Preview

Correlating the attributes that suppliers show on G-Cloud with sales data shows what buyers look for

There are a number of attributes that successful products tend to exhibit and which unsuccessful products do not

This article publishes the attributes with the highest positive sales correlation

Benchmark your product, first against this general case – is there good reason to ignore buyers’ preferences?

Reperform the exercise with successful products comparable to your own to identify the effective differentiators

Beware ‘attribute inflation’ we expect buyers to increasingly raise attribute expectations

What we mean by attributes

Get-found: Vital First Marketing Step looks at the important Product Summary, Features and Benefits. In this article we are looking at the 140 questions of detail which SaaS vendors respond to about their offering which forms the bulk of data displayed on the Digital Marketplace. Concentrating on the structured responses (pull-down menu selection), we can correlate what successful suppliers have responded against the responses from suppliers with no sales. Although we are looking at a ‘general case’ where buying decisions covering 8,000 products are going to blur the relationships, what emerges is a clear signal of some of the attributes that buyers favour.

How to use the table

This analysis shows us some of the general attributes which buyers are looking for. In a competitive situation, where your product is being compared to another and one possesses an attribute and the other does not – the question of whether it will influence selection will vary from case to case. So the list needs to be applied with knowledge of your specific domain. For example, in the general case buyers seem to have a propensity to select products where access to a public sector network is available. In your specific domain, this may be a significant differentiator, in other areas it may have little impact, which is why when the effect is averaged over all 8,000 products the correlation is diluted to ‘modest’.

To make this form of analysis specifically relevant to your services we need to take a look at the subset of successful comparable and competitor products. Then you will be able to identify the active differentiators that influence buying behaviour in your market – if you have some gaps, these are worth considering for the product/marketing plan. The sales data and product data are openly available. If you need help in making the correlation and want to use the latest available sales data, please get in touch.

Attribute inflation

In discussions with CCS and buyers we have noted how some buyers like to use categories and filters despite the inference that these indiscriminately eliminate products and so lead to poor value decisions (see the argument in Get-found: a vital first step in marketing). This is one sign that buyers, faced with many competing products in some areas may use a random tool to reduce the long-list of candidate products to manageable proportions. CCS advice in the Buyers’ Guide when faced with too many candidate products is to ‘add filters’. This, and the addition of ‘nice-to-have’ requirements for the same purpose will cause attribute inflation, the tendency to specify higher requirements than necessary.

Attribute inflation is also a feature of a maturing and evolving marketplace. For example, accessibility standards to WCAG 2.0 AA are mandated in the Technology Code of Practice. While this is not universally applied at the time of writing, education and pressure will, we feel, increasingly put this standard in the ‘must have’ requirements which buyers have a duty to source. This provides a nimble SME with a fresh and evolving product an opportunity to differentiate from vendors who are slow to react to the needs of the public sector buyer.

This makes correlation analysis a useful tool to research and identify trends. As Ivanka Majic pointed out in her 2014 GDS blog on G-Cloud:

…I invite you to think less about being at the top and more about being the last one left after the elimination round. So when writing your description, keep asking yourself, “what requirements does my service meet?”

This analysis points towards the requirements that are being sought.

A footnote on methodology

In this section we reviewed the (mainly) non-functional information on the catalogue. For SaaS products there are approximately 140 possible fields of data which may be presented (some are conditional upon a positive response to an earlier field) and of these, roughly half have defined responses presented in pull-down menus when completing a submission. These defined answers may be a simple Yes/No or up to 9 alternatives some of which may be unique selection or multiple selection.

There were, at the date we extracted SaaS catalogue data from the Digital Marketplace (December 2017), 6,634 SaaS products listed and 1,755 uniquely identifiable suppliers of which 217 had at least some sales on G-Cloud 9.

To get a signal into what is emerging as current best practice (or state of the market), we approached the question from 3 angles.

1. Attributes which are likely to be common for all products in a supplier’s portfolio – such as ISO27001 accreditation – we make a test database of 1-product from every supplier using an algorithm to select a representative product. We then compare the attributes to relative sales success and highlight those that do appear to have a positive correlation with sales success. This is the data published in this article in the above table.
2. Attributes which may be different across products in an organisation’s portfolio – such as ‘users can customise service’ – we cannot correlate against sales success because sales data is published by organisation (not product). In this case we can interrogate the database of all 6,634 products and look for frequency of occurrence. If more than 75% of products feature an attribute, we consider it is ‘best practice’ or ‘industry standard’. This data will be published in another Insight article.
3. For the qualitative assessment of free-text information i.e. without defined responses in a drop-down menu, we took a random sample of 100 (5.7%) of the 1-product test database created in (1) above and reviewed the content. Completing a scorecard for each supplier against a defined set of criteria. We then compared the result of this review against relative sales success to find a signal indicating good practice. This information informs and qualifies the opinions expressed in Service Definition Non-conformance and other articles .