Typical. You wait ages for a great Analytics conference and then two come along at once! Following on from our coverage of MeasureFest on Monday, today I will be taking a look at all of the talks from today’s eMetrics.
This post will be updated through the day and we hope you find the round-ups useful.
Multinational Analytics – A Different Kind of Numbers Game
Multinational Analytics involves identifying three tiers – the overall targets of the business as a whole, the various importance to each country, and the various markets in each country.
There are many different lines of businesses with their own objectives, which can often act as silos. Analytics can help break down these silos and bring brand messages together by encouraging relevant lines of business to work together.
Consider how you will integrate Analytics into everyday digital life. You need to consider the Analytics maturity in your various target countries – not everyone needs to be at the mature end of the scale. Some countries may only require basic metrics, whereas others may be able to use much more in-depth data. You need to provide the right level of analytics applicable to the team you are dealing with.
Understand what it is you are going to measure in order to evaluate brand perception and awareness. This needs to be a universal agreement across all teams. This will help various teams take responsibility for the metrics most important to them. You need to gain organisational excellence by ensuring everyone is on the same page, but also have the flexibility in place to alter the market drivers dependant on what’s relevant to each line of business.
It is important to understand what the various KPIs are for each line of business. Then, put these KPIs into buckets (e.g. reach, engagement, and completion) for simplicity. Spend 10% of your budget and the other 90% on people.
Stock: You can have a website with outstanding CRO but if the product is rubbish then the user won’t convert. Even if you have the best product in the world, then users still might not convert without CRO. You can set up Google Analytics (GA) event tracking to monitor which products are best sellers but are out of stock or fractured.
Price: If your average value of basket is falling over time, you can bump up sales by twixing (placing cheap products which can be added by impulse at checkout). Event tracking can be set up to monitor which products are then removed at this stage to monitor whether or not you are in fact cannabalising your sales I.e. two twixs added, one expensive product removed leading to a lower overall spend.
Wish listing: Encouraging people to add items to a wish list then allows you to send promotions about these products. I.e. 20% off your wish list if purchased in the next 24 hours.
Only 13% of people on GA use site search functionality within the GA account. Look at what items have been brought, and what other items were searched for in conjunction – you could then show these specific products on pages with the original products – for example, showing cheese on a wine page. This also highlights new product areas to expand into.
AABB testing can mean more accurate results. Test one red apple, one green apple and two oranges. What’s your favourite colour apple? What’s your favourite colour orange? If the two oranges end up the same then the more significant apple wins. If the oranges do not match, then the test is void.
Use site search to identify questions which your content team can provide answers to.
Drawing the Line on Privacy in the Technical, Ethical and Legal Sands
Your name is not strictly unique and therefore could be argued it isn’t PAI, but in some instances it can identify someone with an unusual name.
Visitor ID is not identifiable in itself, however in conjunction with third party software it would be possible to identify the person.
A lot of companies are uncomfortable with retaining postcodes, however, these aren’t overly granular in comparison with some of the other information not consider to be PAI.
Collecting certain elements of data is fine but when you are linking it to other elements of data, you may be entering dangerous territory. You need to consider the context in which the data was supplied by the user and what their intent for its use is.
Measurement is broken because we measure data within silos, our KPIs don’t tie into profit, our data is unorganised and the data consists of complex, anonymous interactions.
Measure is made up of two models; Top down: do we have the right spend in each of the key market buckets and Bottom up: using individual level data to gain insights into what is working and what is not. A Top down approach alone does not provide actionable data.
Optimisation is capable due to data addressability. For example, data from a TV is not specific, whereas digital allows us to collect much more granular data.
Last touch can lead to inaccurate data interpretation. For example, if PPC is considered to be effective because it is the last point of touch, you may invest further budget there, detracting it from radio which is actually an important source for providing good leads. As a result, this makes your PPC conversions drop.
The various media that we can’t currently track at an individual level, is moving towards being able to be tracked, e.g. TV, radio etc. This will allow us to draw upon the Bottom up approach more.
Tag management systems are sold on the premise that one tag fits all and will be easily deployable, but in actual fact only 1 in 5 can be replied as a template, the rest require quite heavy customisation.
Don’t deploy your tag management solutions on a Friday as you don’t want to run into any major issues over the weekend. When implementing a full role out, do this when it can be closely monitored.
It is recommended that different levels of authorisation are used within your TMS so that only your quality/technical team can set the custom tags live on the site.
It’s important to have written documents covering the process and an Analytics agreement with each company. It can also take years to get a good TMS implemented properly using data layers, so if this is something you are considering for the future, start thinking about your data layers now.
The less you can depend on the jQuery in your customisation on a TMS, the more you will benefit in the future.
Focusing on sales leads to a loss, focusing on cutting expenses leads to low customer service. We need to focus on three things; raise revenues, lower expenses, increase customer satisfaction.
Personally provided information – is valued information which is provided by the user and available information – is readily available information about your customer.
Amazon were the first to collect personally provided information by collecting the email address of the user. They then collected other targeted information to provide a service to users by giving them targeted offers and exclusive first-hand knowledge through the power of pre-ordering.
For the reward of recording personally provided information, you must provide them something in return. Each new piece of personally provided information builds on the relationship with the business, especially if extra value is added. For example, one-click buying from Amazon (think about how much data you give them in return). Don’t try and force the relationship to grow too quickly.
Data is an asset and therefore should go in your records as an asset. Rules of collecting Personally Provided Information; 1. Always ask for opt-in, 2. Fair perceived value, 3. Collect data for customers not everyone, and 4. Standard Analytics practices apply.
We will be back tomorrow with a writeup of the second day’s talks.
Unlike 08 numbers, 03 numbers cost the same to call as geographic landline numbers (starting 01 and 02), even from a mobile phone. They are also normally included in your inclusive call minutes. Please note we may record some calls.