TMAI #474: 🎀 Rules for Dashboard Revolutionaries!

[The Marketing Analytics Intersect, by Avinash Kaushik] [1]

TMAI #474: RULES FOR DASHBOARD REVOLUTIONARIES!

[ Web Version [2] ]

The purpose of a dashboard is:

80%: To recommend actions

20%: To provide data (where the Analyst does not have biz context)

If I had to hazard a guess, I would offer that most dashboards you

bump into have this distribution:

20%: Recommended actions

80%: Regurgitated data

Dashboards as data pukes also reflect a cultural incitement:_ I do

not trust you, Analyst/Data Team, just give me all the data and I'll

figure out what's going on._

The problem of course is that you don't hire Extremely Senior Leaders

(ESL) because they are fantastic Analysts, neither do they have access

to all the data to segment, drill anywhere, and validate hypotheses to

find insights. Obviously, they don't have time to do that either -

they have BIG compelx jobs.

Implication? Dashboard review sessions are about asking a hundred

questions about aggregated data, poking holes in it, and.... No

action.

Effective dashboards drive action. To accomplish that, they don't

contain data pukes illustrating_ what happened_. Instead, they lead

with_ why it happened_. Quickly followed by_ what should we do_.

Ending with_ how much impact will I see if I say yes to the action._

_Why > What > How Much._

In language you've seen me use: Insights. Actions. Business Impact.

IAbI.

To help you convert each module of data on your dashboard into a

high-impact thing of beauty, here's a collection of specific

recommendations curated from my successes and failures.

1. PICK KPIS FOR THE DASHBOARD AND NOT METRICS.

A METRIC is a number. Usually good for tactical and diagnostic

purposes. Moves small things.

A KPI is a metric that helps you understand how you are doing against

your objectives! Flows directly to your business bottom line.

Impressions, tickets opened, page views, bounce rate, footsteps,

inbound calls, visits, likes, employees hired, cost-per-click,

transportation costs, number of tweets, Facebook followers, leads

response time, website domain authority, email open rate… ARE ALL

METRICS.

Let your Analysts, Agency, Managers of Marketing tactics should use

Metrics to improve bits.

Profit, cost per acquisition, leads close rate, Net Revenue Retention,

checkout abandonment rate, incremental sales, Earned Growth Ratio,

accounts receivable, absenteeism rate, churn rate, project schedule

variance, monthly active users… ARE ALL KPIS.

Dashboards should have KPIs because you want executive attention

focused on levers that most closely impact the bottom line. Everything

else is someone else's job.

[BONUS TIP: If you have more than six KPIs on your Dashboard, you’ve

not interrogated enough what really matters to the business.]

2. (WHERE POSSIBLE) SHOW TRENDS OVER TIME.

Everyone overreacts to the now. Especially when the now is missing

context.

In part, this is behind my dislike of pies, treemaps, speedometers,

et. al. All of them provide snapshots in time. Usually, KPIs aggregate

detail up, which makes snapshots in time even more useless.

Trend your KPIs over time, _m’kay?_

A tip from a lifetime of experience: It is important to use optimal

time splits in your trends.

Daily trends are good up to the 90-day horizon, then switch to weekly

for up to 13 weeks (so you have _same week last quarter_), then switch

to monthly for up to 13 months (to get s_ame month last year_), and

beyond that use quarterly.

Your goal with this recommendation: Make it easy to notice _is this

data point an anomaly, are things really going down/up over time_.

Prevents premature celebrations/freak-outs.

3. SEGMENT, SEGMENT, SEGMENT.

If I loved segmentation any more, I would marry it.

Smartly selected segments spew insights, they accelerate

answering_ why_ you are seeing the_ what_. From there to_ let's do

x, y, z_ is so much easier.

On occasion, smart segmentation can also help answer _WHY _you might

be seeing the performance you are - and when that happens, what action

to take.

Here’s an example. For your SAAS product, your dashboard will surely

report the KPI Cancellation Rate. Good. But, in aggregate, the number

going up or down is less helpful.

Instead, why not have your dashboard module show a segmentation of

_why _people are canceling...

[Segmented Cancellation Rate.]

Following advice #2, the little arrows hint at the trend. (X-axis

hidden for privacy reasons.)

If you can’t segment a data module, you should have a _VERY_ good

reason not to delete it.

4. SHOW THE END-TO-END (COMPLETE) PICTURE.

In my writings on digital analytics, I’ve advocated for a framework

called ACQUISITION, BEHAVIOR, and OUTCOMES (ABO) to ensure you

always show the end-to-end picture to an ESL - helping them not just

be informed but actually be smarter.

An example: If you show the results for Cost Per Sale (an Acquisition

KPI), and it went up by 400%, everyone would freak out.

Understandably. But. If you showed Acquisition and Outcome, CPA &

Revenue, they might observe that while CPA went up by 4x, Revenue went

up by 9x!

The frown turns upside down.

ABO. Always.

Here's one of my fav examples, from one of my students... It brings

together ABO and Segmentation recommendations into one helpful

dashboard module...

[End to End Segmented Data Module.]

You can use recommendation #7 (below) to streamline the visual to

sharpen the ESL focus even more. But, you can see the usefulness of

the end-to-end view clearly.

5. COMPARE TO TARGET / BENCHMARK / COMPETITORS.

Dashboard that fail at IAbI, fail at this simple task:_ Is the

performance I'm looking at good? Bad? Otherwise?_

Ex: Your dashboard shows a 7% conversion rate. Is that good?

7% vs. a target of 5% is fantastic.

7% vs. an industry benchmark of 11% is less so.

7% vs. a direct competitor’s conversion rate of 9% implies _not bad,

but we can do a bit better_.

Never parade your KPIs naked.

My preferred order: 1. Targets. If not, 2. Direct competitor

performance. If not, 3. Industry benchmark.

Because setting targets is a clear sign of intentionality - and that

is a winning strategy.

6. COVER ESL BLIND SPOTS.

Our ESLs are very good at the obvious. Great dashboard artists figure

out the blind spots of the ESL, fill them in.

An example: CMOs obsess about Paid Media too much. A common blind spot

is not realizing that Owned and Earned Media drive a majority of the

Outcomes.

Hence, my dashboard modules at the very minimum report Conversions &

Assisted Conversions for each channel. That helps illuminate two blind

spots at the same time:

A. A whole lot of credit that Paid Media (PPC, Meta...) is claiming,

were assisted by Owned (Email, SMS..) and Earned (SEO, Referrals...)

channels.

B. Owned and Earned are responsible for 60 - 80% of all Conversions.

Now the ESL has more intelligence to allocate headcount, budgets, and

rank order priorities.

Bonus Gift: This helps illuminate the consumer journey, every visit to

the site is NOT an opportunity to convert. That in turn leads to

inquiries about helpful metrics like_ Days to Conversions _and

_Visits to Conversion._ Which will change your landing pages, your ad

creative, your... So. Much. More.

Get to know your ESLs. Identify their blind spots (which everyone has,

that is not a bad thing).

7. FIGHT THE BLOAT: FOCUS ON MOVERS, OUTLIERS, AND THE SIGNIFICANT.

Two common dashboard problems: Strategic: It is very hard to stick to

the six or so KPIs that truly drive the business - dashboards balloon.

Tactical: Each data module ends up with too many rows and columns -

values in which don't really change WoW or MoM.

Fight the bloat.

Statistics can be your first stop. It can help you identify the

biggest movers (up or down) or rows that happen to represent threats

on the horizon. For ESL dashboards, I'll often focus on showing data

for values three standard deviations from the mean.

You don’t have to focus on the outliers, you can do two standard

deviations.

You can also choose to show data that is 90%, or higher, statistical

significance.

Likewise, you can also apply more advanced algorithms that’ll sort

your data to draw out nuance and also reduce bloat. One of my favorite

filters in GA (now dead) was the weighted-sort [3], which surfaces

what was hidden below the top ten or twenty rows AND shows promise of

being important for the business.

8. DON’T BE BORING / MAKE DATA _CONNECTABLE_.

Look, I find no shame in admitting that data can sometimes be boring.

Ok. Fine. A lot of the time.

We can make it less boring.

Or, at least more memorable.

Do you know my definition of bounce rate? _I came. I puked. I left._

Less boring and more memorable than _the percentage of sessions with a

single hit_. Yes?

Another example: Here were the segments for Site Engagement as "Less

than one page", "Three pages or less," "More than three pages & did

not meet goal," "Placed an order," and "Placed more than one order"

I'm bored typing them! Imagine the snoozing for the ESLs.

Unboring version: Abandoners, Flirters, Browsers, One-off-Wonders,

Loyalists.

The first time folks saw the descriptions, they smiled. Even better,

they immediately got the behavior being described. They reviewed the

data a little more carefully. After all, data reflects behavior, and

behavior is about people… Why not connect the dots in the actual

presentation of the data?

Then, they started to use these terms to talk about the various

segments.

_How can we create more one-off wonders?_

_Who is going to focus on making sure the Flirts become Browsers?_

Productive discussions. Actions. Progress.

9. VISUALIZE SIMPLY, FOCUS OBSESSIVELY.

You've read a million words by me on this topic. Ok, a slight

exaggeration.

But, I do tend to write a lot about data visualization because it is a

key contributor to _death at the last mile_ - the distance between

completed analysis and someone taking action.

Nowhere is it more true than on dashboards.

An example: Both the left and the right visuals are communicating the

same information...

The one on the right demands so much less re brain processing to

internalize. No?

It unpacks different stories that make the patterns easier to see.

You can keep pushing to make things as simple as possible (but no

simpler).

For example, one of my beliefs is to never send a visualization to do

a table’s job.

Same data as above, this time as a table...

Depending on your leadership team, the table might work better than

the visual. It is probably not a stretch to say that they are both far

superior to the original.

Visualize simply. Focus obsessively.

Checkout 14 more examples in this post on Occam’s Razor to

practice: Great Storytelling With Data [4].

10. STAY RELEVANT: KILL KPIS.

Once finalized, we tend to treat dashboards as carved in stone.

Everything evolves. Businesses change, people come and go, high-level

priorities evolve, we become smarter, our competitors think of new and

clever things, and so forth.

Why should our dashboards and KPIs stay the same over the span of a

year?

The best dashboards have a high bar to change, but change they do over

time. Around 40% of your KPIs will likely never change (they are just

that important). I find that 20% OF THE KPIS WILL CHURN over the

course of a year - at the minimum.

Apply the three-layers of the so what [5] test to fuel this metrics

lifecycle process:

Define > Measure > Analyze > Action > Improve/Eliminate.

Pretty picture, with some tips...

[DMAAI!]

I’ll be the first to admit that this is incredibly hard to put into

practice - organizations (like humans) resist change.

But.

Ensuring the rigor above, ensuring evolution, will ensure your

dashboards remain relevant and useful.

BOTTOM LINE.

If having access to data is all it took to make the world a better

place, the world around you would be a better place.

Data needs careful curating, smart context, and a pinch of

storytelling.

If you apply a majority of the rules for dashboard revolutionaries,

you’ll convert your dashboards into agents of change.

_Carpe diem._

-Avinash.

Thank you for being a TMAI Premium subscriber - and helping raise

money for charity.

Your Premium subscription covers one person. It's fine to forward

occasionally. Please do not forward or pipe it into Slack for the

whole company. We have group plans, just email me.

[Subscribe [6]]  |  [Web Version [2]]  |  [Unsubscribe [7]]

[8]

[9]

[10]

©2022 ZQ Insights  |  PO Box 10193, San Jose, CA 95157, USA

Links:

------

[1] link

[2] link

[3] link

[4] link

[5] link

[6] link

[7] link

[8] link

[9] link

[10] link