational, is the human cost 3,000 euro per month for the supervisor, or only the cost of 5 hours of the work of this person? Again, if you only count the 5 hours of work, ROI is higher.
Depending on these decisions, ROI will already differ.
| Timeframe | Investment | Investment up to now | Return | Return up to now | ROI |
|---|---|---|---|---|---|
Month 2 |
100,000 $ + 5 people x 3,000 $ x 2 months |
130,000 $ |
0 |
0 |
negative |
Month 12 (Year 1) |
10 x 3,000 $ |
160,000 $ |
10 months x 3,000 $ x 4 people |
120,000 $ |
-25% |
Year 2 |
12 x 3,000 $ = 36,000 $ |
196,000 $ |
12 x 4 x 3,000 $ = 144,000 $ |
264,000 $ |
+34.6%
(264,000 $ – 196,000 $)/196,000 $ |
(If you do not calculate with the running staff costs and only factor in five hours of investment at 100 $ a month, ROI after two years is +78%, not +34.6%.)
As you can see ROI clearly depends on the overall timeline and the available initial investment. By running a relevant ROI analysis you have a picture which all stakeholders can review and decide to buy in or not.
Another interesting aspect is the internal rate of return, which is the timeline when the investment moves out of the loss zone.
initial investment + monthly operating cost * months < return per month * months
In our case the break-even is at 14.45 months with the example detailed in the table.
While the percentage values of the gains of this investment are significantly smaller than that of the first example, the actual monetary gain is much higher.
Now let’s look at some of the risks that our ROI model couldn’t take into account:
- The project will not finish on time. Return for those months remain either zero or lower than expected, affecting ROI.
- The project will not save as much as expected. Costs for those months increase.
- The ongoing costs of change or maintenance will be higher.
- The automation will become obsolete. If it becomes obsolete in year one because, for example, a change of systems from above, the ROI will remain negative.
- The mathematics are made up. In the first case, how can you prove if it really saved you 60 minutes a month? There is no empirical data for checking it – ROI is speculative. In the second case, if you really only have one person working on what used to be the work of five, that’s considerably more evidenced by the payroll.
The scope and scale of projects varies, it is important to do an ROI analysis for every project, ensuring the depth of the calculation matches the importance of solving the problem. Now let’s look at when to run an ROI analysis.
Decision 3: When to calculate ROI?
An ROI analysis is not needed for every case. ROI calculations should really be reserved for situations when you will make an investment. It costs money to do an ROI analysis.
When you are not making an investment, use your gut on the ROI feasibility.
You can’t buy what you don’t know. As mentioned, it can take time and effort to run an ROI calculation. However, you need to understand the alternatives and the risks. You should run an ROI calculation when you know to a decent degree what the set of solution options are.
To see the full range of potential automation options you will need to spend time and money up front. In many cases this step is skipped because decision makers default to a previous incumbent solution. Just because an integration exists off the shelf, does not mean it is the best option for your use case. An example is where TMS systems have a bespoke connectivity solution to CMS systems, and this is considered the solution option to go with.
In terms of effort, factor in at least 3-5 work days effort to develop an ROI case. This should happen at an early planning stage, as many times once concluded, it can change the whole shape of a program. So do it early, aim for accuracy and write down assumptions.
By running an analysis you pre-determine that the solution needs to fit within a preexisting toolset. Sometimes this means you miss an opportunity to use simpler tools to solve the problem.
Making these three decisions before you decide for or against automation should give you the right outcomes every time.
Can automation be useful for you in this specific scenario? If yes, check if the scope of the project is a good fit. Finally decide when the ROI should be completed, and how much you investment is acceptable to come to an informed decision. Sometimes you can evaluate options immediately, other times you need to spend six months researching before you understand the options that exist to solve your problem.
Are you making these decisions in each step of your automation journey?
