TCO: cutting hidden costs, 3 pitfalls to be avoided

Purchasing any Analytics solution (as is the case for any other domain) is an act of management, which will significantly impact your results, requiring concentration, precision and the ability to project.

The different people involved in the negotiation process are buyers and salespeople.

• What is important to salespeople is to able to lead a potential buyer to adopt their way of thinking in order to get their argument across, based on their strong points. A salesperson will have a governing principle which will always allow them to come back to the beaten track, laid out by their marketing service or VP Sales.
• However, what is important to a buyer is actually the reverse. Buyers want to take salespeople off the beaten track. To do so buyers need to present the salesperson with a subject that they are comfortable with and which does not fit into the salesperson’s way of thinking. Buyers therefore need to make sure that they take the initiative as it is they, and only they, who will define the path to be taken in the future.
• Whenever purchasing any solution it is a potential buyer’s needs and not the actual solution which must be considered first: therefore the first step is to accurately define the need.
• The different needs must be accurately expressed in functional terms and in terms of qualitative and quantitative criteria (presence, limitations, internal and external explicit and related costs).

What is TCO?

TCO, the “Total Cost of Ownership” is the advantage of studying not only the initial purchase but also the overall cost on the product (or service) life cycle.

In addition to the acquisition costs, it also takes into consideration direct and indirect function costs, training costs, upgrade/update costs, maintenance/administration costs and support/assistance costs.

For example, a car salesperson who explains the difference between purchasing a car, leasing a car with intent to purchase and a long-term lease is a first step towards TCO, taking into consideration maintenance, assistance and insurance.

Add to this the consumption of petrol, parking, washing, and possible reductions associated with car-sharing with your neighbour, the estimated value of the vehicle when selling it, and you are starting to get closer to TCO …

Calculating the TCO of an Online Analytics solution

Calculating TCO involves studying the solution much more than just reading the technical specifications. It forces us to evaluate the solution in more qualitative terms, and requires that costs associated with its use over time, and the impact that it has on related tasks, be taken into consideration.

In the field of IT a TCO forecast study generally covers a period of three years.

Complicated? No, I would rather use the term rigorous when it comes to defining needs (present and projected) and the list of the costs associated with meeting them.

Accurately defining needs at the beginning

It is therefore very important to couple a TCO calculation with an accurate functional analysis. A simple yes/no answer is not sufficient as it will hide significant differences which will be translated by additional, sometimes very significant costs and/or gaps, which are to be made up (segmentation, real time etc.).

Let’s now take a look at the notion of cost.

We often tend to only consider the total costs invoiced by service providers. If it is a significant part of the cost, we will also see that such costs can prevent us from seeing the bigger picture.

Main pitfalls revealed by TCO (and tips on how to resolve them)

Pitfall number 1: considerable gaps between the invoicing method used and the basis used for calculating costs

At first, the different methods used for invoicing appear to be simple and uniform: Service providers invoice server calls according to a price list that is drawn up based on quantitative levels.

As a result, the comparison between the two is generally limited to comparing the prices for traffic volumes from the same cohort: for example for 10 million server calls, service provider X will invoice £X, whereas service provider Y will invoice £Y. Does this mean that all we need to do is choose the service provider which charges the lowest?

Big mistake: even though this might all seem clear, depending on the solution used, there is a significant number of differences in the number of server calls counted for the same piece of content.

• Simply displaying the home page may be worth 5 times more server calls (SC) according to the solution used.
• As far as simple transactions are concerned, some solutions will count the transaction as being 1 SC, whereas others will count it as 1 SC + the number of articles contained in the basket (for example a transaction with 4 articles in the basket will be counted as 1+4 = 5 SCs).
• Another source of the differences that exist are the different “events” taking place: the most frequent case (and most numerous) relates to clicks. Almost all of the different solutions available count 1 click as being 1 SC, whereas AT Internet counts 1 click as being 0.5 SCs.
• What is even more important are the advertising displays, emails opened, updates of RSS feeds, podcasts etc., which AT Internet counts as being 0.1 SCs.

Comparing price grids is far from being sufficient, because according to the protocol used to calculate TCO, price grids which are said to be identical reveal some very significant differences.

For example:

Let’s consider the following simple sequence: An email campaign which has been opened 20,000 times generates 1,000 visits on page A which contains 4 web banners. 460 visits go on to page B which has 3 web banners, and 320 onto page C which also has 3 web banners. 220 leave the site by clicking on ads.

Below is a detailed breakdown comparing Sever Calls between AT Internet and its competitors:

On one hand the data shows the distribution of the different actions (actual server calls) and on the other, how much AT Internet or its competitors will invoice you.

In our example, our competitors will invoice 30,900 server calls whereas AT Internet will only invoice 5,628. We can easily imagine the difference between the two figures whenever a complete website is considered. All types of sites are affected with an even greater difference if objects such as Rich Media, Video, Podcasts, RSS feeds etc., are measured.

Pitfall number 2: the hidden costs associated with the duration and costs of implementing the solution

Some solutions are well known (if not legendary) for taking several months for the initial implementation of the tags.

We can only but gasp when we think about the actual amount of time wasted: calculating TCO will restrict you not only to calculating how much the implementation deadline will cost you in terms of the service provided (comparing the number of man days required and the cost for each solution) but also in terms of the number of your own man days required (technicians and project managers required to work on the project) preventing your employees from working on more productive tasks. When all of these different factors are considered the costs will begin to add up very quickly.

Pitfall number 3: support, assistance and Service Level Agreement (SLA)?

TCO, which allows us to calculate the real cost of the customer support for each service provider, points the finger at the continuity (or discontinuity) of the service.

The question on the cost of the support (it is free for AT Internet) invites us to further investigate the qualitative aspect: who is involved? Is the support used internal or outsourced? How available is the support team? What are the response/intervention times etc.? The same can also be said for the support provided: in addition to resolving bugs, is any real assistance provided?

The costs involved will also depend on these qualitative factors: your results will be affected by deadlines for resolving problems, and in some cases the time required to respond to simple questions may also disrupt your time management. Everyone knows what could cost the mobilisation of a team, which should concentrate on improving profits rather than on resolving problems: this shortfall is also a type of cost.

The final point to be studied is the closeness between the support centre and the product teams. An internal support team will be in contact with the marketing/technical/development teams and will provide precious information, whereas any outsourced support will not share any common goals and will remain focused on their own business.

According to Forrester Research AT Internet has “the highest level of satisfaction with technical support responsiveness, knowledge of support staff and ability to resolve issues”

In more general terms SLAs, should they exist, can reveal significant differences. If a level of guarantee for data collection is generally at 99%, guarantees relating to the availability of reports, data reports in particular, real time and the responsiveness of the support teams are more differentiated. Some SLAs are free, whilst others are fee-paying.

Other potential costs

Amongst all of the unpleasant additional costs associated with the functional limits, I have listed some of the ones that I feel are undesirable. Please feel free to add to the list:

• additional URLs
• additional users
• additional segments
• additional reports
• additional API requests
• SLAs
• improving the accuracy of real time (read the article Real time: gadget or ROI maker?)
• limiting/cancelling sampling (read the article The myth of virtuous sampling).
• Etc.

When considered all together these costs represent very substantial amounts.

Conclusion

Despite the fact that there are many hidden costs it is, after all, quite easy to track them as long as we concentrate on the real, specific needs at hand.
TCO is a precious aid, it is one of the key elements involved in the general process for choosing an Analytics solution.