Experience in building a b2b product: 3 continents in 6 years and half a mile of stuffed cones

Whatever the idea and implementation, but in B2B without good sales people it’s hard to fly. We are proof of that - by 2007, we already had experience in a start-up company and not a single dozen projects, but none of them turned into a full-fledged product.

And all because in previous years, we were wobbling from logistics to biology, from insurance to recruiting, from medicine to e-commerce. And although all the projects were created on a common engine (multi-agent systems + knowledge management + data / text mining + optimization engine), each remained in a single instance, solving the problems of a specific customer. And all sales were carried out on a residual basis, if somewhere it was possible to catch the tongues with the right people.

Therefore, when we finally managed to hire a sales person who was ready to develop knowledge management, we had a list of 20 ideas, each of which, with due diligence, was supposed to bring millions.

Some of those ideas still seem promising, the rest during this time were already invented and implemented by others and, as far as I know, they did not bring much riches to the creators. Therefore, what we have chosen as a result of the field of online advertising, to some extent, luck.

When we say that we are engaged in the optimization of Internet advertising, for some reason, many people associate with SEO. Slightly more advanced recall Google. AdSense and Yandex. Direct. Some people think that we create advertising ourselves, determining which video to shoot or which marketing slogan to deliver.

In fact, almost none of our developers can even see what the banners and videos of the advertised goods and services look like. We solve a complex optimization problem - based on the statistics of many advertising campaigns, adjust their progress so as to maximize certain performance indicators for each specific campaign and for the entire advertising network as a whole.
Our main focus is the optimization of premium display advertising inventory on large sites - those where hundreds and thousands of advertising campaigns are competing at the same time for impressions. In general terms, we intelligently automate the process that is extremely difficult to manage manually.

And since all of us in ordinary life do not like advertising for intrusiveness and irrelevance, we have chosen an extremely simple mission for ourselves - to make Internet advertising more useful, maximizing the likelihood that every particular person will be shown only what he personally interested, and only there and then, when he may be interested. This is a win for everyone - advertisers and users.


Now it is interesting to look at the very first technical project and layouts that we showed the earliest potential customers. Maybe 10% of what is there, eventually logged in. And not to say that the remaining 90% were completely irrelevant to the matter - these were real problems. Just not so painful that customers want to pay for their decision. But the Pareto principle once again showed its applicability - the remaining 10% was more than enough to build a business.

With the first client, we were very lucky - and at the same time pretty unlucky. I was lucky that they turned out to be experts in their field, who are perfectly able to optimize advertising manually - because as an advertising agency, in accordance with the specifics of the UK market, they lived solely on the earned margin between what they bought from the site owners and what they sold to advertisers. And we have learned a lot from them and spied.

And no luck, because at that moment, without having enough experience in this market, we began to develop the product, assuming that all players are as smart as these guys from MediaRun. Therefore, the first version of the interface is very similar to the dashboard of the aircraft.

We have created a very complex system of intellectual decision-making support, with a training mode, a huge number of options and configurations, opportunities for adjusting for themselves, and so on, so forth. And then, all subsequent years, were forced to simplify, hide and minimize everything. For, as it turned out, the dream of an ordinary client is one button “Do me well”. Well, plus, the reporting system for the authorities like "I have everything under control."

Unfortunately, we lost our very first client before we started working (another company, AdKnowledge, bought them, and their priorities changed). Therefore, in order not to lose pace, we decided to take root in the next willing. But access to the data was given to us after the decision was taken to implement and coordinate the deadlines, as a result of which it turned out that the new client runs on another advertising server (for definiteness, AdTech Helios), which we hadn’t seen before. And if the data integration and interface at the time of delivery somehow managed to finish, the whole system logic (recommendations for improving advertising campaigns and predicting the behavior of the advertising network) was in a state of deep rethinking and processing.

So, the process of delivery looked like this: during the month, every day, we told the client how much automatic optimization is cooler than what the experienced operator team could do manually. At the same time, they clarified exactly how they are optimizing now in this or that situation. And then at night with pens they did what, as they themselves convinced the client, it is impossible to do. And the next morning they gave out results for the set of recommendations for improving the network that was calculated at night by the program. Received a review where and what the program did not take into account, and promised to finalize. And they “finished” the next night. At the same time, they shared with programmers knowledge of how optimization should work.

But after a month of such activity, we not only had a system that could automatically do everything that had previously been considered by hands. As a bonus, we have become experts on the internal logic of the advertising server, understanding it no worse than its developers. What made it possible in the future to win many clients at the first meetings, due to the demonstration of the knowledge and techniques that were not known by the most experienced technical and operational specialists on the customer side.

We arrived to Channel 4, being already fully armed. And because they were on the ad server with which we initially integrated (OpenAdStream by 24/7 Media). Plus, unlike the first client, our system already knew how to work by itself.

Not to say that everything went smoothly. By the time we arrived, they were just deciding which ad server to go to, choosing between OpenAdStream and Google DART for Publishers.

Although optimization was critical for them, the situation was complicated by the fact that Google had an optimization solution - DART Adapt. Moreover, it was provided to the client free of charge and was part of the overall offer.

And we had to directly compete with Google. A series of test advertising campaigns was launched on one advertising server, their “twins” on another, and then after a while the results were compared.

We won. There are two reasons (well, except that our solution is really good and of high quality). First of all, a bunch of “person + computer” is usually always stronger than just a computer - and by that time we could give a head start to a computer, at least for a short distance. And secondly, we competed with Google seriously, but he didn’t hear about us then, so there wasn’t much effort.

And it launched a full implementation. As always, not without problems and jambs. In constant attempts to justify the benefits of the decision, and sometimes just find out how the client himself calculates the ROI of their own actions. But once…

Once their clients, an advertising agency, called their sales department. And they literally stated the following: “We don’t know what exactly you did there. But the past few months, the effectiveness of all our campaigns that run on your sites is 50-100% higher than their performance on all others. Therefore, do not want to - do not tell, but you are great, keep up the good work - we already have new contracts for you.
After such a sign a long-term contract with Channel 4 was a matter of technology.

At this moment, practically on takeoff, there was the first serious failure in development. The classic situation is dividing the skin of an unkilled bear. The salesman and CEO did not share the annual bonuses put for the first successes, quarreled, the salesman quit, and we lost the only person who could really sell, and not just get puffed up, swelling from his own greatness.

No, we still have sellers, but these were tears. From week to week they talked about progress, reporting the same Excel file that had not changed for months (they probably thought that we had a bad memory). We ourselves had to prepare all the marketing materials, including what advantages we need to tell customers, what to say about competitors, etc.

Moreover, seeing complete impotence on the sales side, we formed our own lists of potential customers - those who had sufficient volumes and used the advertising servers we needed. A separate question is that such information is not publicly available, and we had to write a web crawler, recognizing by tags, which client the ad server uses (at the time there were no solutions like ghostery yet). But calling from far-away Russia to large European corporations and explaining the advantages of the product in broken English, the existence of which they didn’t suspect, is not the fastest way to success.

In general, the rule for salespeople is extremely simple - he must be a beggar at the base salary. And if at the expense of the bonus system, he earns more than the general director, this is great, because in this case he will bring more money to the company. But if he is already comfortable with the basic salary ... Especially considering that in B2B, even with good sellers, it takes about 4-6 months before the first sale at a new place.

The situation was worsened by the general director himself, who, by analogy, received a very good salary and didn’t really want to bother. And on all the requirements of changing sellers, intensive sales, marketing research, etc. he just kindly patted us on the shoulder, saying that the little ones still didn’t grow, they don’t understand how a serious business is organized.

The favorite illustration of that time is a typical first show to a potential customer. The show is live (say, in London), and our technician is on the phone in Samara. And Gender very solidly and confidently tells the introductory part about the company, for three minutes. Then he gives the word to the learned monkey to a technical specialist, who already shows the system, telling about its advantages and, in fact, selling as much as possible. And absolutely any question addressed to gendera, having made a significant pause, will necessarily redirect back to us (simply because it is in such high matters, where any knowledge about the product and the subject area is completely superfluous).

The apotheosis (or, more precisely, the nadir) of our failures and throwings was the introduction of Sky. By that time, it was the largest and most complex client, with almost unrealistic demands and volumes several times higher than our maximum in productivity.

Therefore, we made a typical mistake of techies - solving problems that we like to solve, and which we are well-versed in. But completely without addressing and controlling those that eventually became the cause of failure. During the implementation period, we accelerated the system several times. We have achieved all the performance and efficiency indicators stipulated in the contract. And, nevertheless, the client safely ignored the signed contract, and the long-term agreement was not concluded.

But because it is necessary to pay more attention to client management - as it turned out at the very end, there were three competing groups in Sky itself, each of which promoted its decision (and with success - an additional budget, career growth, fame, money and other worldly pleasures ). And the competition was won not by bare figures of performance and efficiency.
And all these problems we had to either recognize initially and not get involved in this adventure, or even then fully untie the Gordian knot, participating in all political games within the client.

By the way, what is significant: recently, on their own initiative, they came to us with a proposal for cooperation. But out of more than 10 people who were part of those three warring factions, no one is already working there. The snake devoured itself.

At this point, investors finally got it - something wrong is happening.
Yes, investors are generally a separate conversation. Even in a personal conversation it is almost impossible to reach them. They operate exclusively with money and prospects. And until they themselves see the risks, who will listen to the insane bear from Russia?

When, finally, they ceased to blindly trust the words of the general director and inquired about the real numbers, the decision to find any worthy alternative to the existing CEO turned out to be obvious. And what's more, for the first time they seriously thought about the prospect of business as such.

Here we were very lucky. At this point, we have already successfully cooperated with Channel 4 for more than a year, and since everyone in this industry knows each other, rumors about our product slowly spread, at least in the UK, and the Guardian came to us, very interested in increasing its monetization. content through possible optimization.

In fact, it was a crucial moment. At the beginning of the path, the failure of even one major implementation can seriously shake the company (both in terms of money and reputation). And two failures in a row - and you can go home.

And we threw all our strengths - the best technical specialists drove in to implement, customization was completed faster than immediately, and the leading business analyst every day and a half months, sat with 15 operators, teaching our system, clarifying the requirements, answering every question. And literally every week, the client received a new release, including one or another of his "Wishlist."

But once again - we are very, very lucky with the client. The Guardians are absolutely fantastic guys - smart, demanding, open to everything new. After they were convinced that the system really gives a tangible increase, they became our hottest fans. Each next client we sent to the Guardian for a review of our work and they “sold” to us much more than all our own salespeople put together.

Started white stripe. Channel 4 and Guardian were two of the largest clients in the UK OpenAdStream ad server (24/7 Media). And we have demonstrated that we can consistently achieve good results.

It is necessary to say that the war of advertising servers is constantly going on - for large site owners usually conclude a contract for two or three years, and each time they closely monitor the market - whether a solution that is more profitable in terms of money or functionality has appeared. And without hesitation, sites migrate from one vendor to another almost every year. To a greater degree, this applies again to the Western market, since in Russia each major player has his own decision.

Therefore, the competition in the provided functionality is one of the most fierce in business - whether it is targeting by geography or demographic features, support for keywords, or, in our case, increasing the effectiveness of advertising campaigns.

At that stage, Google had a DART Adapt solution (of course, only suitable for their server), while OpenAdStream competing with it had its own optimization at a fairly basic level. What is generally typical and explainable? The required skill set for creating an ad server differs significantly from the experience needed for a high-quality optimization solution. In this case, the budget for the team is always limited.

And so 24/7 Media signed an exclusive partnership agreement with us; we became an official optimization solution on their server and all their clients, if they wanted optimization, could only work with us.

The Guardian was so pleased with the first results (and we really allowed him to increase the revenue from premium advertising by about a third) that he agreed to nominate us for one of the most prestigious awards in the industry.

And we won it, including defeating solutions from Google (DART) and from Microsoft (Atlas).

In many ways, of course, the victory was due to the fact that for the first time we were seriously lit up at such events and on our side was the effect of novelty. And the fact that together with us played the Guardian and 24/7 Media. But the victory, however, was unexpected and completely fair. And it happened very on time.

A partnership agreement with one of the five largest advertising servers in the world. Satisfied Guardian. Won the award. Investors are convinced of the great prospects of the project. And they were ready for serious investments - primarily in the US market, because there is at least 50% of the total advertising budget in the world.

But by this time we were already so full of mistakes and incompetence of some of our colleagues that we were ready to stop immediately if a number of organizational issues were not radically corrected and we didn’t have more opportunities to influence the development of the company.

Therefore, it is time to seriously talk and agree on what the company will be.
What can I say now, after a few years? If you enter into an agreement, unscrewing your partner (or simply if the partner perceives it in this way), then do not hope that the agreement will be respected by the time when the partner can unscrew your arms. And no contract will not save it. As a result, we managed to find a fragile balance that suits both sides, but could not return the time and effort spent on internal squabbles.

Thus, somehow we agreed, we found and hired a new CEO in New York on our own, completely changed the sales team and moved with optimism to a bright tomorrow.

While the white line continued. The winning competition, in which all the major European online advertising market players participated, made it possible to shine a name like we could have dreamed of. The company began to appear in reviews and on classic slides of the eco-systems of the advertising market in Europe and the USA.

But more importantly, potential customers began to find us themselves. This is how the leading European telecommunications company came to us, whose name I cannot name according to the NDA, but it will soon be three years since we successfully cooperate with them.

At the time, they were sitting on Google DART for Publishers, which we have not yet supported. Plus, they already had an optimization solution - DART Adapt. But since the volumes are very large, even a small increase in efficiency ultimately results in hundreds of thousands of euros per month. Therefore, they were ready to experiment with our approach.

As a result, we got involved in another heroic project, at the same time integrating with the new advertising server and pleasing the largest client at that time.

And the client, in turn, gushed forth with ideas, pleased that, once and for all, they come to life in weeks, not in a few years. At some point, it reached the point that more than half of the company worked only on this project. And he just stopped paying off. Despite the substantial money that we were paid, we spent a lot more on it, and therefore could not develop further - to refine the product, connect new smaller customers, etc.

We had to harshly enough to stop the process, putting out all the cards and demanding an increase in pay plus a decrease in work (dream, right?). But by that time we had so strongly promoted the client to the convenience, quality and additional functionality that, after becoming suitable for negotiations, he happily agreed.

At this moment we made another serious mistake, which cost us a lot of time and resources. Instead of stabilizing the existing solution and reaping the benefits from those potential customers, whom the product functionality already covered, the thirst for money and obligations to investors drove us to new feats.

Among all the possible products, we, as “smart”, chose the one that our new sellers promised to sell as quickly as possible, a lot and easily.

Needless to say, in the end, they failed to sell what they ordered themselves? No one. At all.
, , – , — , , ..

, (.. , , , ), - (, ). .

, ( ), SAP Business Objects (, , SAP?!)

, - SAP , .

, , , , ,
.

, . —

. , , , .

, DART. , 2010 , , , , , .

– 24/7 . , Forbes Digital, , , 10 , - ( , , ). .

24/5 24/7.

, . DART DFP6, , – .

, — . . . SWOT-. . . . Etc.

- , . . – .

. . , , FT , ( – ).
, ( – ) , , , . , . FT, , .

, ( AIR), .

. , , . , -, . .
. . .

, Forbes , , , .

, , , .

, , , .

, , . , , , . . , , , , . 15-30 , , – , – . – . , b2b-, , .

– , 24/7. , - .

, , , / , – . , . – , , « ».

, , , , , - .

, – , , , , . , . , .

. Monster.com , , , -, — 2012.

, , , Financial Times , , , , , , , – .

. , , , , , . , , .

, , .

. , , , .

, , . Dentsu, - , . , – .

, - , – , Maxifier , .

, , .

: 24/7 Media – .

, , . , . – API .

, , ( ). , - , , ( 3.5 , ), , , .

, . , , . – : - , ( – NDA).

– , , , .

– , . , , . – - . .

, , « ». . , . , , , . , , – , , , .
, : « , , ».

, . IAB ComScore ( ), : US — 6%, UK — 11%, Europe (w/o UK) — 4%, Japan — 10%.

. , . , . - . . . . .

. , , . , – , . . , – , . – , .

Nobody knows what will happen tomorrow. , , . . , , . , , . .