The Allure of Flat Rate Pricing

Moving away from a per-word translation financial model can increase quality while reducing operating costs.

For decades now most translation buyers have become used to paying for services by the word, plus some negotiated “overhead”. Though this continues to make good sense for some organizations, using it also means ongoing business friction, which can become crippling at scale.

In a flat rate model, you assign a fixed value to the variable (usually word count) that you and a language service provider agree to, that will be applied over a span of time. An immediate benefit is that it provides a predictable set of future financial transactions. It works best when you can scale across languages or over time. The same model applies well to QA and other services you may need to acquire. When applied across services, languages, and time, the results are magnified.

The Problem

Though it is common practice to pay a content developer by the hour, it is not common for most organizations to do so when paying content translators. Consistently, a huge amount of time is spent on per-word price setting for translation services. Per-word pricing is an abstraction for the time a translator will take to do a particular job. By doing business on this basis, we agree to several unfortunate things and encourage several other side-effects:

  • We agree to spend time, intellectual capital, and money in its development, and therefore agree to write content and code for its implementation into agreements and software products.
  • We agree that we will only find value in a translators work by the number of words they have affected, and therefore agree to devalue their work when we find perceived gaps in it’s quality.
  • We agree to have a transaction basis that is inconsistent with that of any other mature industry, and therefore to spend the effort of explaining it … over and over and over… for years.

Unfortunately none of these agreements and side effects line up with agreements that might have been made internally when the work was taken on: “I agree to give you the right French words for the upcoming release in France in exchange for some money” would be much more along the lines of reality. We do this because someone has agreed to deliver product to a market expecting a return from that market for the effort invested.

The Solution

Some organizations can be a lot more successful with funding sources, measurably improve customer experiences, and better incentivise language suppliers by implementing a fixed rate pricing model. Wait, a win for your customers, for your organization and for your language partners? Why isn’t everyone doing this!?

Everyone isn’t doing this mainly because it’s a big change – which itself costs time and money to do. Secondly, you need to be really specific in what you are asking your partners for, so you need to have good data. And thirdly, you need to have a good relationship with your language service partners to succeed in unlocking the real value this change can bring about.

When you need language services at a very large scale, the per-word cost becomes a less and less meaningful part of the equation when you need millions of words translated a year. When you are fielding requests across large organizations, the change in day-to-day translation volume makes it very difficult to predict near-term costs. Your price per word costs have the most dramatic impact on your overall costs of going global when you are in the mid-range as a language consumer, this is a great time to begin thinking on how you will scale further by reducing this burden.

I have had the opportunity to be financially responsible for the acquisition and ownership of a wide variety of services, organizations, budgets, and business decisions. Only in the language sector do people buy rice by the grain. When you need a boxcar or rice each month, you can’t spend the effort to count them individually. In translation reality, we know we can’t buy that 100-word PR blurb for ~22 euro, we will get caught up in minimum fees. Successful merchants learned long ago to sell rice by the sho or cubit, not by the grain.

Invariably you need to pay for the language services costs and have a discussion to get funds. Often buyers feel like their finance folks don’t understand the unique nature of how they transact for services. You could view this need for deep knowledge of globalization business practices as a strength, but as a requirement, each day it erodes your organization’s ability to grow more dramatically.

Assumptions and Measurement

Other organizations do not share the inability to predict their future operating expenses, having usable formulas for others to rely on. In this void, assumptions are made that become hurdles for the future. If I own a budget and intend to hit it I need to primarily make safe bets. For example, maybe you work in a silicon valley tech company and report into product or engineering leadership. Engineering leadership thinks you need to add a scrum team to hit product feature goals. You know how much a new team of 8 developers will likely cost, but can infer little about how this impacts language costs. If you make the assumption they add 20% to your product feature set and allocate 20% of the initial translation costs (of $100k) for the updates this new teams work would create, call it $20k, hire the team, and move on.

What are the odds the language work will run $20k? Likely worse than playing the dice in Las Vegas. More likely? 0 dollars were budgeted for the added work to go global.

What if we could predict that cost, with better than 95% accuracy, every time?

When I worked at a small IT products company, we had roughly 2 hardware-driven releases a quarter, and I had a simple formulae based on the number of actually unique hardware products to come up with total costs for my org per quarter. Development produced features steadily, we had implemented topic-based authoring and VDP, which had eliminated publishing and testing costs with our vendors, plus eliminated 3 weeks of time from the product release schedules.

So, my cost to support the bulk of development were fixed – my writers. I budgeted illustration costs by the height of each new box we would ship, and print fulfillment costs in the same way. It was simple, I could tell people what they could expect if we delivered the things that we really needed to each other. This gave me the dollars I needed when I needed them, as a result our content won awards, made our customers measurably happier, and the operations and services teams came to thank engineering for the work.

In a small environment, look at and track your fixed and variable costs to produce. These are the same categories your language costs will fall into. If you are able to equate these with results like decreased customer contacts in support, or increased customer contacts in marketing and sales, that can form the basis for assigning a financial value to your content, and the value of your team’s outcomes. If you are good at doing this in one language, growing to many will be a lot easier.

I moved to a large software company where I was regularly running ~40-60 campaigns and projects and needed to account for every penny spent and get the funding up front. At first, the simple question of “what will X cost” was really tough and the whole team suffered from the burden of balancing the funding and work. Eight people all managing that volume of work, all spending 40% of the time focused on the dollars and not the customer. Knowing who usually did what over a long term, enabled by some simple tracking and a data analyst on contract, enabled a trove of utility.

We reduced our overhead to less than 10% had a simple cost model that could readily be budgeted for by our customer teams, enabling us to be allocated the funds on a just in time basis. The funding came in at the front end for what was significant, and covered the costs of the 2 services we offered for ‘free’, while also enabling our supply chain partners to pre-book the revenue and the resources. As a result the real cycle time was reduced to 1/3 to 2/3 of the prior amount, and it all got done within 2% of budget each quarter.

However, quality and it’s cost were still significant problems. We had great processes, tech, and a great team; the problem remained deeper than that in the stack.

Many large organizations suffer from quality issues, and many spend more dollars on the cost of quality than on the cost of translation or content (or both). Consider how changing the incentives and how we pay for language services impacts the cost of quality. (I will do a future article going in to further depths on Quality and more of these synergies.)

When you abstract payment by word count, the most margin comes out of translating the most words in the least time. The most words is very far from the best words. If you have a large volume of words going out for French each month across a wide variety of asset types on typically no notice or predictability, your language partner is going to resource the work as best they can afford to. In an ad-hoc per word and project model, time is spent by each actor in the supply and administration chain counting the billable grains. They all want to be compensated for this of course, but this doesn’t directly accomplish your work: it’s enablement overhead that removes time and therefore quality. So work goes out to whomever is best suited and available at the lowest cost per grain.

Well, great translators are in great in demand and can pre-book work, so are much less likely to be doing ad-hoc work. These core business incentives drive toward a sub-par outcome that is not solved by people, process, or technology. The bottom line is when you need something really bad, that may be just what you get.

What if we changed the equation from where most of the actors in the translation supply chain are incented to just be done as soon as they can, having done the best job that they can to get the quality level needed?

Your next step might be to implement flat rates for language support. To do so, you need to know what you are likely to need, even when your internal partners have limited data available to you. This is easy when you have a large amount of data to work with regarding how you have operated in the past and how that was reflected in the profile of the translation projects worked on. By default, this is easy for any established MLV to do. They have both paid for and billed for this work.

Let’s use this hypothetical example: you and your language partner have collected data about volumes and timelines over the past 2 years of work. You are spending $100k/year on German translation in a mix of product UI, legal, and marketing content. Each project is issued, quoted, approved, executed and delivered and billed individually. You have a lot of lines on your invoices about file transfers, conversions, and a PM overhead you have negotiated (Buyers can feel nickel-and-dimed by the non-translation lines, which the service providers needed to cover costs incurred as well as make a margin off the ridiculously low price per word you have all agreed to.) Tallied up, these numbers have significance and create a huge gap in the costs for the words and the real costs on the invoices.

Chances are you have fallen into a trap, spending too much time counting grains of rice and how you can affect their cost, rather than counting smiles on your customers faces.

What if you simply agreed to 8k a month to give over the best German words for the next 6 months? You give them simple metrics about the expected word volume and frequency, and it’s expected quality. “I agree to send you 25% +/- 10% of last years volume over the next 3 months and you do the best job you can on it. We build a simple meter to see how are we on productivity target as a measurement metric, not a billing metric.” This makes the productivity part easy. Though there is opportunity to also use this as a billing metric – e.g. here in the US our electricity bill changes very little when we stay within our allocation, but can go up dramatically when we suddenly start to exceed it.

With all the time both your team and your partners teams will save on transaction overhead, the time becomes available to focus back on the more important part – the outcome. This is where you will have needed to clearly define expectations around timelines, quality, and, ideally, end customer impact. If you have baked in metrics with delivery time or defect penalties, 1. own your defects and 2. these penalties can be carried as credits against the next billing round. Balance these against the outcome goals, and be very mindful of pushing too many of your measurement metrics into billing metrics, or you will quickly begin to lose out on the benefits of a flat rate. The best scenarios support direct synergy between the core goals of both organizations being reflected in the billing model. Directly incented not to lose money, people fix problems.


Using this model allows language service suppliers to predict work and revenue. Enabling them to pre-book translators who can more consistently work on your materials, building up knowledge and experience that will continue to enhance their ability to deliver higher quality. (When given effective knowledge and feedback, and when available, tooling to help them reach the outcome goals.) PMs and other required resources can also be better allocated, or even afforded to work on projects, or to spend time in advance learning about a customer, their terminology, style, and being incented to do it rather than just rush through the words at hand. This persistent revenue planning also enables their teams to unlock more opportunities, thus growing the capabilities of your partner rather than simply utilizing their existing ones.

Go back to finance and say: It’s less work, flattens out my Operating Expense variability, and it’s cheaper overall. Because in reality you can get this done at the same cost each month, plus get a couple of points off by paying at the beginning of the interval, and reduce the number and complexity of our financial transactions!

In reality it should be all of those things and makes the model consistent with other common models. This makes finance buy-in more likely. Then, when you teach them that you can spend money consistently, you will find a welcome finance partner when your needs change.

This all creates a much clearer basis to expect benefits to the customer. You reduced your effort on activities other than defining and reaching the outcomes needed. Now you should be able to answer two questions to quantify some results:

  • What was the positive impact to customer balance-sheet interactions with teams such as sales and marketing?
  • What was the negative balance sheet impact from customer interactions such as service and support?

Now you have the full basis to calculate the cost to produce a good customer outcome. Coupled with Market Size and Potential Revenue data you have the basis for RoI calculation across any number of markets! Excellent work – go speak at a conference!

This shifts the payment basis to be customer based, but not yet customer focused. Truly unified alignment of goals to the shared customer could be in our all futures. It’s time to take a step.


At Spartan Software, we have both the linguistic and technical expertise you need to implement the right solutions, to learn more about our services, contact me at: