Skip to content

Tag: digital marketing

Example of Google’s Moral Hazard: Pooling in Ad Auctions

Google has an incentive to group advertisers in ad auction even when this conflicts with the goals of an individual advertiser.

For example, you’d like to bid on ‘term x‘ and would not like be included in auctions ‘term x+n‘ due to e.g. lower relevance, your ad might still participate in the auction.

This relates to two features:

  1. use of synonyms — by increasing the use of synonyms, Google is able to pool more advertisers in the same ad auction
  2. broad match — by increasing the use of broad match, Google is able to pool more advertisers in the same ad auction

Simply put, the more bidders competing in the same ad auction, the higher the click price and therefore Google’s profit. It needs to be remarked that pooling not only increases the CPC of existing ad auctions by increasing competition, but it also creates new auctions altogether (because there needs to be a minimum number of bidders for ads to be launched on the SERP).

A practical example of this moral hazard is Google’s removal of ‘do not include synonyms or close variants‘ in the AdWords campaign settings, which took place a couple of years ago.

There are two ways advertisers can counter this effect:

  1. First, by efficient use of negative keywords.
  2. Second, by resorting to multi-word exact matches as much as possible.

In conclusion, I always tell my students that Google is a strategic agent that wants to optimize its own gain — as far as its and advertiser’s goals are aligned, everything is fine, but there are these special cases in which the goals deviate and the advertisers needs to recognize them and take action.

Modern Market Research Methods: A Startup Perspective

EDIT: Updated by adding competitive analysis, very important to benchmark competitors.

EDIT2: Updated by adding experimentation (14th April, 2016)

Introduction

Somebody on Quora was asking about ‘tools’ for validating viability and demand for a startup’s products.

I replied it’s not a question of tools, but plain old market research (which seems to be all too often ignored by startup founders).

Modern market research methods

In brief, I’d include the following options to a startup market research plan:

  1. market statistics from various consultancy and research institution reports (macro-level)
  2. general market (country, city) statistics generated just for your case (macro-level à la PESTLE)
  3. competitive analysis, i.e. benchmarking existing solutions — will help you find differentiation points and see if your “unique idea” already exists in the market
  4. (n)etnography, i.e. going in-depth to user communities to understand their motivations (micro-level, can be done offline and online)
  5. surveys, i.e. devising a questionnaire for relevant parties (e.g., customers, suppliers) to understand their motivations (just like the previous, but with larger N, i.e. micro-level study)
  6. customer development, which is most often used in B2B interviews as a presales activity to better understand the clients’ needs. Here’s an introduction to customer development (Slideshare).
  7. crowdfunding, i.e. testing the actual demand for the product by launching it as a concept in a crowdfunding platform – this is often referred to as presales, because you don’t have to have the product created yet.
  8. experimentation, i.e. running different variations against one another and determining their performance difference by statistical testing; the tests can relate to e.g. ad versions (value propositions, messages) or landing pages (product variations, landing page structure and elements). Here’s a tool for calculating statistical significance of ad tests.

So, there. Some of the methods are “old school”, but some — such as crowdfunding are newer ways to collect useful market feedback. Experimentation, although it may appear novel, is actually super old school. For example, one of the great pioneers of advertising, Claude Hopkins, talked about ad testing and conversion optimization already in the 1920. (You can actually download his excellent book, “Scientific advertising“, for free.)

How to combine different methods?

The optimal plan would include both macro- and micro-level studies to get both the “helicopter view” and the micro-level understanding needed for product adoption. Which methods to to include in your market research plan depends on the type of business. For example, crowdfunding can be seen as a market validation method most suitable for B2C companies and customer development for B2B companies.

The punchline

The most important point is that you, as a startup founder, don’t get lured into the ‘tool fallacy’ — there’s no tool to compensate for the lack of genuine customer understanding.

Dr. Joni Salminen holds a PhD in marketing from the Turku School of Economics. His research interests relate to startups, platforms, and digital marketing.

Contact email: [email protected]

Dynamic Pricing and Incomplete People Information

One of the main problems in analytics is the lack of people information (e.g., demographics, interests). It is controlled by superplatforms like Google and Facebook, but as soon as you have transition from the channel to the website, you lose this information.

So, I was thinking this in context of dynamic pricing. There’s no problem for determining an average solution, i.e. a price point that sets the price so that conversion is maximized on average. But that’s pretty useless, because as you know averages are bad for optimization – too much waste of efficiency. Consider dynamic pricing: the willingness to pay is what matters for setting the price, but it’s impossible to know the WTP function of individual visitors. That’s why aggregate measures *are* needed, but we can go beyond a general aggregate (average) to segmentation, and then use segment information as a predictor for conversion at different price points (by the way, determining the testing interval for price points is also an interesting issue, i.e. how big or small increments should you do —  but that’s not the topic here).

Going back to the people problem — you could tackle this with URL tagging: 1) include the targeting info into your landing URL, and you’re able to do personalization like dynamic pricing or tailored content by retrieving the targeting information from the URL and rendering the page accordingly. A smart system would not only do this, but 2) record the interactions of different targeting groups (e.g., men & women) and use this information to optimize for a goal (e.g., determining optimal price point per user group).

These are some necessary features for a dynamic pricing system. Of course then there’s the aforementioned interval problem; segmentation means you’re playing with less data per group, so you have less “trials” for effective tests. So, intuitively you can have this rule: the less the website has traffic, the larger the increments (+/-) should be for finding the optimal price point. However, if the increments become too large you’re likely to miss the optimal (it gets lost somewhere in between the intervals). I think here are some eloquent algorithmic solutions to that in the multi-armed bandits.

A Quick Note on Bidding Theory of Online Ad Auctions

Introduction

This is a simple post about some commonly known features of online ad auctions.

Generalized second-price auction (GSP) is a mechanism in which the advertiser pays a marginally higher bid than the advertiser losing to him. It encourages the bidder to place a truthful bid, i.e. one where the price level is such that marginal returns equal marginal cost.

Why is this important?

Simply because:

truthful bid = incentive to bid higher

In other words, if you know a bidder behind is bidding say 0,20 € and you’re bidding 0,35 €, under a standard auction you’d be tempted to lower your bid to 0,21 € and still beat the next advertiser.

In any case you wouldn’t directly know this because the bids are sealed; however, advertisers could programmatically try and find out other bids. When you’re using GSP, manually lowering bids to marginally beat your competition is not necessary. It’s therefore a “fair” and automatic system for pricing.

Of course, for the ad platform this system is also lucrative. When advertisers are all placing truthful bids, there is no gaming, i.e. no-one is attempting to extract rents (excessive profits) and the overall price level sets higher than what would take place under gaming (theoretically, you could model this also in a way that the price level is at equal level in both cases, since it’s a “free market” where prices would set to a marginal cost equilibrium either way).


Joni Salminen holds a PhD in marketing from the Turku School of Economics. His research interests relate to startups, platforms, and digital marketing.

Google and the Prospect of Programmatic

Introduction

This is a short post taking a stance on programmatic ad platforms. It’s based on one single premise:

Digital convergence will lead into a situation where all ad spend, not only digital, will be managed through self-service, open ad platforms that operate based on auction principles

There are several reasons as to why this is not yet a reality; some of them relate to lack of technological competence by traditional media houses, some to their willingness to “protect” premium pricing (this protection has led to shrinking business and keeps doing so until they open up to the free market pricing), and a host of other factors (I’m actually currently engaged in a research project studying this phenomenon).

Digital convergence – you what?

Anyway, digital convergence means we’ll end up running campaigns through one or possibly a few ad platforms that all operate according to the same basic principles. They will resemble a lot like AdWords, because AdWords has been and still is the best advertising platform ever created. Why self-service is critical is due to the necessity of eliminating transaction costs in the selling process – we don’t in most cases need media sales people to operate these platforms. Because we don’t need them, we won’t need to pay their wages and this efficiency gain can be shifted to the prices.

The platforms will be open, meaning that there are no minimum media buys – just like in Google and Facebook, you can start with 5 $ if you want (try doing that now with your local TV media sales person). Regarding the pricing, it’s determined via ad auction, just like in Google and Facebook nowadays. The price levels will drop, but lowered barrier of access will increase liquidity and therefore fill seats more efficiently than in human-based bargaining. At least initially I expect some flux in these determinants — media houses will want to incorporate minimum pricing, but I predict it will go away in time as they realize the value of free market.

But now, to Google…

If Google was smart, it would develop programmatic ad platform for TV networks, or even integrate that with AdWords. The same applies actually to all media verticals: radio, print… Their potential demise will be this Alphabet business. All new ideas they’ve had have failed commercially, and to focus on producing more failed ideas leads unsurprisingly to more failure. Their luck, or skill however you want to take it, has been in understanding the platform business.

Just like Microsoft, Google must have people who understand about the platform business.

They’ve done a really good job with vertical integration, mainly with Android and Chrome. These support the core business model. Page’s fantasy land ideas really don’t. Well, from this point of view separating the Alphabet from the core actually makes sense, as long as the focus is kept on search and advertising.

So, programmatic ad platforms have the potential to disrupt Google, since search still dwarfs in comparison to TV + other offline media spend. And in the light of Google’s supposed understanding of platform dynamics, it’s surprising they’re not taking a stronger stance in bringing programmatic to the masses – and by masses, I mean offline media where the real money is. Google might be satisficing, and that’s a road to doom.

Dr. Joni Salminen holds a PhD in marketing from the Turku School of Economics. His research interests relate to startups, platforms, and digital marketing.

Contact email: [email protected]

A major change in AdWords – How to react?

Introduction

Google has made a major change in AdWords. Ads are now shown only in the main column, no longer in the right column. Previously, there were generally speaking eight ads per SERP. For some queries, Google didn’t show ads at all, and additionally they’ve been constantly testing the limit, e.g. running up to 16 product listing ads per results page.

But what does that mean to an advertiser?

Analysis

The change means the number of ads shown per SERP (search-engine results page) is effectively reduced. Since the number of advertisers is not reduced (unless rotation is applied, see below), the competition intensifies. And since the visibility of search ads is based on cost-per-click auction, ceteris paribus the click prices will go up.

Therefore, logical conclusion is that when ad placements are cut, either CPC increases (due to higher competition) or impression share decreases (due to rotation). In the former, you pay more for the same number of visitors, in the latter you pay the same click price but get less visitors.

Why Google might in fact prefer ad rotation, i.e. curbing down an individual advertiser’s impression share (the number of times your ads is shown out of all possible times it could have been shown) is because that wouldn’t impact their return-on-ad-spend (ROAS) which is a relative metric. However, it would affect the absolute volume of clicks and, consequently, sales.

In some of my campaigns, I’m using a longtail positioning strategy where this will influence, since these campaigns are targeting positions 4+ which, as said, are mostly no longer available. Most likely, the change will completely eradicate the possibility of running those campaigns with my low CPC-goal.

Why did Google do this?

For Google, this is a beneficial and logical change since right column ads are commanding lower CTRs (click-through rates). This has two implications – first, they bring less money for Google since its revenue is directly associated with the number of clicks; second, as commonly known Google is using CTR as a proxy for user experience (for example, it’s a major component in Quality Score calculations which determine the true click price).

Therefore, removing the possibility of poorly performing ads while pushing the advertisers to an increased competition is a beneficial situation for Google. In the wider picture, even with higher click prices, the ROI of Google ads is not easily challenged by any other medium or channel, at least what I can see taking place in the near future.

However, for advertisers it may easily signify higher click prices and therefore decreasing returns of search advertising. This conflict of interest is unfortunate one for advertisers, especially given the skewed distribution of power in their relationship to Google.

(On a side-note, the relationship between advertisers and Google is extremely interesting. I studied that to some extent in my Master’s thesis back in 2009. You can find it here: https://www.dropbox.com/s/syaetj8m1k66oxr/10223.pdf?dl=0)

Conclusion

I recommend you revise the impact of this change on your accounts, either internally or if you’re using an agency, with them.

Dr. Joni Salminen holds a PhD in marketing from the Turku School of Economics. His research interests relate to startups, platforms, and digital marketing.

Contact email: [email protected]

The correct way to calculate ROI for online marketing

Introduction

This is a short post explaining the correct way to calculate ROI for online marketing. I got the idea earlier today while renewing my Google AdWords certificate and seeing this question in the exam:

Now, here’s the trap – I’m arguing most advertisers would choose the option C, although the correct one is option A. Let me elaborate on this.

The problem?

As everybody knows, ROI is calculated with this formula:

ROI = (returns-cost)/cost*100%

The problem is that the cost side is oftentimes seen too narrowly when reporting the performance of online advertising.

ROI is the ‘return on investment’, but the investment should not only be seen to include advertising cost but the cost of the product as well.

Let me give you an example. Here’s the basic information we have of our campaign performance:

  • cost of campaign A: 100€
  • sales from campaign A: 500€

So, applying the formula the ROI is (500-100)/100*100% = 400%

However, in reality we should consider the margin since that’s highly relevant for the overall profitability of our online marketing. In other words, the cost includes the products sold. Considering that our margin would be 15% in this example, we would get

  • cost of products sold: 500€*(1-0.25) =425€

Reapplying the ROI calculation:

(500-(100+425)) / (100+425) * 100% = -4.7%

So, as we can see, the profitability went from +400% to -4.7%.

The implications

The main implication: always consider the margin in your ROI calculation, otherwise you’re not measuring true profitability.

The more accurate formula, therefore, is:

ROI = (returns-(cost of advertising + cost of products sold)) / (cost of advertising + cost of products sold)

Another implication is that since the ROI depends on margins, products with the same price have different CPA goals. This kind of adjustment is typically ignored in bid-setting, also by more advanced system such as AdWords Conversion Optimizer which assumes a uniform CPA goal.

Limitations

Obviously, while the abuse of the ‘basic ROI’ calculation ignores the product in the cost side, it also ignores customer lifetime value from the return-side of the equation.

Dr. Joni Salminen holds a PhD in marketing from the Turku School of Economics. His research interests relate to startups, platforms, and digital marketing.

Contact email: [email protected]

Carryover effects and their measurement in Google Analytics

Introduction

Carryover effects in marketing are a tricky beast. On one hand, you don’t want to prematurely judge a campaign because the effect of advertising may be delayed. On the other hand, you don’t want bad campaigns to be defended with this same argument.

Solutions

What’s the solution then? They need to be quantified, or didn’t exist. Some ways to quantify are available in Google Analytics:

  • first, you have the time lag report of conversions – this shows how long it has taken for customers to convert
  • second, you have the possibility to increase the inspection window – by looking at a longer period, you can capture more carryover effects (e.g., you ran a major display campaign on July; looking back on December you might still see effects) [Notice that cookie duration limits the tracking, and also remember to use UTM parameters for tracking.]
  • third, you can look at assisted conversions to see the carryover effect in conversion paths – many campaigns may not directly convert, but are a part of the conversion path.

All these methods, however, are retrospective in nature. Predicting carryover effects is notoriously hard, and I’m not sure it would even be possible with such accuracy that it should be pursued.

Conclusion

In conclusion, I’d advise against being too hasty in drawing conclusion about campaign performance. This way you avoid the problem of premature judgment. The problem of shielding inferior campaigns can be tackled by using other proxy metrics of performance, such as the bounce rate. This would effectively tell you whether a campaign has even a theoretical chance of providing positive carryover effects. Indeed, regarding the prediction problem, proving the association between high bounce rate and low carryover effects would enforce this “rule of thumb” even further.

Dr. Joni Salminen holds a PhD in marketing from the Turku School of Economics. His research interests relate to startups, platforms, and digital marketing.

Contact email: [email protected]

Chasing the “true” CPA in digital marketing (for Pro’s only!)

This is a follow-up post on my earlier post about “fake” conversions — the post is in Finnish but, briefly, it’s about the problem of irreversibility of conversions in the ad platforms’ reporting. In reality, some conversions are cancelled (e.g., product returns), but the current platforms don’t track that.

So, my point was to include a ‘churn coefficient’ which would correct for the CPA calculation. In other words, it adjusts the CPA reported by the ad platform (e.g., AdWords) in regards to churn from “conversion” to conversion (as per the previous explanation).

The churn coefficient can be calculated like this:

1/(1-churn),

in which churn is the churn from the reported conversion to the lasting, real conversion.

However, I got to think about this and concluded this — since we consider the churn taking place due to real world circumstances as a lift to the reported CPA, we should also consider the mitigating factor of customer-to-customer references (i.e., word-of-mouth).

Consider it like this – on average, converted customers recommend your company to their friends, out of which some convert. that effect would not be correctly attributed to the referring customers under normal circumstances, but by attributing it uniformly to the average CPAs we can at least consider it in aggregate.

So, hence the ‘wom coefficient’:

1-(Cn / Cm), in which

Cn: conversions from new customers non-affiliated with any marketing channel
Cm: conversions from all marketing channels

The idea is that the new visitors who convert can be attributed to wom while conversions from marketing channels create the base of customers who are producing the recommendations. Both pieces of information can be retrieved in GA (for Cn, use an advanced segment).

So, the more accurate formula for “true” CPA calculation would be:

1-(Cn / Cm) * 1/(1-churn) * CPA

In reality, you could of course track at least a part of the recommendations through referral codes (cf. Dropbox). In this case you could have a more accurate wom coefficient.

Limitations:

Consider that in period t, not all Cn are created by Cm. Hence, it would be more realistic to assume a delay, e.g. compare to period t-1 (reference effect does not show instantly).

The formula does not consider cases where the referred customers come through existing marketing channels (this effect could be eased by not including branded search campaigns in Cm which is a good idea anyway if you want to find out the true performance of the channel in new customer acquisition).

Finally, not all customers from non-marketing channels may not originate from wom (especially if the company is using a lot of non-traceable offline marketing). Thus, the wom efficient could have a parameter that would consider this effect.

Dr. Joni Salminen holds a PhD in marketing from the Turku School of Economics. His research interests relate to startups, platforms, and digital marketing.

Contact email: [email protected]

Online ad platforms’ leeching logic

I and Mr. Pitkänen had a discussion about unfair advantage in business – e.g., a gift card company’s business model relying on people not redeeming gift cards, investment banker’s relying on monopoly to take 7% of each new IPO, doctor’s controlling how many new doctor’s are educated, taxi driver’s keeping the supply low through licenses, governments inventing new taxes…

It seems, everywhere you look you’ll find examples of someone messing with the so-called “free market”.

So, what’s the unfair advantage of online ad platforms? It’s something I call ‘leeching logic’. It’s about miscrediting conversions – channel x receives credit for a conversion while channel y has been the primary driver to it.

Let me give you two examples.

EXAMPLE 1:

You advertise in the radio for brand X. A person likes the ad and searches your brand in google. He clicks your search ad and buys.

Who gets credited for the sale?

radio ad – 0 conversions
google – 1 conversion

The conclusion: Google is leeching. In this way, all offline branding essentially creates a lift for search-engine advertising which is located at a later stage of the purchase funnel, often closing the conversion.

EXAMPLE 2:

You search for product Y in Google. You see a cool search ad by company A and click it. You also like the product. However, you need time to think and don’t buy it yet. Like half the planet, you go to Facebook later during that day. There, you’re shown a remarketing ad from company A but don’t really notice it, let alone click it. After thinking about the product for a week, you return to company A‘s website and make the purchase.

Who gets credited for the sale?

Google – 1 conversion (30-day click tracking)
Facebook – 1 conversion (28-days view tracking)

In reality, Facebook just rides on the fact someone visited a website and in between making the purchase also visited Facebook, while they learned about the product somewhere else. They didn’t click the retargeting ad or necessarily even cognitively processed it, yet the platform reports a conversion because of that ad.

For a long time, Facebook had trouble in finding its leeching logic, but now it finally has discovered it. And now, like for other businesses that have a leeching logic, the future looks bright. (Good time to invest, if the stock’s P/E wasn’t somewhere at 95.)

So, how should marketers deal with the leeches to get a more truthful picture of our actions? Here are a few ideas:

  •  exclude brand terms in search when evaluating overall channel performance
  • narrow down lookback window for views in Facebook — can’t remove it, though (because of leeching logic)
  • use attribution modeling (not possible for online-offline but works for digital cross-channel comparisons)
  • dedupe conversions between channels (essentially, the only way to do this is by attribution modeling in 3rd party analytics software, such as GA — platforms’ own reporting doesn’t address this issue)