If there is one slogan that your analytics team should live by, or indeed, die by, it is trust, but verify.

The spirit of it is built into the job description: never take anyone's word for granted.

I urge you not to give in to pure faith. (Without proof.)

Push back.

I am not implying that perfection in measurement is possible. Don't be the person who insists on 99.9% measurement accuracy - except when lives are threatened.

But, try not to accept 0%. Pure faith. Because in most cases, we can do better.

To make this idea concrete for you, let me share one common example where faith currently rules in many Marketers, and help illustrate how you can practice trust, but verify.

Context

An impression is the delivery of an ad (banner, video, billboard, television). Depending on the publisher, that ad unit could be partially, fully or not all have been viewed.

If there is an interaction with the ad unit, a click or active engagement is recorded. Conversions are recorded if they happen.

If there is no interaction with the ad unit, but a tracking mechanism detects that 35 days later you visited the site and converted, a Viewthrough conversion is recorded with credit claimed by that original impression.

Viewthrough Value

I am sure your brand media agency has worked very hard to convince you of the value of display advertising purely based on impressions delivered.

Look, all these people are driving by the highway looking at your ad, then they walk into the store.

Or, look, all these people are seeing your ads on Facebook, they took no action but they came to your site via organic search and converted which should be credited to Facebook!

My learning from years : Viewthroughs are the backstop of every Marketer when they know there is no provable value. (Click to tweet this.)

Color me skeptical.

My recommendation: Trust that your brand agency has good intent, but verify the results rather than accepting their faith.

And here's the beautiful thing, you can verify with relative ease when it comes to Viewthroughs. You have two excellent choices...

1.

The simplest strategy is to do a controlled experiment. For example, you can segment by Geo or publisher. Or, you can run the experiment for a particular campaign, offer, audience, or any combination that you can imagine.

Controlled experiments, done well, will help you measure incrementality. Across cells they will show that Viewthroughs – i.e., an ad that potentially has been seen by a human, but has no recorded interaction - caused incremental conversions.

Trust, but verify.

Without controlled experiments the massive display/video marketing budget is simply being judged based on faith, or, perhaps worse, judged by victory being defined as an ad being delivered.

It is not easy to do controlled experiments. It is completely worth doing controlled experiments.

[Bonus Read: Measuring Incrementality: Controlled Experiments to the Rescue!]

2.

If the optimal method of measurement is not possible, due to whatever worthy reason, there are other choices.

For example, if you are using a tool like Google Analytics Premium, you can look at the attribution reports in the Multi-Channel Funnels folder. They include impressions, at least for your display ads being delivered via Google's stack.

The report will help you surmise the value of impressions (basically viewthroughs). Use this easily available information to challenge faith with data. Based on this initial proof, go to strategy #1.

.#winning

If your company is in a situation where neither of the above two ideas is possible, I recommend stopping any expenditure that is based purely on faith that viewthroughs are valuable. There are simply too many smart ways in which you can spend that money to deliver real measurable business impact.

Bonus.

The mindset of trust but verify applies in all other business scenarios as well.

Why believe that Lab Testing is the all mighty god of UX? Take some lab tests that failed along with those that passed, and measure how they do in the real world.

Trust, but verify.

SEO is the bedrock of all marketing - online or offline! Why not try to measure incrementality? Or, cruder, take all the money you put into SEO and put it into Email marketing for three months (or whatever other rich opportunity). Does your company collapse? Maybe it does, there's your verification of the SEO assertion.

Trust, but verify.

Your boss' boss wants you to hire their third cousin twice removed, and you don't feel you have a choice not to... Hire the cousin as a temp to perm.

Trust, but verify.

The use cases are endless. You are an Analyst. Verification is your middle name - it is not for pretty much all other departments at your company (except Accounting!).

Bottom line: Analytically-minded friends don't let friends make business decisions solely based on faith. :)

Avinash.

PS: I am not saying there is no room for faith in context of new ideas. There is. In Analytical roles, our job is to validate faith, to prove hunches are right (or not).