Network costing for builders: getting it right with automation - interview with Alain Meller


In this interview, we showcase the benefits of using automation in defining network cost, this is inspired by real cases of Setics Sttar software used by our clients, mainly operators and contractors (called « Builders » later in the document). This interview focuses on the difficulty of defining the price of an FTTH network construction and the impact of Setics Sttar on this matter. In economic terms, it could be termed a “price formation” problem.

This article doesn't showcase the Setics Sttar software by itself. The reader is welcome to consult the product’s information given by Setics Group or visit the website.

NN: Let’s start this interview by discussing the landscape. In your opinion, what are the changes in fiber optics and some of its challenges?


Initially perceived as an auxiliary infrastructure, FTTx significance has undergone a profound transformation over the years. What was once considered merely an ‘accessory” has now evolved into a vital component of government national plans, elevated to the status of a fundamental necessity alongside water and electricity.

Fiber to the home now seemsto be the most appropriate vector to sustain the exploding need for a high-quality, high-broadband connectivity. However, to connect every housing and every company by deploying a new network is expensive.

To construct a network, first of all, a price must be « fair » to ensure for all of the contractual parties that not only their economic risks are reasonably managed, but also that they can be formed in a « fast and low-overhead » way in order to be efficient and lead to an industrialization of the deployment.

When an operator or an FTTH infrastructure owner requests a Builder to build an FTTH network, the problem of determining the cost/price arises, among other things, as a subject matter for the contracting between both parties. Because of the financial challenges (hundreds of millions of euro for such projects), it seems obvious that a few percent of error can translate into massive absolute numbers.

First, let’s consider that the most preferred way of estimating a price is by the cost by Home Passed (HP). We often see debates about the cost of HP among network construction actors, and this debate seems to be central to the relation between the implied parties.

This debate is also of utmost importance in the context of the volume growth of deployments knowing that with growing volumes, approaches must be industrialized.

NN: Discussing in more detail the cost estimation, is it the same between dense and less dense areas? And how does the complexity arise?


Historically, the fiber optic network deployment has concerned densely populated urban areas and projects of a limited scope.

  • In densely populated areas like urban areas, the frequency of fulfilled road works and the good knowledge of infrastructures make it so that those works’ risks are broadly under control by the field actors. With the experience of time, the involved parties have converged to ratios and prices that allow a contracting that warrants a good risk control for the Operator and a fair revenue for the suppliers.
  • In the case of projects of a limited scope, a prior fairly detailed study is doable in a short period of time. A competition can be organized between the suppliers via the issue of an RFP. The pre-sale study costs incurred by the competitors are then integrated into the construction price of the successful deals.

Outside of the dense areas: complexity rises

The matter is completely different in the cases outside of the above-mentioned situations: rural or moderately populated areas and in the context of the growth of the volume of deployments.

The object we seek to find the price is of great complexity and is potentially submitted to various and significant uncertainties.

An access network that possesses a great capillarity is complex to conceive, the possibilities for finding pathways are multiple, and in the end, it will bear the inevitable hazards coming from an incomplete knowledge of the field situation as well as the right-of-way negotiations that will have to take place in the future.

It would be too long and too costly to obtain the entirety of the information necessary for a perfect awareness of the situation and therefore reach a precise estimation of the costs and consequently of the price.

We notice retrospectively that cost by HP greatly varies from one area to another and that the gap cost between HPs and the average cost in one area can drastically change from one area to another. As a matter of fact, it is common that 5% of the HPs constitute more than 20% of a project’s global cost.

NN: What are the various approaches for setting the price?


On a situational analysis of the methods used by the Operators to negotiate the prices, three main trends can be combined and, of course, are by default applied to these new areas: the call for tenders, the prices catalogs and the HP target cost.

1. The call for tenders

In the case in which the Operator establishes a competition between the respective Builders on the areas where he plans to deploy a network to obtain a fixed-price commitment.

In this approach, the Builders deem that the cost of the response for the call for tenders will be covered by the price on projects were they are successful. It simply is the most classical way to deal with the projects. Because the studies are more expensive in these areas, it happens that the Operator compensates the suppliers who did not win the deal for a fraction of their expenses.

This approach raises many questions:

  1. It requires time and is costly for the Operator and for the Builders. For every area, the Operator must conduct a call for tenders, analyze the responses, contractualize…
  2. On their side, the Builders must mobilize rare resources to respond to those calls for tenders. A significant effect is that the Builders can’t respond to all of the calls for tenders and therefore this approach based upon competition isn’t effective anymore.
  3. The early studies are not valued as such and are taken as pre-sale activities. Hence, the Builder will have a limited trust in the studies he will have produced and will therefore inflate the prices to prepare for any risks...

In the end, it is impossible to face project volume increase and industrialize this approach: the studies become a real bottleneck. Here, an economic price inefficiency will be the result, mainly because of the risk perception among Builders and the weak level of competition.

2. The price catalogs

In this fairly traditional system in the world of construction, the Operator and the Builders agree upon the unitary prices of a catalog of typical tasks; for instance, the cost of digging a trench per linear meter, the junction cost from an OLT to an ONT etc. and will contractualize around those catalogs.

The counterpart to those commitments will be, for the Operator, to commit to a volume of pre-attributed works for every Builder.

In this approach, the Builders don’t commit to quantities. The followed process is then generally the following:

  1. The Operator decides to launch the realization of an area that is pre-attributed to a Builder,
  2. The Builder proceeds to make the early studies and provides them to the Operator which will validate the engineering schemes and the studies’ estimated quantities. The cost of studies has been generally integrated into the prices beforehand.
  3. The Operator validates the project and the construction may start.

The difficulties of this approach are that:

  1. The Operator doesn’t get enough time and resources to control the studies and will make use of partial surveys to check if the engineering rules are properly applied,
  2. The Builders, knowing that the work is already attributed to them and that they don’t commit to a fixed price globally, don’t have any incentives to make more reliable quantities numbers given through the studies.

In the end, in this approach, the Operator is the one taking an inflation price risk due to the construction’s risks and the Builder is slightly motivated to find optimal solutions to minimize costs. 

3. The HP target cost

This approach is a direct transposition from the densely populated areas where, as we already mentioned, the knowledge of the costsis better known. In this approach, the Operator enforces a lump-sum price to the Builder based on ratios supposed to be typical of the deployment zone and that are recorded on abacus table over time. For instance, the population density, the collective habitation versus the residential areas ratio, the second homes rate etc… can be taken as ratios.

In practice, this approach has the following weaknesses:

  1. As we said earlier, the variation in area types makes the abacus tables not very reliable. They are the result of a fastidious work, and need a long time to learn from the realized cases.
  2. This approach tends to considerably tense the relationships between the Operators and the suppliers, who perceive the price as arbitrary; the model is therefore often relaxed, for instance by excluding the civil engineering cost from the HP target cost. 
  3. Strictly applied, this way of proceeding can lead to flaws quality-wide: the Builders under pressure, can on some territories experience over-costs, that they may manage by lowering the quality of the final network infrastructure.

Besides putting most of the risks upon Builders, this approach contains great deals of risks for the Operators on the final quality of the work, and eventually on the price of some of the works that are excluded from the target HP cost.

NN: With the existence of different approaches, what is the core of the problem?


The difficulty to estimate the price comes from the complexity and the hazards the network construction is subjected to: in that sense, the risk perception of the parties can only lead to an important price gap between them.

When stepping back, the core problem to improving the estimation of the price and allowing a price gap reduction between the parties is clear, it comes from the capacity to produce early studies that:

  • are doable in a short period of time with a limited rare resources consumption, in order to limit the costs,
  • are reliable from the perspective of engineering rules compliance,
  • correctly reflects the potential of savings through the usage of already existing infrastructures,
  • are typical to the particular territory at hand,
  • are precise when it comes to cable quantity and estimated equipment,
  • are reproducible in a consistent way, and especially don’t rely on the particular human agent that did the study, in order to get an all-around trust between all of the parties implicated.

NN: So, what is the solution?


This is where software tools like Setics Sttar are needed. Indeed, one of the considerable benefits of this tool is that it makes possible the realization of early studies that respect the just previously listed criterion.

Of course, this is not done without fully understanding the requirement for the Operator to invest in early studies and not to consider them as included in construction costs. Indeed, the modeling quality done through a tool such as Setics Sttar depends on the quality and the quantity of the entry data. Gathering that information beforehand is a necessary cost that has to be taken by the Operator, directly or indirectly. Here the new techniques of field survey automation can further make the Design Automation even more effective with the capture of more precise data sooner in the process at reasonable speed and costs.

NN: How does design automation work to tackle this issue? And how to define which model is best to use?


First, the Operator will do with Setics Sttar a rough early study in an area and estimate a price that he could, if he wishes to, summarize as a per-HP price. Please note that thanks to this fast approach, he may be able to setup a smaller perimeter (by excluding the more expensive HPs for instance) and also a phasing that will allow him to optimize the deployment over time.

We will call Psttar(A) the estimated price on an area “A” given by Setics Sttar. Reasonably, everything can’t be predictable at thistime, and hazards may happen (unusable infrastructure, wrong endpoint numbers etc…). A percentage must then be introduced that will reflect those hazards: it will be called “H%”.

The Operator can therefore consider that the price P(A) of this area is: P(A) = (1 + H%) * Psttar(A).

This H% coefficient is of course calibrated over time as experience makes it possible to better capture its variability and magnitude1. For example, annually, the theoretical quantities and the effective final quantities are compared and the hazard coefficient is adjusted.

The Operator will then be able to offer a construction flat rate for the area to the Builders based on that price.

On their side, the Builders can do by themselves these same studies with Setics Sttar, in a short period of time. It is in fact beneficial to the Operator to share with the Builders the data used to do the modeling. Builders can then acquire a trust in the price proposed for the area. They can also, with their own knowledge of the field, do models with Setics Sttar with different parameters and sometimes even improve their bottom line by further optimizing the design.

They can also modify the data to do sensitivity analysis. What would happen to the cost if the aerial infrastructure wasn’t usable?

Lastly, a debate around the model resulting from the use of Setics Sttar can occur between the Operator and the Builder allowing a beforehand discussion of the potential challenges.

In the end, the contracting can be made through a good command of the risk taken by both sides, and therefore drastically reduce the initial gap between both parties.

  • The benefits of this model-based approach

The benefits of this approach based on a network model created with Setics Sttar are numerous. Price-wide, it can be fast established and be in accordance with the at-hand territory. Therefore, it is an approach that allows to industrialize the deployments without jeopardizing the network.

This approach can reduce the time and the investment needed for reaching a good level of trust in the control of the construction risks, by the involved parties. Consequently, it can bring anew the usage of competition for the price, as the Builders have access to better information from the Operator that allows them to use Setics Sttar to

  1. produce a network model with their own vision of the project, and
  2. allow the Operator to evaluate the project offers, notably to avoid any drift quality-wide.

The approach also helps to:

  • improve the command of the infrastructure by the Operator: indeed, the Operator can internalize the knowledge of the network without having to heavily invest in resources dedicated to network studies, when today it relies mainly on the Builder to get that insight,
  • fluidize the relationships between the contracting parties by objectifying the discussion around the network model from Setics Sttar, rather than being stuck in the discussion about an HP target price, at a time, no party can possibly apprehend the reality of the field impact,
  • by providing practical elements from the modeling that allow a documentation of the technical aspects of the contract between the parties.

NN: Can you share your last word on this subject?


Thanks to the possibilities of an automatic FTTH network modeling with Setics Sttar, an efficient model to converge on the price and to contractualize between Operators and Builders has been successfully implemented.

This model allows the industrialization and the volume growth of network deployments and allows the involved parties to manage their economic risks for the projects, without compromising the quality of the work

In reality, the model may be more complex: a different coefficient of hazard can be applied to the various objects of the model (linear length of cable, number of splices, number of boxes etc...), the vertical part for buildings can be treated separately etc..

The editorial staff had no role in this post's creation.