A bit of both. Our first website release concentrated on the buy-to-let market alone, which is projected to grow in the coming years if more institutional landlords enter the market. This is good for the investment side of our business. However, we have taken a risk in investing a lot of effort in making our ratings more relevant to owners-occupiers, who felt under-served by Rankdesk initially. We will be launching this rating service in September.

Rankdesk isn’t a single product, is it? There are degrees of data depth available, at different costs. Is it possible to summarise the different services?

We offer two simple services:

One-off property ratings: A user simply submits a web link of any London property as it appears on a portal or an agent's website and our analysts will carry out a desktop analysis, testing the data provided against 60 quality and investment parameters. A four page rating is then sent to the user within 24 hours. This costs £70.

A £29 monthly subscription service for our property rankings: We've teamed up with some of London's most innovative estate agents. They provide us with their data so that we can rate their properties and publish the LONDON100 rankings every Wednesday morning.

I’m assuming there’s a certain amount of clever (and well-hidden) secret sauce mathematics involved, because – presumably – the raw data’s available to all of us?

We bring together estate agents' property data, land registry data, neighbourhood intelligence and rental and capital growth projections, which can all be gathered freely with a bit of spare time! We then test this data against housing quality standards (anything from room size, storage space to quality of local schools) and the rental and sales prices of comparable properties on the same street. All the answers hold a specific score and weighting, which is the secret sauce (a combination of buyer surveys, regression analysis that looks at how a property feature shapes the asking price and time on market of the property, and macro-level neighbourhood and city performance data). The algorithm is being perfected constantly.

Would you deny that there’s an element of 1) subjectivity 2) micro-geography to the market? For example, 1) Different “experts” (agents) will give vastly different valuations to the same property, and will take different views on whether, say, end-of-terrace properties are more or less desirable. 2) To take an obvious example, the HBOS index reflects with a reasonable amount of accuracy where house prices have been heading nationally, but is of minimal value to an individual vendor deciding how to price their property. How does the Rankdesk system deal with subjectivity and the micro-geographical nature of price trends?

Of course there is an element of subjectivity in valuations and ratings of property features, whether this is done by a buyer, an expert or the collective market! But there are ways that this subjectivity can be partially 'corrected'. Rankdesk applies two corrections (although there are many more!). One needs to first tackle the problem of information loss. For too long, property valuations, specifically those derived from AVMs [automated valuation models], have been reliant on too narrow an information base, mainly location, price, transaction date, type of property and lease. But what about end-of-terrace properties, properties with a porter, ones with southerly aspects? How do these features affect prices and time on market? Our algorithm 'learns' from these linkages so that any future ratings we assign to a property would have taken all of these conditions into consideration. That said, we need to be careful of common market failures. To give you an example, if we relied purely on market prices and buyer surveys we would give maximum points to properties with a private car park. But same buyers would find it annoying that their neighbourhood or city is more congested! Finding the perfect balance between market demand and sustainability (both in terms of quality & price) is the long term key to Rankdesk's success.

How has it been tested in the marketplace? How do investors/landlords know that it reflects the realities of the market?

We can track this quite easily. Properties that are rated the highest in the LONDON100 are the ones flagged under-offer more quickly. So we know that our ratings track the market quite well. I am sure that has nothing to do with our ratings! We are way way too small to influence the market!

Very many thanks for your time, Savvas, and the best of luck with Rankdesk. We'll be following its progress closely, and hope to chat again sometime soon.