Several approaches may be used together, ie a multi-criteria analysis is required, to achieve the best outcome.
The overall aim is to obtain the greatest improvement in LOS for the greatest number of people (existing and potential users), and for these benefits to outweigh the money spent to achieve it.
The cycle review or LOS criteria could usefully be combined with cycling usage data and cross-compared with crash data and project costs. Together, these may indicate a programme focus on a particular geographical area, bringing forward other lower-ranked projects and removing identified blockages. This treatment could then be repeated for the next highest-ranked area, and so on.
This approach should not, however, neglect the value of some demonstrable achievement through implementing easy or cheap network elements or some quality flagship projects. Similarly, a focus on a particular area should not neglect particularly strong needs identified elsewhere.
During implementation, it may be useful to advance a lower-ranked cycling project and combine it with the timing of a mainstream project. This could be for strategic/political reasons, to make certain projects more ‘palatable’ to groups who may oppose them. It might also help retain a balance of equity, for example by giving certain users an advantage through one project where they will be disadvantaged or left out by to another. (See Infrastructure projects under Integration in Implementation).
The prioritisation process may reveal some factors that had been omitted in the evaluation process and it may be necessary to consider this feed-back and adapt the chosen route alignments or provisions. Some iterations may be necessary.
One should not get too hung up on the precise means of determining priority; this can lead to ‘paralysis by analysis’ rather than actual work to implement parts of the network. So long as a suitable method has been used to identify the routes to be included, any process of ordering them for implementation based on logic and common-sense should suffice.