RFP. ‘Request for proposal’. Throughout a long career in treasury operations, it seems I have been continuously involved in thinking about them, drafting them, discussing them, assessing them and scoring responses to them. I can barely think of a six-month period when I haven’t been involved in the process at some point. Some of them were for small services, some of them for major system purchases. I’ve written some, consultants have drafted others. They are ubiquitous.
Bankers and software houses have entire teams of people to reply to these things. After you send them off, you can almost hear the sound of analysts cutting and pasting from their standard responses into your templates. Sometimes they forget to change the previous client’s name.
The RFP process is usually regarded as a linear process, with clear milestones along the way. The steps usually go something like this:
There are many sub-steps in the process (for example, do you allow a Q&A phase for suppliers once the RFP has been issued?), but broadly speaking, that is how it goes. As these things eat up a lot of senior management time, they also tend to be thought of sequentially. Each phase isn’t considered until the previous one is finished. This often causes issues if the actual implementation date is a fixed point. That’s not to say there isn’t a plan in place – there will be – but the exact content of each phase won’t have been predetermined.
The most difficult phase in practice tends to be the scoring and assessment. That’s not to say that it will be the longest. Some of the RFPs I’ve been involved with are mighty documents in their own right as every possible consideration is crammed in. Paragraphs requesting a general description of various aspects of the product or service jostle side by side with simple yes/no questions (for example, does the payment service include domestic US dollar settlement?) and straightforward requests for information (for example, please advise cut-off times for all settlement currencies).
Often, everyone in the department will have a view (“Can you make sure you ask them…?”), although mostly people are backwards in coming forwards when asked to write one. It’s the same when it comes to assessing the replies (“I’ll just read one or two… Who are the favourites?”).
That leads to my number one rule: If you’re involved in an RFP process, be committed. You either read all the replies or none. You can’t compare offerings if you haven’t read all the data.
The important thing with scoring and assessment is to have a structure to the process so it is transparent how the answer has been arrived at. It helps greatly if this is kept in mind when drafting the RFP itself. Consider at each point how you are going to differentiate between the differing replies you are going to receive. At its most extreme, it is almost like writing an exam model answer.
The RFP process is usually regarded as a linear process, with clear milestones along the way
The structure should come directly from the RFP. To that end, the RFP needs to be split into five or six sections, for example, ‘supplier’, ‘functionality’, ‘disaster recovery’, ‘usability’, ‘price’. Each of these sections should be self-contained and you should have in mind their relative importance to you. You can do this simply by choosing an arbitrary number of points – for example, 50 – and dividing them up between the sections. So you might have five points for supplier, 15 points for functionality, five points for disaster recovery, 10 points for usability and 15 points for price.
Then you can look at each section individually and compare them with each other across each supplier. If the RFP is a small document, you can almost do this on the fly, ie by reading all the sections, placing them in the order of preference, and then allocating the scores afterwards. Thus, the most creditworthy supplier automatically gets five points, the least one and so on. It will often be the case in this area that the scores will be close together, with all the A-rated banks on three points and all the AA-rated banks on five points.
This approach enables you to look at sections in a lot of detail where necessary, without small, but important, items being buried. You will have more questions on functionality than disaster recovery, and so when differentiating the suppliers on functionality, you might end up with scoring the section out of 50 questions, which you then translate to a score out of 15 for the overall assessment. Disaster recovery might only have half a dozen questions. The important thing is to keep the balance between the overall sections correct so that what is important outweighs the merely useful.
The other things to consider are the ‘gateway questions’. There are some things you must have, or you can’t proceed, ie they are those things that are the ‘gateway’ to proceeding with the project. So they are important to identify. After all, it is no use to you to appoint someone with brilliant client support if their product doesn’t do what you want.
Gateway questions are digital, yes/no questions of the type discussed above. In addition to the question on domestic US dollar settlement, you might also have: Do they have a disaster recovery site outside the M25? Or does the system operate on Unix? These are questions you can’t score out of five or 10. The supplier either does it or they don’t. If they don’t pass your gateway questions for each section, then they can’t be scored and are eliminated. It’s no good the supplier having top-notch disaster recovery arrangements if they can only make payments in British sterling and you need all currencies.
You have to take care with the questions, however, to ensure that they are genuine gateway questions and aren’t just following company policy without giving thought to whether it is applicable.
The other thing to guard against is being taken in by the apparently objective nature of the whole process. If you simply score every question out of five, ignoring gateway issues and subject groupings, you will end up with a big number and almost certainly a clear winner. As every score is essentially subjective, however, adding them together may simply compound error upon error.
Having gone through this process, it should be clear who the favoured supplier is. The next question is how to combine your scores with those of other people involved in the process.
You can, of course, avoid that problem by getting everyone in a room together and agreeing the scores as you go through. But this can be time-consuming and also creates a kind of ‘groupthink’.
Just adding together everyone’s scores is a simple approach, but you need to be wary of distortions that arise from adding the scores of those who are ‘hard markers’ to those who are more easygoing. There are a number of ways to create a more level playing field, for example, by looking at how many sections in which each supplier was the top scorer, and taking the one with the most first places, but this then evens out the previous section weighting. Of course, you could be lucky and have everyone come to the same conclusion immediately. Stranger things have happened.
Graham Evans is an independent treasury operations consultant with Trebian Consulting.