Here are some introductory words. All advertisements on the Net can be divided (very crudely, so experts are kindly requested to find no fault) into three big groups: media, contextual and product advertising.
As for media ads, it is needless to discuss it in the strict payback perspective, as its strategy is different from that of contextual/product advertising, namely: to put the brand into prospective customers" heads as much as possible (in media advertising, it is impressions (shows) that are bought; visits to the advertiser’s site are a good bonus, no more), so we’ll concentrate our readers" attention on the real thing: the payback of contextual and product advertising.
Why contextual? As the search engine analyzes your query, it compiles a selection of both information drawn from Internet directly — structured in the search results in the descending order of relevance (+ your resource’s utility to the audience) and paid advertisements placed by millions of companies in the search engine’s advertising system (Yandex.Direct, Google AdWords).
And from that giant array of ads, Yandex (we’ll focus on it for simplicity, as Google AdWords" contextual machinery is generally the same) will select exactly those that seem of greatest interest to you (with a click most probable) proceeding from how you worded your search query.
Thus, after typing, "shirt ironing classes in Moscow", you are almost sure to see ads of irons and laundries. Advertisement shows are linked to the user’s region (geo-targeted), so if you live e.g. on the Franz Josef Archipelago, you’d better indicate a more inhabited area in Yandex settings to stage this experiment. And for a few days to come, the irons and laundries will keep haunting you on all the websites containing the Yandex.Direct/Google AdWords ad block.
It’s clear as noonday that contextual advertising is many orders more efficient than ordinary one. This is like shooting at a crow either from a telescopic-sighted rifle – or from a naval cannon and with your eyes closed; the combinations of round cost and target engagement probability are quite similar (it is also instructive to compare the tons of indignation that Net users pour on those obtrusive banners / pop-up boxes, while contextual ads are taken quite neutrally). The whole secret is how to make an ad contextual, and in the Russian market this secret, for now, is only known to Yandex and Google. On which they earn their well-deserved billions.
Let us examine this situation from an advertiser company’s perspective. An advertising campaign in Yandex.Direct is paid on a per click basis. To put it simply, the click price multiplied by the number of visits will form the final budget.
Importantly, the click price is a floating one. Advertisers compete with one another in sort of an auction that is constantly held by Yandex, where bids are click price and advertisement efficiency (CTR, Click-Through Rate). Where two advertisements match a search query, the one whose author worded it better (for a higher CTR) and is willing to pay more for a potential click will be shown first. The resultant of these factors will determine the ad’s place in the search results. Clearly, and quite in line with the ideal competitive market model (a limited advertising space on the screen vs. a virtually unlimited number of potential occupants), the more popular is a query, the more the advertiser is expected to pay. An ad for a query like "buy iPhone" will only be shown if you agree to pay several tens of dollars (just for a user’s visit to your page), while for an advertisement like "building nuclear plants in the Pskov Region" to be shown, a one-cent click will be quite enough.
Once again: we’ve been discussing the click price only, just a visit to your site, long before an order is actually made. A user may go from the Yandex page to your Internet shop’s page and buy nothing, but Yandex will write off your money anyway. The ratio between the number of purchases made to the total number of visits to the site is termed the advertising campaign’s conversion. We should note that the nuts and bolts of contextual advertising were described in large strokes. It includes not just many, but thousands of details, tricks and strategy options. So let us leave any further enlargement on this topic to professionals.
Yandex has a special tool, Yandex.Metrics, to measure conversion. It shows an advertiser fourteen megatons of imposing figures and handsome plots describing his campaign — in addition to conversion, these include the number of unique users, visitors, user geography and time breakup, etc.
Let’s concentrate on the gadget industry, well known to everybody, and take any popular Smartphone X.
To simply say that the X smartphone market is a highly competitive one is to say nothing. A click on an advertisement shown after any search query like "buy X smartphone in Moscow" costs tens of dollars, as we said above (for clarity, we’ll take it to be 50 as we continue).
Suppose you agree to pay 50 dollars for each click, your advertisement’s wording is not bad, and it is shown on good places. The next question is conversion. Its values for advertisements in this sector of trade vary between 0% (utter failure) to 5% (fantastic). If your advertising campaign results in more than 4%, this is a Success. So, if we continue with our Smartphone X example, it is already cool if only every twenty-fifth visitor to your site from Yandex makes a purchase.
Assuming a Success, we ultimately get the following:
The amount of money we must spend to make one sale, $ 50, divided by 4%, produces US$ 1250.This is debit. On the credit side is our margin from the sale of one Xsmartphone: we assume its price to be some RUR 20,000 and a 5% margin, so onthe credit side we get 400 — Roubles.
Converted into evergreen squids, our skim from that assumed transaction is minus 1235 U.S. Dollars. Immediately and reasonably, the question of "why all the headache" arises.
Our example with popular smartphones and an exorbitant click price is rather an extreme one. Still, in the case of less popular gadgets the above calculation will also produce a grim result in the TOTAL line. We shall not err much if we say that this pattern applies to whole set of popular goods selling via Internet in market quantities. Now let us record where we stand in this context — and look at product advertising.
For simplicity again, we’ll look at Yandex.Market, a leading goods offer placement floor. The buyer browses a well-structured catalogue of goods from different shops that are offered in the market. To help the visitor find the required product, the site includes filters, "assistants", interactive "wizards" and collation tools. After deciding what he needs, the visitor sees a list of different stores" offers of the item. These differ in price, delivery terms and timing, and the shops" ratings. The shops" offers retrieved are sorted in the order of a "relevance" — actually the product of the "click price" that the advertiser is willing to pay multiplied by the offer’s CTR. And here, like in contextual advertising, he goes to your site (is it well designed, by the way?) — still far from buying anything. And again conversion varies prosaically between 0 and 5%.
And now imagine that your assortment includes over twenty thousand items and all the stuff is uploaded
to Yandex.Market. And each item finds its buyers, and for each visit to your site the advertiser is charged the click price that he offered. His money may fly away in no time.
And the calculated payback of product advertising costs will be generally the same as the above figure for contextual advertising.
So every alley is a dead end. "And what’s the trick of it?", an inquisitive reader will ask. "Why does the market of both contextual and product advertising grow in volume with every passing year, despite this situation (that is, more and more companies place more and more ads at growing cost) and the computers of the banks keeping the Yandex and Google accounts will soon hang up due to overflow? Are corporate advertisers complete idiots? Or "millions of flies can’t be wrong", as enemies of the country "rising from its knees" say?"
Dear reader, we proclaim that the idiots are NOT complete ones. There is good reason behind their intentions to keep posting contextual/product ads and to step up the respective budgets.
This includes (but is not limited to) the following considerations:
As a result, more and more money is spent on contextual and product advertising exactly because of a persistent perception that it works. Just a perception. Nobody can calculate which ads work and how (except happy client of dia$par, but we’ll dwell upon this below): it takes (costly) experience, exceptional art-like advertising skills and a lot of time to get some idea, so we have too much risk and too much money at stake. But we can’t avoid taking the risk, for more active competitors will eat away our market share.
So the old maxim that half of a budget is wasted, but God knows which one, applies to Internet advertising as well. And the megatons of figures generated by the use of primitive tools like Yandex.Metrics are of little practical use — and simply dangerous if you are not skeptical about them. Back in the 1980s Mr. Trout said that "It used to be that a CEO would wonder which half of his advertising budget was being wasted. Now he is beginning to wonder if it’s the whole budget."
A constant reader of our website can already feel that now dia$par will go on stage and have all others for breakfast.
Yes, it will. meta-system stores all the information about your advertising campaigns: goods/goods categories, source of traffic (where the ads are actually posted), geography, and cost (can be indicated/calculated dynamically).
When somebody calls at your online store, the latter will leave an ID key (a cookie file) on the visitor’s PC, linked to information on this advertising campaign that is stored in dia$par. All his/her further actions (registration and making orders on the site, purchases at stores, etc.) are recorded in dia$par, attached to his ID key.
This is just a general description. The actual mechanism is far more complicated, but the details are of no interest to a layperson. Just in case, we’ll note once again that the dia$par mechanism described above works with any type of Internet-based advertising, and not necessarily with contextual or product ads; dia$par get curious results on SNS and even media commercials.
As a result, for each advertising campaign we know the clients that it has attracted. It is not important when the client registered or bought something. He may come in six months, and the system will log everything. We can look and see what exactly that individual campaign’s clients bought immediately (as it was still running) or within three months or three years; we know all the history. We can calculate the return from our expenses on each advertising campaign (or a group of these, selected at our discretion) for each goods item (group of items), region, time period, shipment time, and sales of related goods / services — as you wish.
How much is earned or lost and where — your eyes are open, and your telescopic sight is attached. Prepare a large basket for the crows you’ll bag.
A few figures from the Ulmart example. We compare the following periods: the first month after system launch (before the launch, no written records had existed due to the above causes), actually plus or minus the general market figures, and the results obtained by optimizing the advertising campaign a year later.
ROI: 70% and 400%;
ROMI: −25% and 300%;
CAC (customer acquisition cost) decreased by 20%;
The CAC payback period shortened from eight to three months.
A perfect insight into the efficiency of their online advertising investment enabled Ulmart to consciously increase their budget more than tenfold over two years and to become a major Russian Internet advertiser.
Perhaps the most efficient one. Incidentally, the Ulmart CEO calls this system U-Metrics.
Still, the above notwithstanding, perhaps things are different in your case and you’ve got no questions about the efficiency of your advertising costs? Then you needn’t stir.