Some moments in time are perfect to reflect on where you are, what your priorities are, and then consider what you should start-stop-continue. In those moments, you are not thinking of delivering incremental change… You are driven by a desire to deliver a step change (a large or sudden discontinuous change, especially one that makes things better – I’m borrowing the concept from mathematics and technology, from “step function”).
In those moments – common around new years or new annual planning cycles – the difference between delivering an incremental change vs. a step change is the quality of ideas you are considering. In this post, my hope is to both enrich your consideration set and encourage the breadth of your goals.
My professional areas of interest cover Customer Service, User Experience and Finance, though here on Occam’s Razor my focus is on influencing incredible Marketing through the use of innovative Analytics. To help kick-start your 2019 step change, I’ve written two “Top 10” lists, one for Marketing and one for Analytics – consisting of things I recommend you obsess about.
Each chosen obsession is very much in the spirit of my beloved principle of the aggregation of marginal gains. My recommendation is that you deeply reflect on the impact of the 10 x 2 obsessions in your unique circumstance, and then distill the ten you’ll focus on in the next twelve months. Regardless of the then you choose, I’m confident you’ll end up working on challenging things that will push your professional growth forward and bring new joy from the work you do for your employer.
Ready?
First… The Analytics top ten things to focus on to elevate your game this year…
The Step Change Analytics Obsessions List.
A1. Improve the Bounce Rate of your top 10 landing pages by 50%.
(Improving Bounce Rate results in reducing it. :))
You'll be surprised by the steep drop in Cost per Acquisition.
Google Optimize will be one of your BFFs in this quest. You’ll know you’ve moved beyond basic improvements when you start setting Custom Objectives – they require deeper thinking, which is a good sign.
A2. Eliminate 40% of the numbers from your dashboard.
Take the newly-created white space to explain what to do based on performance of 60% of the numbers that remain.
What your boss wants most this year, more than love, is to be told what the data wants her to do. Don't leave her guessing.
(Bonus, with actionable ideas: Smart Dashboard Modules.)
A3. Take your first steps towards unlocking smart algorithms.
Learn what Session Quality is in Google Analytics, then learn how to use it in your campaigns to improve conversions. In the Audiences section, go to the Behavior folder.
Learn what Smart Bidding is in Google Ads, then learn how to use it in your campaigns to improve outcomes.
Machine Learning algorithms will make our data smarter in unparalleled ways; Session Quality and Smart Bidding offer early clues about the scale and type of intellect. In both instances, it is immensely valuable to really understand how a smart algorithm uses billions of data signals to calculate likelihood of a conversion.
Across all your analytics data, algorithms will take you places humans simply can't. This should be the year you invest in an expansion in skills and practice to take advantage of these possibilities.
A4. Take a class in data visualization. It will save your life.
Anyone can make a complicated visual, it takes someone very special (you!) to draw out the essence of the story data is trying to tell.
My recommendations:
Affordable: Data Analysis and Presentation Skills at Coursera.
Occam’s Razor: Start with this one: Closing Data's Last-Mile Gap: Visualizing For Impact. And, there are five more linked to here.
Through all these courses remember the most important thing about data visualization: It’s not the ink, it’s the think. Obsess about improving the think, just as much as I’m encouraging you to improve the ink.
A5. Obsess about what happens after campaigns end.
In our analytics practice we tend to celebrate victory too early (at the end of the campaign) or with insufficient breadth (the full scope of impact).
Did you get customers with high lifetime value? How long did the brand lift – say Awareness – last? What was the average order value of the second purchase by people you acquire via Search, compared to those via Retail?
Is there a difference in behavior between people who signed up for email over the last year vs those who did not? What the cost of getting a retail customer to make subsequent purchases over mobile apps lower?
A6. Understand your personal impact, obsess about improving it.
Grab the revenue number for the company. Now work out how much of it is influenced by you directly. Make a note of what it is (likely to be a couple percentage max).
Double that number this year.
What are the first five things on your list?
None of them will be easy, but converting insights into action via influence rarely is. But, you don't have to stretch too far to see how amazing it would be for you (and data too!) if you double your impact.
A7. Run one super-large controlled experiment.
To prove what your Executives believe purely from their gut. Or, to disprove it.
Does Facebook advertising really work better than TV? Can you create premiumness for your brand using digital? Is a 15% coupon now better than 20% off the next purchase? Does swapping out male model posters for cute animals triple sales?
Does sponsoring a fashion show lead to an increase in brand equity? Does free pickup in store result in higher attach rates?
A8. Identify four relevant micro-outcomes to focus on in 2019
(In addition to the macro-outcome of revenue).
Businesses win when you optimize for a portfolio, because at any given time only a tiny fraction of people want to buy. Solving for micro and macro-outcomes is directly connected to the holy grail of solving for short-term AND long-term success.
Employees also become smarter when they have to optimize for more than one thing. :)
A9. Throw away your custom attribution model. Embrace data-driven attribution.
For some things, humans are already less smart than machines. Trying to guess what might be happening across millions of touchpoints on and off site, on and offline, is one of those things.
Skip the first five steps of attribution’s ladder of awesomeness, jump to DDA. From the tens of hours saved per week, figure out how to feed offline data into your data driven attribution model.
With an obsession with data-driven attribution, you are also solving for a portfolio rather than a silo. Super cool, super profitable.
A10. Hire an experienced statistician to be a part of your analytics team.
There is too much goodness in modeling that you are not taking advantage of. From segmentation models to identifying incrementality to predictive modeling to survival analysis to clustering to time series to… I could keep going on and on.
2019's the year you get serious about serious analytics.
A11. Bonus: Reporting kills, analysis thrills.
If that is true, and it is, :), then what % of time are you personally spending between Data Capture – Data Reporting – Data Analysis?
Outsource or eliminate half of your data capture and data reporting responsibilities, and allocate it to data analysis and driving action.
You'll be surprised at the increase in your salary and bonus (oh, and the company will benefit too!).
In context of Analytics are you aiming for something special in 2019 that I've not covered above? Will you please share that with me by adding a comment? Thank you.
Switching gears, here are ten things to obsess about to collectively deliver a step change via your Marketing game this year…
The Step Change Marketing Obsessions List.
M1. Improve the Bounce Rate of your top 10 landing pages by 50%.
(Improving Bounce Rate results in reducing it. :))
Same as the #1 on the Analytics list. :) Far too many Marketers ignore this simple strategy to make lots more money. You work so very hard to earn attention, why then let your ads write checks your website can’t cash?
An additional delightful benefit: I find that getting Marketers to obsess about landing pages forces them to audit the user experience, something worth its weight in gold.
M2. Put up or shut up time for your social media strategy.
99.999% of corporate social media participation yields nothing.
Your CMO wants people to love your brand and organically amplify its goodness. It genuinely is a good thought. Except, a cursory glance at your social contributions show nothing of that sort over the last three years.
So, why are you spending all that money?
I recommend using that money to buying your team iPhones every Friday, I assure you that'll have a positive ROI.
Or. Focus on social media primarily as a paid media strategy. Bring the same discipline to the application of accountability to social media ads that you bring to your Display or Video ads anywhere on the web.
Here are five brand and five performance metrics that'll be your BFFs in 2019, as you social strategy lives up to that now famous mantra: Show me the money!
M3. Keep control of creativity, give up control of the creative.
Machines are much better at optimizing the latter for short or long term.
(For now) You are still better at the former – do lots of it, then hand it over to smart algorithms.
It is hard, especially for creative types who confuse creativity with creative. But, with every passing day you are harming your bottom-line more if you don’t follow the formula above.
Also consider the Machine Learning opportunities for Marketing beyond creative.
Aim to shift 25% of your marketing budgets in 2019 to opportunities that are powered by ML algorithms and rejoice at the boost in profits that results.
M4. TV works, solve for each factor that drives success.
Most TV campaigns are sold and bought based on reach (GRPs FTW!).
In my experience you should optimize for reach AND one overarching story AND creative consistency AND ensure each successfully tested creative has enough frequency to wear-in.
And, if you can't solve for three ANDs… Shift money to max out the Performance Digital opportunity, then with the left over money buy every person in your team – and at your agency – a new car. Your TV budget is big enough , and trust me when I say that giving out a new car will have very high motivational and bottom-line ROI.
M5. Seek to understand the customer journey.
What drives the first purchase? What drives the second? What drives the support calls in between? What does using the product really, really feel like? What drives advocacy?
All advertising that fails does so because the Marketer behind it understands only one sliver of the experience, then solves for that sliver with heart-breaking short-term focus.
When the Marketer understands the answers to the above questions, it influences the creative, it influences targeting, it influences retail store displays, it influences frequency, it influences product design, it influences…. it changes everything. Including profits.
Journeys are better than tinder dates.
M6. Solve for intent. It is more possible and more critical with every passing day.
See-Think-Do-Care is a great intent-centric business framework, if I may say so myself, for challenging your current marketing strategy.
What intent is your current marketing content (tv, digital, ads, emails) targeting? What happens once your ads meet that intent? What meaningful content are you publishing, on and offline, to engage audiences before and after the BUY NOW (!) moment? Is your measurement aligned with the intent your marketing is targeting, or are you judging a fish by its ability to climb a tree? How do you know?
Shifting to See-Think-Do-Care is the single biggest force multiplier when it comes to your marketing. Help shift your organizational thinking to the current century in 2019.
M7. Your marketing budget allocation can be improved anywhere from 50% to 50,000%.
Allocating budgets is the hardest decision a Senior Marketer will make. Most will use strategies like Digital had 27% of budget last year, this year we should do between 28 and 30%. History, gut-feel, inter-company-politics, etc. are primary reasons why this silly mindset is pervasive across companies.
A better way? Profitable opportunity size.
I don't think you can argue with the first part: Invest where you make more profit. The second part takes a bit more work. It comes from plotting diminishing margin curves with confidence intervals. In English: How high can the investment goes before every $1 you invest returns less?
You are a Marketer, so it's unlikely that you'll plot these curves. Make it a priority for your Analytics team to do so; without them massive chunks of your budget is being flushed.
(Also, see obsession #10 on the Analytics list.)
M8. A grandmother's Marketing strategy for grandmothers only.
A bit provocative, but I want to challenge how most Marketers just make little tweaks to their strategy. The bigger the company, the more that this pernicious problem exists. Don't let that be you, and allow me to share two views that'll challenge your reality.
Here's the average time spent per day by US adults with media devices…
My humble description of a "grandmother's marketing strategy" is the bar on the right (65+).
It is eminently sensible for our marketing for our fellow 65+ aged Earthlings to be reflective of the implications of that right-most bar.
The problem arises when our entire marketing strategy is an extension of that right-most bar. For our entire marketing strategy to be structured on that 6:55 you see above, when our products and services are not 65+ centric is… A bit silly. Perhaps even reflective of failing our fiduciary duty.
Note the difference in total media consumption (time, place, device, more). Note the products and services your company currently offers. Reflect on this: How misaligned is your current marketing strategy?
I get really excited about something super-cool, but subtle, in the data above: The implication of the difference between active vs. passive consumption!
The difference between leaning-back and letting content wash over us vs. leaning-in and pulling content you desire is huge. It dramatically changes what your marketing should be solving for (beyond the obvious investment alignment by platforms issue).
One more reality-check for your 2019 Marketing strategy: Here's a helpful deep drive into the shifts in consumption of TV across US adults – in just six years (!!)…
This possibly explains why Toyota's entire Marketing strategy seems to be TV-centric (with the incredible frequency of 48 per day per person here in the bay area!). It seems Toyota is only trying to sell cars to 65+ (whose TV watching has actually increased).
In 2019, resolve to align your marketing strategy with your 1. products 2. goals 3. audience, and 4. amount of expressed intent on the platform.
Credits: Originally created by Sara Fischer of Axios, the first graph is via my buddy Thomas Baekdal's newsletter. 100% of you need to sign up for it. The second chart is from the lovely team at The Economist.
M9. Suck less more.
Every campaign you are currently executing can be made to suck less – especially if you think end-to-end experience.
Ex: Expedia's emails are so long they always trigger "[Message clipped] View entire message." Suck less and maybe use my past behavior to send shorter emails so I know you care about me?
Ex: Nordstrom sends me one email a day with exclusive deals – how many clothes do they think I need? Suck less and maybe send me one a month? Or, base it on shopping patterns in store to deliver delight and not just a deal?
Ex: Macy's email I just received (titled "Resolution #1: get an extra 20% off before it ends") has promotions for Women, Men, Shoes, Bed & Bath, Kids, Juniors, Jewelry, Plus Sizes, Handbags, Home, Kitchen, Beauty. All above the fold. Below the fold: Large pictures with promotions for White Bedding, Biggest Underwear, Biggest Mattress (yes again), Best Face Forward, 25% off Adidas, Macy's presents the Edit, Fresh Pastels (the image does not make clear what this is), Free, Fast Pickup. PHEW! This can be unsucked at so many levels, with just a little bit of love and focus.
Ex: Even really good programs can use sucking less. Companies like Google and Microsoft have so many divisions. Each team/department optimizes for itself, emails are pretty good, hence each thinks they are doing really well. But, if you flip the lens to me – the recipient – I get a lot of email from each company. I wish someone at G/M would track Emails Sent/Humans Sent To, and reflect on the sad reality. It would create a culture of Marketing with me at the center instead of a company department – you can imagine the benefits.
I'm using email marketing as an example of activating the power of suck less because I love email marketing. It is an effective and profitable strategy. It has loads of behavioral data available. It needs a comparatively small team to execute well. Yet see how much opportunity there is to suck less at even the largest companies.
Substantially bigger opportunities to suck less exist in all other Marketing you are doing. TV. Print. Radio. Display (omg, sooooo much opportunity!). Video. Website. Mobile app. Everything else.
All you need to do is take a quick peek under the covers.
Your 10x goal for 2019: For every $1 invested in chasing a shiny object (VR ads! Influencer marketing!!!), invest $10 in sucking less in existing large clusters of your Marketing.
Profits that follow will also be that lopsided.
One last bit, culture eats strategy for breakfast. Create a quarterly Most Unsucked Team award, and celebrate this dimension of success. Incentives matter.
M10. Bring your great taste and expectations to work.
You can easily recognize when something is mediocre – even when others put lipstick on the pig and run it around the organization as the greatest success of the month.
You know what exceptional looks and feels like – you are not just a Marketer, you are an intelligent customer.
Yet, my experience is that most Marketers stay in their lane. Often, company cultures encourage that non-beneficial behavior.
In 2019, speak up.
You have great taste. Don't leave it at home when you leave for work.
Speak up.
When you see low quality work being pushed out by your Marketing organization… Create alternative mocks. Push for your version of the brand's tag line (not the generic MBA buzzword puke-fest). Ask for a better balance between Earned-Owned-Paid marketing. Politely challenge your Leader's assertion that creative x is better because he feels like it will be. Recommend experimenting with reckless ideas, instead of directly putting 30% of the budget on them. If you see lipsticked pigs being paraded around as exceptional examples, humbly, privately, flag the corrosive implication on culture to the most senior leader who'll listen to you.
Speak up.
You deserve to be heard.
When you speak, it'll give others around you the courage to speak up as well. Smart people tend to run in packs.
That’s it. :)
A slight repetition: Reflect deeply on the impact of the 10 x 2 obsessions in your unique business environment. Then, distill down to a total of ten you’ll focus on in the next twelve months. Finally, put a start and expected end date for each item. If you get through the list, you would have contributed a step change to your company’s bottom-line, and discovered unexpected personal joy.
As always, it is your turn now.
If you had already identified obsessions for Analytics and/or Marketing for the next twelve months for yourself, what obsessions did you choose? I’m super curious. Are there a couple in my lists above that would be particularly impactful in your company? Some of my recommendations are quite straight-forward, what do you think get’s in the way of focusing on them?
Please share your obsessions, tips, culture-shifting strategies, and critique via comments below.
Thank you.
New year resolutions tend to be fluffy and unachievable. This is a really serious list of ideas for us to focus on. I also feel with focus our company can achieve a minimum of half. We are taking that half, and making a plan.
The most difficult obstacle in our case has been trying to get enough of a pause from all the new stuff coming down the pike to do things that are important but not sexy enough for leadership.
Thank you for keeping it real.
An amazing collection of actionable ideas!
I can't decide if M8 or M9 is my absolute favourite. They are both so powerful, in some way in opposite directions. With marketing portfolio in M8 you are advocating pushing forward in new directions, while with M9 you are asking us to go deeper into what we are already doing. There is a good kind of inherent tension between the two.
My first action though is to more closely look at the emails we are sending out. -)
As someone focused primarily on analytics bullet number 9 to move away from custom attribution modeling was a pause point for me. We use custom attribution modeling to move our clients beyond last click.
Is there more detail on how the data driven models work? Could you please share any helpful links?
Great post.
Erik: My perspective on attribution is that for the most part moving beyond last-click means you are automatically less wrong. So, you are on the right path. I recommend moving to data-driven faster as there is so much complexity in user behaviour, that it is hard for us (humans) to understand it across millions of data points across months and months of data.
Here are two helpful links with a bit more detail and some case studies:
https://support.google.com/google-ads/answer/6394265?hl=en
https://www.blog.google/products/ads/data-driven-attribution-results/
Avinash.
Hi Avinash, the chart by Economist is really amazing. Clear and straight to the point.
My obsession for 2019 is learning AI and Data Visualization. So far in this journey, I learned Python Programming, Pandas library for data manipulation, Data Visualization using matplotlib. At this moment actually, I am writing an article on how to do data viz with python. Looking forward to learning statistics and SQL programming before applying for entry-level Data analyst job in April. And will keep working towards my AI and Data Viz goal.
Is there anything I need to aware off before applying for the job?
Thanks for reading it. Happy new year to you and your family.
With regards,
Bhola
Bhola: I would guess that with the skills you are acquiring, you might be able to apply for more than an entry-level Data Analyst job. I might encourage you to consider roles in Data Science as well, where modeling and visualization require Python skills.
Two things to consider as you continue your wonderful evolution:
1. Being proficient in R, with the other skills you've acquired would be considered quite desirable (and value added for you).
2. Create a portfolio of your work, publish it online. There are a number of free data sets out there (for web analytics the Google Analytics Store Demo account which is free and accessible to all, and lots of data at the UN). Use your skills to analyze it, create stories of insights. Having a portfolio you can point to for any interviewing company will cause you to stand apart.
All the best!
Avinash.
A really insightful post to jump start the new year. Thank you.
Our big focus this year is to finally try to crack tracking offline conversion as comprehensively as we can. Thus far we have done controlled experiments to get some sense for performance. We are tackling big data merges to do better in 2019.
I am surprised killing vanity metrics did not make your list. Always one of your clarion calls here on Occam's razor.
Andrew: My prayer is that people who read this blog would have realized the abject uselessness of vanity metrics. Hence, I decided to skip it!
But you are right to underline its importance.
Avinash.
Thanks Avinash, lots of great ideas to digest here!
Two points that resonated the most for me were A9 [moving to DDA attribution] and M7 [making spend decisions based on marginal profit.] Trying to combine both items, we can conceptually think of a solution that would attribute credit to all channels assisting in a conversion, then incorporating each channel's spend into our calculation, while closing in with variable contribution product margin for items sold.
One of the challenges here is distinguishing between attribution vs. contribution of said channels. DDA does not have visibility to non-digital channels and stops short of offering a holistic view of our marketing efforts, yet along business. After all, DDA's algorithm only considers various marketing touches and doesn't incorporate organizational inputs (pricing strategy, inventory levels, product quality, customer support, UX [including already—mentioned bounce rate improvements across the board] etc), nor does it reflect external factors.
Do you have any thoughts on how we can possible solve for all of the known unknowns (in addition to the obvious, yet gut-feeling/assumption-based models of running various multi-variate analyses [A10])?
Alex: Your comment is asking if there is a grand unified theory that explains the universe, I'm afraid there is not. :)
Our approach is to use different approaches to answer questions that are quite different (altitude, detail, accuracy, strategy, purpose, etc.).
So, you can send offline conversions and some data about other channels (digital for sure) into DDA and it will accommodate for that behavior as it provides analysis.
Once we get into understanding holistic view of our marketing efforts, we start using Media Mix Models based on Bayesian methods to understand impact on Sales (or other outcomes) across the entirety of the online and offline portfolio (and once we have enough data cross two or three years, we also turn this around to do predictive modeling for upcoming budget allocations to improve yield based on marginal curves).
Large scale controlled experiments are another ally, when we are in the known unknown or unknown unknown clusters. We'll take $4 to $7 mil of the budget and run large scale Matched Market Tests to answer some deeply fundamental questions that we simply don't have answers for. Do enough of these and we can identify meta patterns that help us understand the impact of promotion strategies, targeting, retail presence, and more, at a level that is simply magical.
I could keep going on (deep learning algorithms are our latest obsession), but you catch my drift.
Please keep asking the questions you are asking, keep seeking the truth. It is out there. :)
Avinash.
Great post Avinash, thanks!
My in-the-weeds question: does DDA work if you have goals/conversion events defined in your SaaS property for Google Analytics and your website as a different GA property?
Jon: I believe the answer is no. There has to be unified account for the behavior to be stitched together.
I encourage you to reach out to an authorized consultant: https://measurementpartners.google.com/ They can quickly assess and recommend the optimal solution.
Avinash.
I'm always amazed by your ability to strike the perfect balance between high-level strategic advice and very specific deep tactical connections. Suck less more is a good example. The reviews of very specific tactics by the companies you mention is devastating – how could they not know -, it illustrates so clearly the progress yet to be made on something we have all been working on for a couple of decades now.
I'm off to deep review our email campaigns to unsuck them!
Hey Avinash,
My obsessions,
1)Virtualization
2)Improvising bounce rates
3)Sucking less – thats a wonderful concept, i loved it :)
4)trying differnt things and allocate the marketing budget.
I love this blog a lot man :) god bless you :)
Regards
Joby John
Joby: An excellent list, I'm confident you are going to have an exciting and fruitful 2019.
-Avinash.
PS: Just make sure your list for #4 does not end up with…
#kidding
This really got me! haha
2019 is for you AK
Thanks so much Avinash.
IMO A10 is super important, its what science is built on, but even scientists struggle with. There's a book called Statistical Rethinking by McElreath, where he interestingly compared our models to simple, reliable, but dumb golems and how he found so many studies were throwing the statistical treatment they thought was best without thinking if it really fit their data – sometimes the only data of its kind in the world!
There is very little content on actual *causality*, the thing we are after, not just correlation. I found out about that recently and I'm consuming information on that as much as I can.
That's where out of the box ML (or even just plain safe stats) can trap you – or never fit well to your data. There's a point where you really need to know all the cogs in the machinery. Asking for error bars can be hard, like finding the actual revenue an assisted revenue your website is generating – but in the same way I think it can promise a similar bliss of clarity.
When are we going to be graced with your Web Analytics 3.0, or 2.5, or maybe even an online course? :)
Skipping to data-driven attribution would indeed be awesome – but simply too expensive for 99 % of companies. At least in a country like Finland.
Thank you, Kaushik, for sharing an amazing obsession list of marketing and analytics.
I strongly agree with your #M2 (Put up or shut up time for your social media strategy) obsession. Organic reach on social media is super low these days & does not always justify the marketing spend.
My Obsession for 2019 will be:
1. More efforts in Data Analysis.
2. Learn more about Google Data Studio
3. Continue working on my 2018 obsession list :)
Thank you for this post, Avinash!
I found the Nielsen study you referenced in section M8 particularly interesting, I brought it up in a meeting with one of my broadcast TV clients, and it was an eye-opener!
– Justin
As always, you are the best blog for marketing analytics. Thanks for the great post.
Cheers,
Leonardo Candoza
Hi Avinash, as always expose your concepts clearly and always go to the point with your articles. Also this time you opened my mind on what to do in the near future and how to optimize my technical actions.
Great article.
220% Agree with you on M9, We don't need to push irrelevant ads to irrelevant audience, but I think it important to use different marketing services to help your site grow when organic traffic is low.
Great article by the way.
Great post Avinash, thanks!
love this blog a lot man :) god bless you
I just read the article now. It helps me to stand up and say that the marketing relevant decisions like budget distribution along the customer journey have to be data driven and cannot result from a gut decision (to land at the grandmother's model). Advertising, on the other hand, has to be more daring (guerilla marketing) and arouse emotions and should never be boring.
I wish all of you who have fun analyzing your data a lot of joy and success, so that you can underpin your opinions with results.
Regards from Austria
An amazing obsession list of marketing and analytics is shared here that exposes every strategy of marketing clearly.
It makes for a must-read article. Thanks for sharing it.