Towards a Bill of Rights for Online Advertisers
Benjamin Edelman - September 21, 2009

I offer five rights to protect advertisers from increasingly powerful ad networks -- avoiding fraudulent charges for services not rendered, guaranteeing data portability so advertisers get the best possible value, and assuring price transparency so advertisers know what they're buying. I explain the need for these rights by presenting specific practices causing particular concern.


Related Projects

How PPC Conversion Inflation Increases Advertisers' Costs

Delaying Payment to Deter Online Advertising Fraud

Resources for Affiliate Merchants

Other Research by Ben Edelman

Online advertising presents remarkable efficiencies—better targeting, improved measurement, and greater return on investment.  By dramatically lowering distribution costs, online ads cut fat from placements that had assumed significant waste.  Pairing precise measurement with increasingly specific targeting, online ads can show the right message to the right user, with commensurate increases in effectiveness.  Furthermore, online formats invite interactivity—not just text, pictures, and videos, but evaluation, discussion, and customization.  The opportunity is staggering.

Yet there are challenges, particularly when networks of intermediaries place ads through convoluted relationships, and all the more so when advertising powerhouses dictate unsavory terms.  The result is a troubling mess of ads gone wrong—advertisers charged in ways they didn’t fairly agree to, and on terms they didn’t meaningfully accept.  These problems threaten to destabilize online advertising—wasting advertisers’ budgets, slowing transition to online formats, and reducing payments to online publishers.

But online advertising doesn’t have to be a wild west.  In the sections that follow, I propose five specific rights advertisers should demand as they buy online placements:

1. An advertiser’s right to know where its ads are shown.  It is nonsense to pay for ad space without knowing where an ad will appear; sites vary too much in user quality and context.  Even for “blind buys,” advertisers need enough information to determine whether a given site qualifies to show an ad.  Anything less undermines accountability—inviting fraudulent sites that devour advertisers’ budgets.  And with all manner of fraud—from spyware pop-ups to invisible banners to adult sites slipping into networks that claim to be brand-friendly—advertisers need to be wary.

2. An advertiser’s right to meaningful, itemized billing.  Clear records protect advertisers from accounting games.  Otherwise, ad networks can claim “We already credited you for those clicks,” knowing that advertisers cannot prove otherwise.  But some ad networks provide invoices that are opaque at best.

3. An advertiser’s right to use its data as it sees fit.  Campaign configuration details (such as keywords and targeting) are an advertiser’s own creation, to be retrieved whenever and however the advertiser chooses.  Same for records of campaign performance.  Yet some ad networks impede data portability in an attempt to increase their share of advertisers’ spending.  Such restrictions can lock advertisers into needlessly expensive ad platforms—sharply increasing advertising costs.

4. An advertiser’s right to enjoy the fruits of its advertising campaigns.  When a user clicks an ad, the advertiser pays fair value to reach that user.  But in a world of behavioral targeting, a network can later resell that same user to the advertiser’s direct competitor.  Click one airline’s ad, and a network may conclude you’re in the market for travel—then show ads for other carriers.  That’s a poor value for the advertiser whose spending sparked the targeting.

5. An advertiser’s right to resolve disputes fairly and transparently.  Ad networks generally write the contracts that govern their dealings with advertisers. Networks often use this drafting power to tilt contracts in their favor—disclaiming promises that ads will appear anywhere in particular, and denying responsibility for fraud, even when they know about it and fail to take action.  At face value, these contracts purport to grant networks effective immunity from advertisers’ complaints.  But advertisers don’t accept such one-sided provisions in other procurement contexts, and they need not be so lenient in online ad-buying.

In this time of plummeting ad prices, networks are increasingly anxious to attract advertisers.  With increased negotiating power, advertisers can and should demand more.

Let me add at the outset a note on my focus: In many of my examples, I discuss and critique the practices and policies of Google.  I have great respect for Google’s accomplishments; indeed, I use Google services daily.  But in some areas, I am convinced that Google serves its interests at advertisers’ expense, as I explain in detail.  In other areas, Google follows industry norms, but those norms deny advertisers rights they should insist on.  Throughout, Google’s dominant position means that advertisers must do business with Google even if they dislike Google’s policies.  So Google’s practices and Google’s intrusions on advertisers’ rights deserve special scrutiny.

On the subject of Google, one additional disclosure: I serve as co-counsel in trademark holder litigation challenging Google’s use of typosquatting domains to show ads.  See Vulcan Golf, LLC v. Google, Inc., No. 07CV3371 (N.D. Ill.).  But that matter is limited to typosquatting—entirely unrelated to the practices discussed here.

1: An advertiser’s right to know where its ads are shown

Online advertising is a quid-pro-quo: An advertiser tenders payment and receives ad placement in exchange.  But what ad placement does an advertiser receive?  Which users, on which web sites?  This is the essence of the value provided by the site or ad network receiving payment.  Advertisers deserve to know what they’re getting.

Confirming where ads appear is particularly important because fraudulent placements risk draining advertisers’ budgets with no genuine marketing benefit.  Examining pay-per-click ads, I have found Google ads placed through deceptive toolbars that create search ad traffic when users attempt direct navigation; I have uncovered Google ads placed through spyware/adware, again targeting merchants’ own organic traffic; and I have shown Google ads placed through typosquatting sites, including misspellings of advertisers’ own names.  As to banner ads, I have documented impressions faked through nested invisible windows, and I have tracked spyware that loads banners invisibly and even fakes clicks.  These examples are complicated, and interested readers may wish to examine details in the referenced articles.  But the bottom line is simple: No reasonable advertiser would want to pay for any of this traffic.  That said, it can be hard for advertisers to uncover such problems for lack of insight into where, exactly, their ads are being shown. 

Current industry norms largely deny advertisers information about where their ads are shown.  For example, Google’s AdWords Terms and Conditions provide that “Customer understands and agrees that ads may be placed on any other content or property provided by a third party (‘Partner’) upon which Google places ads (‘Partner Property’).”  Try reading that aloud: It’s tautological and circular, purporting to give Google unfettered discretion to place ads anywhere it chooses. 

The problem reaches beyond Google.  Numerous display ad networks provide so-called “blind” ad-buying, where an advertiser is never told where its ads are shown.  Networks believe this approach protects their publisher lists—preventing a competitor from poaching publishers.  As a threshold matter, I doubt withholding a site list actually provides an ad network with much defense against poaching; a determined ad network with a custom crawler can still find sites showing competitors’ ads.  Moreover, ad networks offer important benefits beyond assembling publishers; ad networks provide dramatic convenience, centralized tracking, consolidated payments, standardized contracting, and other administrative functions.  In any event, ad networks could protect their partner list through contractual restrictions preventing merchants from bypassing a network for any relationship initially brokered by the network.  (For example, Commission Junction’s Advertiser Service Agreement includes exactly this restriction.) 

Ad networks may counter that advertisers care only about results.  I’m not so sure.  Plenty of brand-conscious advertisers impose important conditions on where their ads are shown—disallowing placements on sites with adult themes, refusing spyware/adware popups, and rejecting ads on deceptive or fraudulent sites.  Moreover, focusing only on “results” ignores the problem that measured results can be the subject of fraud, as in many of the examples linked above.  Financial auditors well know that robust verification requires verifying data from multiple independent sources—which means comparing ad networks’ performance data with data gleaned from other sources, such as direct observation of ad placements.  But if an advertiser does not even know where its ads appear, such confirmation is difficult or impossible.

2: An advertiser’s right to meaningful, itemized billing

In most states, consumer protection laws require not just itemized receipts, but also customer-facing displays that immediately report the price of each item a customer purchased.  (That “extra” screen on each cash register exists not because merchants choose to install it, but because it is required by statute.)  With detailed purchase information, consumers are better positioned to confirm what they were charged and why—helping consumers identify and resolve any pricing errors.

Online advertising lacks analogous laws or even norms.  Advertisers often receive a perfunctory bill with minimal detail as to what ad services were purportedly delivered.  An invoice may count impressions or clicks, but there is usually no indication of who clicked, when, or, crucially, where. 

The combination of unpredictable ad placements and opaque billing yields an extra reduction in accountability.  Consider an advertiser that found some unsatisfactory placements of its ads—perhaps problems like those described in the prior section.  If the advertiser complains to the responsible network, the network will typically offer a credit in an amount purportedly reflecting the size of the problem.  But the advertiser has little practical ability to determine whether the amount is appropriate.  Should a given problem yield a $90 credit, $900, or $9,000?  Without itemized billing, an advertiser can only assess a problem’s scope by asking an ad network for assistance.  But responsible networks have clear incentives to underreport improper charges. 

I envision a system that gives advertisers more information to improve accountability.  For an ad network combining multiple traffic sources, a meaningful invoice would itemize each traffic source and separate charges accordingly.  Better yet, with each click a network passed to an advertiser, the network could identify what site hosted the ad—helping the advertiser identify any traffic sources where traffic is unproductive, suspicious, or worse.  Passing this information could be as simple as a tag added to a URL as traffic reaches an advertiser’s server—not just “?from=adnetwork” but “?from=adnetwork&source=sitename”.  If an ad network is not prepared to disclose its partner sites, it could run a one-way transformation of site names (e.g. a hash), letting advertisers confirm traffic quantities from any individual site by matching transformed values, even without knowing which site is which.

3: An advertiser’s right to use its data as it sees fit

An advertiser should own its own data and should be able to use that data as it sees fit.  This proposed “right” sounds so fundamental that it might seem to be beyond dispute.  Yet it is under attack by at least one ad network that limits how advertisers can store and copy their own information.

Is Google more expensive?

Ask a typical online advertiser about Google’s pricing, and they’ll say Google is quite expensive.  More expensive than competitors?  Even after controlling for differing conversion rates?  To see for myself, I obtained data on PPC spending by clients of a top agency.  The agency had evaluated pay-per-click ad purchases from Google, Yahoo, and Microsoft (among other providers), and had optimized campaigns to make the best possible use of its clients’ budgets.  For each advertiser, I compared spending at Google with spending at Yahoo and Microsoft, controlling for conversion rates.  Here’s what I found:

- 49% of sampled advertisers found they could buy traffic from Yahoo at average prices (in an effective CPA sense) 30% less than the prices those advertisers faced for the same campaigns at Google.  Of course Google offered more volume than Yahoo; on average, these advertisers bought only 39% as many clicks from Yahoo as they did from Google.  But the Yahoo traffic was dramatically cheaper.

- 72% of sampled advertisers bought traffic from Microsoft at an average price (again, effective CPA) 27% less than the same campaigns at Google.  From Microsoft, these advertisers bought just 13% as many clicks as they bought from Google, but Microsoft’s prices were lower.

- For 40% of sampled advertisers, Yahoo and Microsoft jointly provided at least 10% as much traffic as Google, at prices at least 10% lower than those available at Google.  For 29% of advertisers, Yahoo and Microsoft offered at least 20% as much traffic as Google at prices at least 20% less than Google’s prices.

Of course results vary from advertiser to advertiser.  But the basic result is clear: An advertiser buying placements only from Google is foregoing alternative channels that could offer some traffic at substantially lower prices.

Source: author’s analysis based on price and quantity data from a top ad agency optimizing spending on behalf of its advertiser clients

Consider an advertiser that has built a large PPC campaign in Google AdWords—thousands of ads targeting thousands of keywords.  The advertiser might reasonably want to begin to use other ad platforms too—perhaps Yahoo or Microsoft AdCenter.  The easiest way to use additional platforms?  Export a campaign from Google, then import it as desired.  Using Google’s AdWords API, this export-import could be a simple XML transformation, readily implemented with a small client-side app or even a direct server-to-server transfer.

Unfortunately for advertisers, Google’s AdWords API Terms & Conditions stand in the way.  This contract requires that all AdWords data appear on a different tab or screen from all data for competing ad platforms (III.2.c.i), then specifically prohibits AdWords API clients from “cop[ying] data from one tab or screen into another tab or screen” (III.2.c.ii).  Through these restrictions, Google’s API contract exactly prohibits using the AdWords API to copy data from one ad platform to another.  Nor can tool providers build software to help advertisers make such copies.  Indeed, when tool providers have built such systems, Google has insisted that their systems be disabled, and Google has invoked its right to audit and inspect their offerings.

Google’s AdWords API contract also dictates other decisions that should be advertisers’ sole purview.  For example, the AdWords API T&C’s insist that AdWords data “must be stored separate” from data for other ad platforms (III.2.c.iii).  Indeed, Google requires tool providers to build separate database instances for Google versus competitors—increasing cost and complexity, and preventing straightforward database queries that make comparisons across ad platforms.

These intrusions interfere with competition in online advertising markets.  It seems at least 60% of PPC advertisers currently use Google alone.  (Source: author’s analysis based on 2008-2009 ad data collected by ComScore)  Comparing the cost of ads at Google versus competitors (after normalizing all costs to an effective-cost-per-acquisition baseline), Google appears to be considerably more expensive than Yahoo and Microsoft for most advertisers.  (See sidebar for details.)  So advertisers could reduce their advertising costs by buying a portion of their traffic from competing providers.  More generally, robust competition among ad platforms would serve advertisers’ interests in pricing, quality, and service.  But through these API restrictions, Google exactly stands in the way of this sort of competition. 

It is particularly striking to see Google limit advertisers' use of their own data in light of Google's stated policies on data portability. In a November 2006 interview, Google CEO Eric Schmidt specifically promised that "We [Google] would never trap user data." More recently, Google’s proclaims the “users own the data they store in any of Google’s products” and that “as little as possible” should stand in users’ way in transferring data.  I applaud these promises.  But Google self-servingly offers data portability principles only in markets where Google has low market share (such as photo sharing and office applications), but not in the search advertising sector where Google enjoys overwhelming market dominance.

An ad platform has no proper role in dictating how an advertiser copies or stores its own data.  Nor is there any proper reason for an ad platform to prohibit advertisers from copying data to competing platforms or from trying competitors’ services.  No other ad network has had the gall to attempt such interference, and advertisers should reject Google’s tactics as an improper abuse of market power. 

4: An advertiser’s right to enjoy the fruits of its advertising campaigns

Incidental to an advertiser’s online ad purchases, various data is generated about users and their interests.  “This user just clicked an ad that promoted automobiles,” an ad network might notice.  “Maybe we should show the user ads from other car companies too.”  From the ad network’s perspective, that’s found money, but it risks tainting the value the ad network delivers to the advertiser who bought the first click in the sequence.  The harm is particularly acute when the first advertiser spent big money—perhaps buying thousands of CPM impressions—to find the rare user interested in an obscure subject.

I propose a simple principle: When an advertiser buys online placements, the data generated from those placements should benefit the advertiser, not the network or its other advertisers. 

Of course advertisers may waive this right if they so choose.  For the right discount, advertiser A might be content to let competitors reach the same customer that A already paid to identify.  But the default should be that A, and A alone, enjoys the fruits of its spending.  Anything else will diminish A’s incentive to invest in campaigns to find the customer in the first place.

5: An advertiser’s right to resolve disputes fairly and transparently

If something goes wrong in an online ad placement, current dispute resolution is remarkably opaque.  Typically, an advertiser reports a concern—then receives an answer that is, for many ad networks, effectively non-negotiable.  Without itemized billing data, the advertiser often cannot easily determine the full scope or seriousness of the problem.  And it’s particularly hard to complain since networks claim they can put ads anywhere they choose.  (Recall Google’s circular “ads may be placed on any other content or property … upon which Google places ads.”)

Even when an advertiser uncovers a breach of an ad network’s obligations, it can be hard to respond effectively.  For one, often the individual amounts are too small—hundreds of individual overcharges that are small individually, yet large in the aggregate.  (I often call this “death by a thousand papercuts.”)  Identifying, documenting, and pursuing these many infractions would be costly and time-consuming, yet ignoring the problem is equally unpalatable.

Ad networks’ terms can prove a further barrier to resolution.  For example, in 2007 class action litigation, merchants claimed that ValueClick’s Commission Junction affiliate network allowed spyware and adware to create advertising charges that should not have been paid.  In defense, ValueClick invoked its “Disclaimer of Warranties” to claim it could not be liable for any such overcharges.  Google’s AdWords Terms and Conditions take the same approach: Google “disclaims all warranties” “to the fullest extent permitted by law” “regarding positioning, levels, quality, or timing of … clicks, conversions or other results for any ads.”  I doubt that this disclaimer is enforceable according to its terms; taken literally, it would relieve Google of liability no matter how Google overcharged its advertising customers.  But the mere presence of the text serves to discourage complaints—a real impediment to advertisers who want the benefits Google’s marketing site proudly promises.  These disclosures are particularly troubling because Google’s marketing statements are exactly opposite—for example, promising placement on “high-quality” web sites when the fine print denies any warranties of quality.

For advertisers in some countries, procedural hurdles further impede dispute resolution.  While visiting Australia this summer, I happened to read the Google AdWords Australia Terms and Conditions.  Australian advertisers must sue Google in Santa Clara County, California—even though Google operates a 300-employee office in Sydney complete with an engineering team.  Furthermore, to give official legal notice to Google, Australian advertisers must write to Google in, of all places, Ireland—almost exactly halfway around the world from Australia.  And by “write,” I mean exactly that: The contract demands that advertisers send their complaints by “first class mail or air mail or overnight courier,” requiring a special trip to the post office, extra costs, and delay.  Yet one sentence later, Google’s contract provides that Google may reply to an advertiser’s message simply by simply by sending an email.  Far from establishing an efficient way for the parties to communicate, these provisions intentionally discourage advertisers from raising complaints.

Google’s market power makes its terms especially onerous.  If an advertiser doesn’t like disclaimers in other providers’ fine print, the advertiser can take its business elsewhere.  Better yet, an advertiser can threaten to take its business elsewhere, and obtain a revision to the challenged provision.  But with Google’s market dominance, advertisers have no meaningful alternatives for high-volume sponsored search.  As a result, whatever terms Google insists on, advertisers must accept.  Even where Google’s disclaimers track policies at other networks, Google’s market power gives its provisions substantial extra force.

Structural impediments to advertisers’ rights

Several structural factors further weaken advertisers’ position.  For one, advertisers are typically generalists negotiating with specialist ad networks: Any given ad campaign is just part of an advertiser’s strategy, whereas the ad network sells the same service over and over.  As a result, the ad network comes from a position of strength—expertise on its products, its deficiencies, and its strategy.  When attorneys get involved, ad networks’ specialization again proves useful.  An advertiser’s attorneys must manage all manner of relationships—buying ad space as well as handling legal matters in scores of other areas.  In contrast, an ad network’s counsel are specialized in their specific service, and they can justify investment in what is, for them, a key area. 

Ad networks also benefit from writing the contracts that govern their ad sales.  Some individual sites sell ads using IAB standard Terms & Conditions, but most ad networks insist that advertisers accept terms the networks write internally.  In their dealings with small to mid-sized ad networks, big advertisers find that they need not accept networks’ standard terms; networks want advertisers’ business and are prepared to negotiate to get it.  But when it comes time to negotiate with the largest networks, even big advertisers are usually stymied.  When my contacts have sought custom terms from Google, they’ve been turned away; Google usually says its terms are take-it-or-leave-it, and Google’s market power lets it force advertisers to follow its requirements.

Examining intrusions on advertisers’ rights, many of the starkest problems come from dealings with Google: Google does not tell advertisers where their ads will be shown, omits itemized billing, limits how advertisers can use and transfer their own data, and insists on convoluted dispute resolution.  Why such one-sided terms from Google?  Google’s large market share—reportedly 72% in the US, and as high as 93% to 95% in countries like Spain, Germany, and the Netherlands—immunizes Google from the competitive pressure that other ad networks face.  Ordinary ad networks know they would lose customers to competitors if they imposed unsavory terms, but Google knows that companies wanting search advertising have little alternative.  If competing search ad platforms gain traction, Google may be forced to scale back its harshest practices.  But in the interim, I believe more should be done.  Particularly when Google’s tactics exactly and intentionally impede competition, as in Google’s prohibition on advertisers using the AdWords API to try competing ad platforms, regulators and industry associations should disallow Google’s restrictions as an improper impediment to competition.  And when Google unilaterally imposes contract terms that effectively immunize it from liability, regulators should question those contracts with equal skepticism.

What role for advertisers in these disputes?  For one, advertisers can take steps to protect themselves.  Get technical teams up to speed on types of fraud, and ask them to find ways to search logs and headers for signs of malfeasance.  Get lawyers focused on this too, and ask them to revise contracts to protect advertisers’ interests.  Seek guarantees from ad networks and agencies, and hold them accountable when things go wrong.  Complain to regulators when policies and contracts are restrictive, anticompetitive, or just unfair.  There’s ample room for improvement, and advertisers need not blindly accept the world as it stands.

Consumers enjoy a century of hard-fought consumer protection statutes—requiring pricing transparency, rejecting hidden contract terms, guaranteeing certain key warranties.  Experience shows these protections to be crucial in assuring fair outcomes in the myriad transactions consumers face every day.  With ever-increasing market concentration among ad providers, the online economy may benefit from analogous protections.  But advertisers won’t get what they don’t demand.  I offer these proposed rights in that spirit, and I look forward to advertisers’ efforts to achieve these protections.


Posted: September 21, 2009. Sign up for notification of major updates and related work.