Why optimise at all?
Spotting problems is easy. Knowing how to fix them is different. Enter data analyst. You gotta test to find out if you're wasting your energy! For example, you can pour all the budget into responding to customer emails as quickly as possible, but if all customers care about is getting a solution first time, you're wasting your time.
How can you tell what customers care about?
What else can you test?
Personalisation effects, product success, ad response rates, customer targeting performance, customer satisfaction or anything else you can dream up a hypothesis for and find the data to prove.
Roadblocks
Stepping back for a sec, some data analysts can't get their company onboard with tests like these and they're stuck underwater until they can. At the Digital Analytics Hub (DAH) in London, it was clear that the most common roadblocks to digital analysts (DA) were:- not enough people on the DA team
We don't have time to answer every request we get from other department for metrics so we become the bottleneck in the business because the internal demand is so high. Frustration is born. Or, at the other end of the scale, the one guy running the department doesn't have time to do the analytics, champion his insights for each department AND go to strategy or review meetings.
- not enough budget
The company is data keen and invites their DA team to all relevant meetings, asking them where they can improve and planning experiments to prove their hypotheses. Heaven! But now it's crunch time and the budget goes to marketing for that new product launch (with no left overs for the analytics team to track how the launch went, or hire a DA to look after that product's success).
- earning trust
Data analysts are in the game of optimising the business, not just the website, but we're not automatically invited to support and guide decisions in meetings. So we have to send bodies to tables and bash down the doors to educate our team and prove we're vital brains to have at the table for business strategy discussions, bringing data on what's working and what's not. It often takes a failed project to demonstrate that the analytics team need to be included earlier than 2 days before the staff want results! If you're lucky, they'll come to you earlier next time, you'll dazzle them with insights and from that day forth you're a hero in the business...who now can't handle the demand and you're back to point one!
Prioritisation becomes paramount.
- Automisation
Rather than manually gathering the data every week to write that 40 page report for the staff (which let's face it isn't getting read anyway) set up automated reporting that collects the data you need automatically. Or go one better and personalise the links to your teams so they only see the relevant bits to them.
- Cross-department priority meetings
This can be a great game; get all the heads of each department in a room (anyone who gets a budget from the finance team is relevant here) and a pack of post-it notes. Get everyone to write down their top priorities for next month and puts them up on the board. Everybody will bring their own agenda to the table, so now, you all step back and prioritise (allocate budget to) the most important things for the business; each person marks (tally style) the 3 post-its they think should be top priority for the business that month. This helps bring teams together and focus their efforts in the same direction.
- Fight your corner
Show your team how analytics can support every department; without analytics, the sales team can't see how the product performs in the market; without analytics the IT team can't see how to optimise the website to increase sales; without analytics, marketing campaign effects are unknown; without analytics, the finance department can't see where best to invest the budget; and without analytics, the business can't see its weak spots and the competitors who can will overtake you. It's vital! Do they know?
Use examples to persuade your team:
- to continually A/B test:
When was the last time you saw Amazon do a website update? Exactly - they make continual tiny changes - A/B testing as they go - to play around and see what works. They throw away what doesn't working, using failure to learn and improve.
- to optimise the website:
- to check the GA account configuration:
You buy new tills. Do you check they're working OK on the first few days? You know - counting up the money and checking the gift vouchers, credit cards, debit cards and refunds all tie up with your sales figures. You could be losing money if they're not set up correctly, or accidentally overcharging your customers. Checking your GA is just the same - if you're basing strategic business decision on unreliable data, your revenue is at risk! Check yo'self.
Now how do I get them involved?
Here's a bunch of ideas to light a firecracker under your team
Which A/B test will win? ...find out next week! Not only will this get people personally attached to the outcome of the tests, asking you how it's going, arguing their case and hashing out unforseen issues it also lets the class moron who thinks he knows better than anyone (and that tests are pointless), swallow his words when he loses the bet. It teaches everyone that you can't know what's best, even if you're an optimisation pro! Winning team shares the pot of the losing team's money, or a prize like free lunch or a trip to the cinema.
Suggestions inbox
Set up a survey on your site that asks all internal traffic if they see any spots for improvement on the site (this'll help you see the site from all perspectives). Send all feedback to the optimisation team.
Suggestions inbox
Set up a survey on your site that asks all internal traffic if they see any spots for improvement on the site (this'll help you see the site from all perspectives). Send all feedback to the optimisation team.
Create an easy to read A4 report card for every test
This'll help people to actually ingest the outcomes of your tests and analysis - any longer than 2 sides and it's information overload. You should include:
- A picture of the page before and after (highlighting change)
- The test title
- Why the test was done
- The hypothesis (what you expected to prove)
- Results (labels to explain your graph if your team don't like figures) - remember this is nothing without context - what should the figures be? what's your industry benchmark?
- Annual payback (ROI) of this change
- What we learned (no results = this factor doesn't affect pageviews / CTR etc.)
- Put all report cards into a wiki / database so people can avoid the groundhog day effect, repeating tests unnecessarily, instead just searching the wiki for optimisation test results.
You won't always have to drag them kicking and screaming over to you side to show them analytics is important. For example, if you can see there's a 45% drop-off after people complete a sign up form, and actual registration completion - investigate! You might find the problem is the sign up email (with a "confirm registration" link hidden away in the paragraph text) - nobody will argue that's worth fixing immediately and you'll earn kudos for figuring out the problem.
To summarise:
- Testing is vital to see if your new strategy/product/campaign is working
- Survey your customers to find pain points to work on
- Digital Analysts's biggest challenges are: not enough analysts, not enough budget & earning colleagues' trust
- To overcome time and money issues: hold cross-department priority meetings and automise reporting
- Persuade your team to run A/B tests, optimise the website and check Google Analytics account configuration
- Get your team involved with betting, suggestions inbox, cool report cards or by finding low-hanging fruit to whet their appetite. You could even run an in-house lunch time workshop.
What problems do you encounter as a data analyst? What tricks do you use to get the budget and staff you need? How do you get your team excited and invested in optimising your site? Let me know know in the comments below - I'll reply to each and every one.
