Small Data for Ecommerce
Here is the writeup of the #EcomChat session on “Small Data”.
As ever, there were 3 conversation starter questions, and anyone was welcome to chip in live on Twitter using the #EcomChat hashtag. The questions were:
- What do things like ‘Big Data’ & ‘Small Data’ mean to ecommerce sites? Should they be thinking about these things?
- What data can Ecommerce sites use day-to-day, week-to-week to grow their results?
- What tools are available to put data in the hands of people in ecommerce roles, and what’s coming up in the future?
The full write-up is available as a PDF, kindly put together by @JackSaville1:
Small Data – EcomChat Writeup – Summarised by Jack Saville (PDF)
And details are also below:
Q1 – What do things like ‘Big Data’ & ’Small Data’ mean to ecommerce sites? Should they be thinking about these things?
@carmenmardiros : #ecomchat automation YES, also in surfacing those useful bits from data (less time digging, more time acting)
- @JamesGurd : @carmenmardiros Definitely – knowing how to make data presentable when required in the format the end user can actually use #EcomChat
@mcmillanstu : a data source I used to watch a lot was server performance data for 12 servers, That was data that turned into actions every day. #ecomchat
- @JamesGurd : I think key is knowing what outcomes you want to influence, then putting the data in place to help make the decisions #EcomChat
- @mcmillanstu : agree. An example: we wanted 2 make sure that checkout forms were usable, so made sure that we tagged validation errors #ecomchat
- @allcorknobottle : totally – you’ve got to have specific use for it so you can turn your data into useful infomation
- @JamesGurd Yep checkout error tracking really important (+ easily overlooked) – great to trend to see if errors scale with vol #EcomChat
- @JamesGurd : or if can isolate e.g. browser specific activity #EcomChat
- @mcmillanstu : we’ve already found issues on the site we launched pre-xmas, and have MVT tests planned for this week to improve it #ecomchat
@joeyizzy : personally I’ve worked with at least3 companies that pay for one of the big analytics packages but EVERYONE just uses the GA data.
- @JamesGurd : V true, often due to lack of understanding how to use a tool like SiteCatalyst or awareness of differences with GA #EcomChat
- @mcmillanstu : which is probably because Google have tried to make a user friendly tool! #ecomchat
- @emmabonar : Quite. Have you tried using Webtrends recently? UX nightmare #ecomchat
- @MontseCano : OMG, I had to extract data from there last week. Nightmare, and it’s not reliable. #ecomchat
- @emmabonar : I totally feel your pain, my users can’t (won’t) use it so i end up having to do it for them, horrible #ecomchat
@danbarker : Sites should keep an eye on the potential to improve results via big data tools, rather than the nebulous ‘big data’ term itself. #ecomchat
- @jon_taylor : agreed, big data really means more data, we now have the tools to process more information, in some cases in real time #ecomchat
- @jon_taylor : but we’re still spotting patterns and identify insights, thats no different to what we were doing already #ecomchat
@Reachly : It seems that most people get back to GA because they have questions for which answers can be found there.
- @OptimiseOrDie: Yes – and because you get flexible ways to pull that data (without large bills) #Ecomchat
@AndrewGirdwood : What did we learn in 2013? That an one size fits all dashboard rarely fits anyone. Data has context. #ecomchat
- @JamesGurd : Didn’t we learn that way way before 2013? I just think now there’s a larger pool of smart data driven thinkers #EcomChat
- @AndrewGirdwood : Before 2013 it was just a hunch. In 2013 we had validated that in a dashboard of big data. #ecomchat #ejoke
@mcmillanstu : I start my day with looking at analytics and it’s a tab (usually more than one) that is always open #ecomchat
- @lakey : Real-time analytics can be hypnotic. Whole hours pass by. Seasons change. #EcomChat
- @emmabonar : We used to display real time analytics for 4 brands in the group and made it a competition, surprisingly fun #ecomchat
- @lakey : I’ve been asking for a big monitor in reception for years. In fact, we’ll make a public dashboard. #ecomchat
- @emmabonar : Big data needs a big ol’ monitor 🙂 #ecomchat
- @mcmillanstu : 58% of visitors are from desktop right now… 🙂 #ecomchat
- @lakey : 51.7% = Google organic. #Ecomchat
- @MontseCano : Are you counting on the [not provided] data? *silly joke* #ecomchat
- @lakey : I have completely given up about that. Keywords? Pah!
@AndrewGirdwood : If you can use data to explain the decisions you’ve made then you’re doing it right. #ecomchat
- @RichardFergie : that’s backwards isn’t it?
- @AndrewGirdwood : It assumes you’ve the skills to make the decisions in the first place. Data shouldn’t usurp expertise #ecomchat
- @RichardFergie : I see. I don’t entirely disagree with you
- @RichardFergie : data should combine well with expertise. “Extraordinary claims require extraordinary proof” is mathematically sound
- @RichardFergie : which is how I sleep at night after making decisions based on past experience without running the numbers
- @AndrewGirdwood : Experience trumps data, in most cases. As it happens; it takes experience to know when. #ecomchat
@carmenmardiros : #ecomchat Surely data has different functions: operational, tactical, strategic etc. See Stephen Few, builds dashboards around purpose
- @OptimiseOrDie : Yes Carmen- we spend lots of time on site UX and fraction of that on reporting, metrics, process and dashboard UX. #Ecomchat @carmenmardiros
@OptimiseOrDie : #Ecomchat One key thing ignored is skills level of users. Not enough invested in job roles, skills, analytics training – for self servicing
- @cpbishop: couldn’t agree with you more. Or consistency on reporting tools for junior and senior members #Ecomchat
@danbarker : #EcomChat I broadly think ‘Big Data’ was/is good in pushing forward the agenda of ‘data’ in the press, among company leaders, etc.
- @MontseCano : That’s one my concerns – big data as a load of waffle. #ecomchat
- @JamesGurd : Agreed – Big Data is simply a catchy name but it often is something discussed not acted on. #EcomChat
- @jon_taylor : Big data is important, I think the press take it out of context. #ecomchat
- @jon_taylor : some of ‘big’ meta data analysis for cancer treatment for example, surface unknown insights #ecomchat
- @jon_taylor : totally different purpose to everyday analysis and data though, more long term / strategic #ecomchat
- @MontseCano : @jon_taylor @JamesGurd @danbarker Press usually take everything out context. Sad example the cancer one…
- @mcmillanstu : that’s a good use of “Big Data”, but a lot of ecommerce data is just plain-old relational data
- @jon_taylor : yes, but is why it’s a buzz term by association.
@danbarker : I broadly think ‘Big Data’ was/is good in pushing forward the agenda of ‘data’ in the press, among company leaders, etc.
- @emmabonar : Yup they’ve created a whole new data industry #ecomchat
- @davecatley : ‘Big Data’ has gone some way to highlight valuable technologies.& has raised the stock of people who work with data #EcomChat
@AndrewGirdwood : So what “big data” should Ecom sites wrestle with? Supply and Demand, I think, comes first. #ecomchat
- @KevinWaugh : Peaks & valleys in transactions, which ties back to supply & demand. Also customer data for campaigns too. #ecomchat
@OptimiseOrDie :#Ecomchat A1 : I’m from the ‘Big Data Isn’t Anything New’ camp – are we talking about a new tool driven landscape?
- @mcmillanstu : I’m with @OptimiseOrDie on this one… #ecomchat
- @dergal : big data involves things like Machine Learning ?
- @mcmillanstu :it can do, but it’s such an amorphous term that it can mean anything you want, from “lots of data” to GAs #ecomchat
@DanielJAustin : It’s big insight not big data right? You can get to the same answer with a smaller set. It’s about statical accuracy #ecomchat
- @MontseCano: Consistency in numbers is a problem for some, even if those were not correct in the 1st place! #ecomchat
- @DanielJAustin : You are just looking for patterns. All data has flaws #ecomchat
@OptimiseOrDie : #Ecomchat Need to make sure your data is correct in the first place, big or not! I’ve seen over 20 misconfigured sites in a year – bad.
- @emmabonar : Good point. Although sometimes it’s trends that are more important than data specifics #ecomchat
- @OptimiseOrDie : Yes but in many cases, it’s fundamentals that are broken (key goal recording) so most analysis tainted.
- @emmabonar : True. Any way to avoid getting tagging wrong? Most agencies don’t seem to know much about it in my experience #ecomchat
- @OptimiseOrDie :Keep agency on toes by having 3rd party audit tagging – pref an analytics expert. Many people on #Ecomchat can do this 4 u!
- @emmabonar : I guess. Would be nice to be able to get it right first time though, but hasn’t been the case for me so far! #ecomchat
- @MontseCano : Is that normal practice, Craig? Never done it…
- @OptimiseOrDie : Do ask @peter_oneill @carmenmardiros @danbarker @fastbloke or @timlb if you need any help!
- @mcmillanstu: I don’t think I would have seen trends so clearly if I hadn’t done things like cleaning up URLs in GA #ecomchat
- @OptimiseOrDie : Yes – clumping or splitting things and simplifying core UX/utility of analytics vital, esp for REAL USERS!
@danbarker : #EcomChat I think tools like google analytics, quantcast, etc, are often more useful in the real world – surfacing the bits that matter most
- @lakey : Beware tools that change the goalposts. That you I’m looking at, Google Analytics. #Ecomchat
- @MontseCano : Tools – I asked someone about measuring process the other day and got an answer about tools…#EcomChat
@AndrewGirdwood : If it means nothing to you in spreadsheet form then it’s big data. Right? #ecomchat
- @KevinWaugh : If you can’t make heads or tails of it at first glance, then “Big Data” it is. #ecomchat
@AndrewGirdwood : I suspect ecom sites do more with big data then they realise. You could argue that basic web analytics is big data. #ecomchat
- @danbarker : agreed – web analytics systems tick the ‘big data’ box – the trick of good tools is surfacing the bits that count. #ecomchat
@emmabonar : Big data sounds great in principle, eg Universal Analytics will let me incorporate store data with my web reporting #ecomchat
- @emmabonar : But does anyone need that much data? Particularly on a regular basis #ecomchat
- @MontseCano : Precisely. What I’ve seen so far is people getting over excited and too hung up on data, whatever the size. #ecomchat
- @OptimiseOrDie : Not small or big – the ‘right’ data?
- @MontseCano : That may be a better way to define it, ie, the ‘right’ data. #EcomChat
@AndrewGirdwood : Ecom sites need to think about the data they -need- and can act on. That starts as small data. #ecomchat
- @emmabonar : Completely agree, defining your KPIs first is key, then you can figure out where to get data #ecomchat
Q1 General Points:
- @dannydenhard : companies should be data driven & led, ensure data is actionable for long terms success & not worried about buzzwords #ecomchat
- @lakey: A1: Ignore big data unless you already have a brilliant little data strategy, to personalise the customer experience. #Ecomchat
- @AndrewGirdwood : Ecom sites need to think about the data they -need- and can act on. That starts as small data. #ecomchat
- @AndrewGirdwood : Wikipedia doesn’t have much to say on “Small Data” but here it is https://j.mp/1mufqYq #ecomchat
- @Tony_DWM : A1 For smaller biz’s running ecom sites (w/ smaller budgets), analysing who, what & why data is critical
- @mcmillanstu : businesses rely on information to make decision, information starts with data but isnt the same thing #ecomchat
- @KevinWaugh : #ecomchat A1: “Big Data” can assist in long term strategies (email campaigns), “Small Data” helps you see things now (0 orders today, why?)
- @OptimiseOrDie : If you don’t instrument and calibrate (check) then you may be making bad calls on that data. This is foundational to me.
- @carmenmardiros :But even small data feels big without the right people, processes and tools to leverage it
- @mcmillanstu : does having more data help you meet your overall business objectives? #ecomchat
- @AndrewGirdwood : I think “supply and demand” is interesting because it’s not a KPI. Instead it’s about strategy and growth. #ecomchat
- @danbarker : Measure all you can without adding complexity. Focus day-to-day on the ‘small’ bits that matter most. #ecomchat
- @danbarker : Asking “how can we combine & use the different data available to us to help achieve our objectives?” is quite a valuable question. #ecomchat
- @danbarker : I suppose another angle on the whole ‘big data’ thing for ecommerce is automation. Eg. data-powered product recommendations #ecomchat
- @danbarker : Sites should keep an eye on the potential to improve results via big data tools, rather than the nebulous ‘big data’ term itself. #ecomchat
Q2 – What data can Ecommerce sites use day-to-day, week-to-week to grow their results?
@emmabonar :’email me when back in stock’ requests are useful data for product availability & email campaigns #ecomchat
- @JamesGurd : Yep + tracking conv impact of OOS at category/product/variant level – v useful for fashion retail #EcomChat
- @emmabonar : useful for beating up the merch team 🙂 #ecomchat
@keena_newell : Abandoned shopping carts are pulled, then bundled into lists for targeted emails #ecomchat
- @mcmillanstu : do you then look at the nature of those abandoned carts to look at probabilities of converting? #ecomchat
- @cuzito : Love this – #ecomchat
- @danbarker : I like that. & even sites using automated cart abandon systems can ‘mop up’ at end of week/month using that. #ecomchat
@peter_oneill : My talk tomorrow at @superweek2014 is on See This Do That – creating reports that can be used for instant impact #ecomchat
- @JamesGurd : e.g. SEO is driving higher margin sales than paid search although PPC has higher gross revenue #EcomChat
- @peter_oneill : agree can do – although think less directly actionable – that comes into my diagnostic data matrix… #ecomchat
- @JamesGurd : Yes less actionable directly but can help with budgeting decisions for marketing #EcomChat
- @peter_oneill : definitely agree, also break down by browser, visitor type, country, entry point – identify a range of issues #ecomchat
- @danbarker : eg ‘these products have few views but very high conversion rate – promote em a bit more’ ? #ecomchat
- @peter_oneill : exactly right, ppl want these products when find them but hard to discover (data proves it) – #ecomchat
- @peter_oneill : Old blog post on Merchandising report (may need updating) https://www.l3analytics.com/2010/11/19/what-are-the-most-important-products-on-a-retail-website/ #ecomchat
@peter_oneill :My talk tomorrow at @superweek2014 is on See This Do That – creating reports that can be used for instant impact #ecomchat
- @danbarker : I like that – turning reporting into a rules-based guidance system #ecomchat
- @peter_oneill : Different types of reporting, this one is for non analysts based on analytics data that they are able to act upon
@peter_oneill :Merchandising reports which show products viewed lots, not added to basket & vice versa. Can take actions re products to promote #ecomchat
- @JamesGurd : and can match with cost data to look at margin impact across categories #EcomChat + down to traffic source level
- @peter_oneill : I have a client using a form of this every week to pick products for promotion/price change #ecomchat
- @JamesGurd : Interesting – what is key goals? Drive volume or margin, or both? #EcomChat
- @peter_oneill : They have a fairly constant clearance area, working to select products for that – more volume goal
- @AndrewPMorris : A good measure to try and understand is the effect on total basket value of a product. Does it drive other buys
- @peter_oneill : That sounds a lot more difficult but very interesting
- @AndrewPMorris : it is more of an analysis to look at after all the easy ones have been done. ecommerce marginal gains!
@JamesGurd : A2) As per @mcmillanstu, server monitoring to flag when perf issues and help diagnose causes e.g. is it traffic? #EcomChat
- @jon_taylor : yes, useful data, our page load / journey speed and site availability go into the board report #ecomchat
- @emmabonar : Your Board are interested in page load time?? Wow, you have educated them well! #ecomchat
- @jon_taylor : We have to provide commentary if its outside normal range, like saying ‘how long was the shop open / how big the queue’ #ecomchat
@danbarker : #EcomChat I think most sites pay attention to click, traffic, sales numbers. user/customer data, competitor data, qualitative often missed.
@AndrewGirdwood : Ecom sites should be very aware of their #affiliate data. This shows what’s selling and what tactics affiliates are using well. #ecomchat
- @JamesGurd : What core metrics would you advise for #affiliate marketing Andrew? #EcomChat
- @AndrewGirdwood : Top affiliates, conversation rates, which product/service is selling, returns, ABV, time of day & which creatives work #ecomchat
@mocwoods : I’m still waiting for a client to integrate sales data on their white board in to Analytics… (it’s been 2 years+ already) #ecomchat
- @danbarker : I once worked somewhere where we built ‘web sales’ into call centre sales team targets. worked unbelievably well. #ecomchat
- @mocwoods : Nice, yeah that’s worked on some clients for us too, but their sales team are scared of data and change… #ecomchat
@AndrewGirdwood : Ecom sites should be very aware of their #affiliate data. This shows what’s selling and what tactics affiliates are using well. #ecomchat
- @JamesGurd : What core metrics would you advise for #affiliate marketing Andrew? #EcomChat
- @AndrewGirdwood : Top affiliates, conversation rates, which product/service is selling, returns, ABV, time of day & which creatives work #ecomchat
Q2 General Points:
- @danbarker : #EcomChat I think most sites pay attention to click, traffic, sales numbers. user/customer data, competitor data, qualitative often missed.
- @emmabonar : My basics are Online: traffic conversion & AOV (by device). Offline: returns #ecomchat
- @AndrewGirdwood : I also think Ecom sites need to go through their CRM and returns data carefully. #ecomchat
- @AndrewGirdwood : ASA regulation demands Ecom sites know their sales & supply data. If your sale runs out of product you may be at fault #ecomchat
- @JamesGurd : a2) I’m a big fan of PVV – not just conversion rate or revenue, but marginal value of an additional visit (segmented to hell) #EcomChat
- @SMXChat : A2. the improve results businesses need to use customer behaviour data more often followed by action. #ecomchat
Q3 – What tools are available to put data in the hands of people in ecommerce roles, and what’s coming up in the future?
@AndrewGirdwood : It’s a bit technical but for working with “messy data” then Google Refine is pretty cool! https://j.mp/LGOrLB #ecomchat
@AndrewGirdwood : I’m going off the beaten track for Q3 and will recommend @zapier as a way to tap into a data source, write a rule and action/alert #ecomchat
- @NedPoulter: I’m speaking about @zapier + Automation at #BrightonSEO (April) if either of you guys are attending? 🙂 #ecomchat
- @AndrewGirdwood : I need to work harder on getting to #BrightonSEO.
@KevinWaugh : A3: It sounds dumb to say, but a word processing document to present the data effectively is a needed tool. #ecomchat
- @emmabonar : Not dumb at all, an effective (non manual) way of presenting key data (like a dashboard) is really important #ecomchat
- @iw_expo : So true and not dumb at all. Especially for presenting to snr management / external stakeholders etc.
- @joeyizzy : true, as able to present key data in comprehensive way often decides when stakeholder buy-in is achieved
@JamesGurd : How do you bundle the visitors into lists? How do you define criteria? Is it based on product, or behaviour? Or mix? #EcomChat
- @mcmillanstu : When I used to spend more time on this, I would try to qualify the list as much as possible, to save spam #ecomchat
- @keena_newell : smart, good point and it’s a struggle
- @mcmillanstu : I must try and recreate a bayesian method I used for this and post it… #ecomchat
Thanks very much to the great @JackSaville1 for the excellent write up!
Join us at 1pm UK time on Monday, and do share this post either on Twitter or via email with anyone you think would be interested.
Dan & James.
p.s. If you’re wondering “what is this all about?” there is an about page to explain everything.
No comments