Author: Chris Jones, Independent Multichannel Retail Consultant
In this series of articles, we’re considering how to estimate the likely sales you might be able to make by targeting your ecommerce at a new country – clearly an essential part of any budgeting process. We’ve already looked at a top-down process for doing this (see ‘Sizing the prize: Using a top-down approach - Step 1’ and ‘Step 2’) and in the previous article we started to consider, by contrast, how to make a bottom-up estimate. Step 1, (see ‘Sizing the prize: Using a bottom-up approach - Step 1’) described in the previous article, is to estimate the expected visitors. In step 2 we’ll move on to estimate cart-sizes and conversion rates for these visitors.
Step 2.1 is to consider whether you need to split your modelling by device types (smartphone, tablet, PC). In general, if you are targeting a western European country, USA or Australia, device-usage and conversion profiles are probably close enough to home to be able to avoid this extra complication. In many other countries, it isn’t necessarily quite so simple – tablet penetration is especially variable by country – and you may need to split your model in three and adjust for the proportion of visitor traffic you expect.
In the next steps, we look at the key factors which might affect conversion, compared to your benchmark current conversion-rate back home. Unfortunately almost all of them are likely to have a negative impact! The corresponding budgeting decisions you will probably be taking will be about whether to invest in further localisation to minimise this.
We’ll start with step 2.2: price point. There’s a basic rule of thumb that says that conversion rates on more expensive items are less than on cheaper items. If your expected visitors in the target country are going to be comparably wealthy to those back home, and your price positioning in the new country will also be at parity, then you can assume there’ll be no effect. However if you are targeting a less wealthy demographic – for example because you are aiming at Eastern Europe or South America – then even if you maintain price parity, you’re probably asking for a higher percentage of your visitors’ purchasing-power. This will reduce conversion rates; a reasonable planning assumption is that it will reduce in proportion to the relative discretionary spending-power in the target country compared to back home (using average income figures will not work, because the poorer demographics in less wealthy countries tend to shop online much less, so you’ll need to make some reasonable estimates). So you’ll need a weighting-factor for price point conversion, which will typically be somewhere in the 50-100% of what you are used-to in the UK.
Step 2.3 considers language. In the IMRG / eCommerce Worldwide guide to taking your ecommerce international (see ‘A Nation of Shopkeepers’), we showed some data about how comfortable various European nationalities are shopping in a foreign language:
If you aren’t planning to translate your site, you can expect to see a weighting-factor for language pretty much proportional to these numbers, so for example about 43% for France. Be careful not to make this too depressing by double counting – it’s only relevant if you did your marketing in one language but your site is in another. So if you’re paying for SEM using French keywords, but then land those French customers on an English language site, you’ll suffer this kind of attrition. But if you’re marketing in English, even to French nationals, and then bring them to an English site, you won’t see extra drop-off.
Step 2.4 considers payment methods. In the IMRG / eCommerce Worldwide passports, and also in ‘A Nation of Shopkeepers’ as previously mentioned you’ll find data about payment methods used in different countries. If you are missing a key local one – for example not offering iDeal to Dutch customers – you can expect to see another weighting-factor which is less than 100%. It won’t be completely in proportion, because there will be some willingness to switch, for example from the preferred iDeal to the less-preferred Visa / Mastercard. A reasonable rule of thumb is to assume that about half of customers will be willing to switch. Pay-on-invoice has a 28% share in Germany, for example, and you should probably assume that about half of them won’t willingly use a credit-card, so your weighting factor would be 100%-(28%/2) = 86% if you don’t offer pay-on-invoice to Germans.
Step 2.5 considers delivery. Customer behaviour here is slightly counterintuitive. If an overseas customer is well-aware that they are buying from what to them is a “foreign” site, they will typically be expecting to pay more for slower delivery, and be tolerant of being asked to do so; your weighting-factor will probably be 100% or close-to. If, however, you localise your site, they won’t: a slow promise or high delivery charge will then reduce conversion. You probably already have a business case back home for whether or not it is worth subsidising delivery which includes estimates of increases or reductions in conversion, and this can be applied as-is to overseas customers of a localised site. If not, a weighting-factor of 50-70% is a reasonable starting point assumption.
Step 2.6 considers returns policy. It’s no accident that online retailers such as Asos offer free returns pretty much globally (at enormous expense). If you are selling returns-prone items, especially fashion, and your returns policy is not at parity with local expectations, you can expect to see significant impact on conversion – assume a weighting-factor of around 75%. In some countries, however, especially those where online retailing is less developed, you may be able to exceed local expectations. In this case you may actually see a conversion up-tick, and therefore, for once in this article, a weighting-factor greater than 100%!
Step 2.7 is just arithmetic. Take your home conversion-rate, and then multiply it by all these weighting-factors. For example:
Home conversion (3%) * price-point (100%, Germany is comparable to UK) * language (100%, localised German site) * payment methods (86%, not offering pay-on-invoice) * delivery (70%, slow from UK but at typical local German price) * returns (75%, not quite up to German standards) = forecast conversion rate of 1.45%.
Step 2.8 considers cart-size. Again this is slightly counterintuitive. Overseas customers seeking out your UK site, and paying a premium for delivery, will spend about a third bit more per cart, according to IMRG’s figures. Once you localise, this increased cart-size will disappear as an effect.
Finally step 2.9 is more arithmetic. In Step 1 we calculated visitor numbers. We’ve now found expected conversion rate and cart-size. Multiple the three together, and there’s your sales forecast!
In this short series of articles we’ve look at how to forecast website sales in a target country using both a top-down and a bottom-up method, in order to try and build a reasonably solid business-case. As far as possible, we’ve tried to avoid using assumptions which overlap between the two methods, so that they can be used independently, and therefore to try and sanity-check each other. No one wants a business case based on fantasy, or heavily dependent on one dubious assumption which returns to bite-back over-and-over, and so having two separate approaches is a good idea.
If the two approaches arrive at different answers, be suspicious, and go back to try and understand which assumptions (or weighting-factors in the more arithmetical parts of these models) are driving the discrepancy.
If they arrive at similar answers, be doubly suspicious! Just to re-iterate a point made earlier on in the series: these are consultant’s models. Never use a consultant’s model without challenging it. They are useful to structure a problem, and provide a framework for thinking about it, because the framework and structure has worked elsewhere previously. The default numbers suggested are based on experience. But… they should never supersede your own judgment of what matters for your own business.