0 VisualIQ Connects Attribution to Media Buys

As I threatened – I mean, promised – a couple of weeks ago, I’ll be writing about marketing measurement systems over the next month or two. The hottest topic within this segment is probably algorithmic attribution, which uses advanced statistics to calculate the incremental impact of each marketing contact on final results. Unlike marketing mix models, which work with aggregated data such as total TV spend per month, the algorithmic systems start with individual ad impressions. The systems build a history of impressions for each person, identify groups of people with similar histories, and then attribute different results to differences in those histories. For example, there might be two groups of people, one having received three impressions and the other having received the same three plus a fourth. Any difference in outcomes would be attributed to that fourth impression. The actual details are more complicated, but you get the idea.

The advantage of algorithmic attribution is it is based on actual data, rather than arbitrary fractional weights based on someone’s opinion. (I wrote a more detailed critique of fractional attribution back in 2010;  I’m pleased to see that algorithmic attribution has become more common since then.) It goes without saying that algorithmic attribution makes more sense than assigning all value to either the first or last marketing touch.

VisualIQ was one of the pioneers of algorithmic attribution, dating back to its founding as Connexion.a in 2005. The company has grown rapidly in recent years. This reflects both greater market interest and, perhaps even more important, expansion of the product to connect directly with ad buying platforms. This makes it easier to convert the system’s findings into profitable marketing results.

Like other attribution systems, VisualIQ must start by assembling data about the client's advertising programs. The system gathers individual level data either by embedding its own pixel in Web ads, emails, and landing pages, or by importing impression history from other sources. It can stitch together impressions from several different sources to create a unified individual history. It can also import non-individual-level data, such as TV spend, weather, and competitive conditions.

The system can use this data to build two different types of models.

The IQ Envoy module builds “top down” models based on aggregate data; these are similar to traditional marketing mix models but also show the impact of one channel on another (for example, the effect of TV ads on direct mail response). This lets the system estimate the true net contribution to final results of spending in each channel.

Top down models are fairly customized, but Visual IQ has streamlined its process so it can build a new one in six to eight weeks, which is quick for a mix model. and make regular updates. Once the model is built, marketers can easily see the impact of different budget allocations by moving on-screen sliders. Constraints built into the model prevent users from creating scenarios that are unrealistic.

The IQ Sage module builds “bottoms up” models, which use individual-level data. Users can include real-world constraints such as minimum or maximum spend by channel and diminishing returns from higher spending levels. Once the parameters are set, the system generates a graph showing the best possible results for each spending level. Users can pick the level they prefer and have the system generate an optimal media plan. This can be exported to Excel for manual refinement. Or, users can send the plan directly to real time bidding platforms and online ad exchanges for automated execution.

VisualIQ has two additional modules. Insight IQ allows users to explore the data loaded into the system.  This is important because marketers may have never seen it all loaded and correlated. Audience IQ shows the demographics of best customers and results for different segments.

As you might imagine, algorithmic attribution needs a large volume of data to build an effective model. VisualIQ has found that clients need an ad budget of at least $5 million per year. Those companies typically gain a 15 to 20% increase in media efficiency.

Cost for VisualIQ depends on the modules used, data volume, number of channels and other variables. Pricing for a system including IQ Sage and IQ Envoy starts around $150,000 per year.

0 SetLogik Offers B2B Marketers a Real Marketing Database

I’ve now done more detailed research into the SetLogik B2B data management system I mentioned in my Dreamforce post.  If anything, I’m even more impressed.

I originally saw SetLogik as a tool to associate marketing leads with sales opportunities, even when they are not connected directly within Salesforce.com. That’s important in itself, since those missing links are the greatest obstacle to showing the value of B2B marketing efforts through revenue attribution.

But the bigger story, which SetLogik itself recognizes clearly, is that they’re creating a real marketing database. This has been sadly lacking in most B2B marketing automation systems, which supplement the Salesforce.com database with barely-extensible lead profiles and contact histories. In fact, I’ve recently taken to citing the B2B systems' fixed, built-in database as the fundamental difference distinguishing them from B2C systems, which connect to externally-managed databases with any structure.

SetLogik doesn’t replace the database built into the B2B systems.  Rather, it creates a separate database that merges data from marketing automation, Salesforce.com (or, potentially, any other CRM system), and whatever other sources a company has available. The matching capabilities that initially caught my eye are just one part of a larger suite of functions to load, clean, standardize,  merge, and enhance B2B data, ultimately storing it within a database where it can be used with SetLogik tools for segmentation, selection, reporting (including attribution), and predictive model-based lead scoring. Cleansed data and results such as lead scores can be fed back into CRM and marketing automation systems for direct access by their users. SetLogik’s own diagram expresses this separation reasonably well, although I would have suggested they clarify that there’s an independent, persistent database within their cloud.
 

As consumer marketers learned long ago, building a serious marketing database is a big project. The challenge is even greater in B2B, which manages two data levels, companies and contacts, instead of just one level of consumers.  It’s no wonder that B2B marketing automation vendors avoided the issue by piggybacking on the Salesforce.com structure: otherwise, the cost and complexity of building a separate database would have severely limited their growth.

SetLogik’s addresses the problem directly, by creating a nearly-automated system to build the database. The company promises to deliver a completely functional database within 60 days, and to deliver the database plus predictive lead scoring models in 90 days. Compared with the many months or years needed to deploy a traditional marketing database, this is lightning quick.

I call the system “nearly-automated” because a SetLogik analyst works with each client to set up the data preparation steps, tweaking the standard rules and processes as necessary, and because the predictive models are also built by human analysts. These are advantages, not flaws, since a skilled user adds substantial value to both processes. The system still does most of the work, so the initial data quality set-up takes just a few hours of labor – although the full process typically takes several days because clients need time to make decisions. Similarly, modeling takes about two weeks – again, more wait time than work time.  In fact, the model building is so efficient that the company includes it for free in its Enterprise edition, which starts at $1,400 per month for up to 25,000 records.

None of this would matter if the quality of SetLogik’s results were poor. But, while I haven’t run a test, the company certainly describes the features I'd want. Standard inputs include leads, contacts, campaign members, accounts, and opportunities from Salesforce.com, plus leads and activities from marketing automation. Data preparation includes standardization and verification of addresses in the U.S., Canada, United Kingdom, Australia; phone numbers verification for North America; email format verification (but not sending test emails); table-based transformations and coding for elements like titles and sales territories; and enhancement with client-licensed external data such as D&B listings.

The matching engine uses multiple rule sets, supports both similar and exact matches, and can compare data across several fields (such as mobile vs. home vs office phone number). The system will match at individual and company levels and can link individuals to companies. It will choose the best value for each field and return a consistent best record to all source systems. Predictive modeling can include derived variables, such as number of emails received, as well as raw inputs. The system’s database stores snapshots of old data values so it can track changes and trends. New and changed records run through the system at user-determined intervals that can be frequent as hourly.

The system doesn’t provide an interface for end-users to set their own data processing rules, although one is planned. As SetLogik correctly argues, very few B2B marketers have the interest or skills to do this. In fact, the company’s larger problem is that so few marketers even recognize they need better data cleansing, let alone a separate marketing database. This will likely limit SetLogik's initial clients to the upper tier of sophisticated marketers who do see the problem.  We can hope that the importance of a serious marketing database will eventually become clear to everyone.

SetLogik is a Software-as-a-Service application, available directly from the company or through the Salesforce.com AppExchange. The system has an Eloqua connector today and a Marketo connector in the works. The company promises basic implementation in 60 days, although it is usually much less, and full implementation including predictive modeling in 90 days.

Pricing is based on the system edition and number of records (unique individuals and companies). The Express Edition, starting at $500 per month for 25,000 records, builds the database and feeds the cleansed, enhanced records back to Salesforce.com and marketing automation. Professional Edition, starting at $1,000 per month, adds segmentation, list building, attribution, and other reporting. Enterprise Edition, starting at $1,400 per month, offers all the other features plus predictive lead scores. The price tag gets more serious for large systems – Enterprise costs about $11,000 per month for one million-records – but is still much less than a conventional marketing database. In fact, SetLogik points out that some services built into the price, such as address and phone verification or access to lead profiles within Salesforce.com, would ordinarily cost nearly as much as the entire SetLogik fee if purchased separately.

SetLogik officially released its system in October 2011 and now has several large enterprise clients.

0 3 Ways to Connect Marketing Activity to Revenue

Discussions of revenue attribution often remind me of the famous recipe* that begins “First, catch your hare”.  Specifically, they assume that marketers know which marketing-generated lead is associated with each bit of revenue, and then go on like medieval theologians to debate how credit should be shared among promotions to that lead.  The missing hare is that marketers often can’t link leads to revenue in the first place.

The issues will be painfully familiar to anyone who’s ever tried this. For those who haven’t, let’s start with the mechanics.  In most configurations, leads are created in marketing automation and later transferred to Sales, which creates an opportunity that eventually becomes a closed sale with revenue attached.  If all goes smoothly, the original marketing campaign and marketing-generated lead are named on the opportunity to provide the lead-to-revenue connection. 

But – spoiler alert! – things don’t always go smoothly.  When Sales creates the opportunity, it often links it to a contact record other than the original marketing lead.  Perhaps the salesperson was already working with someone else, perhaps the marketing lead wasn’t the real decision maker, or perhaps Sales just doesn’t want to give acknowledge Marketing’s contribution.  The original marketing campaign is often lost for similar reasons.


All those beautiful attribution recipes are moot if you don’t know which lead is linked to which revenue.  So let’s put down the cooking pots and go hare hunting.



The first approach is simply to get Sales to retain the marketing information when it creates the opportunity.  Let’s not dismiss this out of hand – yes, salespeople can be uncooperative, but appropriate training and management support can convince them it’s important to retain the information.  So it’s worth a try.

But let’s say you don’t have time to wait for better data or can’t get Sales to do what you need.  Now you’ll need to work a bit harder with the data on hand. 

One approach is to look for matches at the account level: build a list of marketing-generated leads, find the accounts associated with them, find the revenues associated with those accounts, and assume there’s a connection.  This could hugely overstate marketing-related revenue, since it potentially takes credit for sales that had nothing to do with marketing activity.  So you’ll probably want to put some parameters on the matches such as only including accounts with no pre-existing contacts, leads that Sales followed up on, and opportunities created soon after the marketing lead was submitted.  Setting these rules may take some serious discussion between Sales and Marketing, but that’s a good thing.

Unfortunately, there’s no guarantee that Sales will retain the leads sent by marketing or attach them to the correct accounts.  Nor is it certain that the companies listed in the marketing automation system will match the accounts listed by Sales.  In this case, you may need to build an even looser relationship, looking at company names in both systems – or even ignoring the Sales system altogether and taking data from accounting records.  Because the same company may be listed differently in different systems, this sort of matching requires either knowledgeable people or comprehensive reference databases that can make the non-obvious connections.  Fortunately, this is a well understood problem and plenty of resources are available to help.

Company-to-company matching casts an even wider net than lead-to-account matching, so it’s correspondingly harder to give marketing credit for every connection.  But you can rate how likely it was that marketing played a role in a given opportunity by looking at factors like timing, pre-existing relationships, and amount of marketing activity.  This could translate into allocating a fraction of the revenue to marketing, ultimately a more realistic, if less satisfying, approach than taking full credit for some deals and no credit for others. 

If fractional allocation strikes you as too complicated, you can also start with a much simpler question: did companies that interacted with marketing programs show more sales than similar companies that didn’t interact with marketing programs?  You won’t be able to prove that any particular contact generated any particular deal, but a strong correlation between marketing programs and revenue growth is good evidence that marketing had an impact.  Once you’ve captured that hare, you can think about the details of how you’ll cook it.

__________________________________________________________________________

*Jugged Hare in Hannah Grasse’s The Art of Cookery Made Plain and Easy, although it apparently  doesn’t include the “catch your hare” part.

0 Marketo's Enterprise Edition and Revenue Cycle Management: Looking Under the Hood

Summary: Marketo continues to follow its own path. Enterprise Edition adds the complex security needed by large organizations but sticks to simple campaign flows. Revenue Cycle Management blazes an important new trail for others to follow.

I finally caught up with Marketo for a briefing on their Enterprise Edition (announced in March) and Revenue Cycle Analytics (announced in May). Since both are somewhat old news, and Marketo describes them in detail on its Web site, I’ll just make a few comments.

Executive Edition shows what Marketo believes is needed to service large marketing organizations. The most extensive enhancements provide finer-grained control over user rights. This is critical in large organizations, where regional and product groups may be responsible for different market segments and where users will have different functional specialties and approval authorities. Enterprise Edition supports these by adding user roles, “lead partitions” to control access to database segments and “workspaces” to make Marketo objects (contents, campaigns, lists, etc.) available to different user groups. User roles (but not lead partitions or workspaces) are now available in Marketo’s Professional Edition as well.

These changes are a big advance over earlier versions of Marketo, which distinguished only between users and administrators and let all users access pretty much everything. Enterprise Edition also adds a “sandbox” environment for training, testing and development – the sort of things that small companies might do on a live system, but large organizations cannot safely allow.

The other major big-company need that Enterprise addresses is more sophisticated integration with other corporate systems. Related features include LDAP integration with enterprise security systems and a Web services API to call Marketo functions and access its data.

Perhaps most interesting is that Marketo did NOT expand the complexity of its actual campaign flows. These remain fundamentally linear: that is, all leads follow the same flow from step 1 to step 2 to step 3, etc. Rules within each step can deliver different treatments to different segments, but everyone still moves to the same next step unless they leave the campaign altogether. Other enterprise-level marketing automation systems can create different branches within their campaigns, so different segments follow entirely separate paths. This makes it easier to design and visualize fundamentally different treatments for different types of leads, something that matters more in a large enterprise with many different lead types. I’ve always considered branching campaign flows to be one of the key requirements for an enterprise-level marketing automation system. It seems that Marketo disagrees.

(Actually, Marketo disagrees with much of the preceding paragraph. Everything in it is factually accurate, but I'm happy to clarify that (1) several campaigns can run simultaneously, sending leads through different flows and (2) steps within Marketo campaigns can remove leads or send them to other campaigns (3) Marketo can connect several campaigns to produce the same flows as single branching campaign in other systems.)

Revenue Cycle Analytics breaks some important new ground. As I commented in an earlier post on purchase funnel measurement, Marketo’s approach is not conceptually unique. The basic idea is to track leads through stages in a purchase funnel, which is similar to pipeline reporting in many sales automation systems. It just starts earlier in the process.

However, Marketo's implementation brings this reporting to a new level. Most specifically, Marketo has introduced a star-schema reporting database, which I’m pretty sure no other marketing automation system currently offers. (Market2Lead had something similar but is no longer sold.) This is important because the structure of an operational marketing database, which most B2B marketing automation systems also use for reporting, makes it hard or impossible to do the necessary time-based analysis.

Other components are similarly sophisticated. These include graphical models that track movement of leads through the stages, detailed analytics with specialized measures such as conversion rates and speeds, statistical projections based on current inventory and historical flow rates, and executive dashboards. The models capture more than a simple linear pipeline: they support skipping and backwards flows among stages, splits within flows for different lead types, complex stage definitions, and transitional stages where leads are processed and reassigned.

Marketo is also tackling the difficult issue of allocating revenue to multiple individuals and marketing touches. Its methods are not particularly advanced: credit can be spread evenly or based on marketing-assigned weights. But no one else has found a much better solution, particularly at the low volumes of most B2B marketing programs.

My only real complaint is that you can't actually buy it all today. Marketo is releasing Revenue Cycle Analytics in stages. The database itself was available for the May announcement and the modeling engine was released in July. Initial analytics are set for delivery this month (August), with the really cool projections and dashboards out during the first half of next year. This delay could prove costly, since funnel-based marketing measurement is a hot topic and other vendors could well build or partner to deploy something similar in the interim.

Pricing of Revenue Cycle Analytics starts at $1,500 per month and grows with database size. Incidentally, I don’t think they’ve published that figure anywhere before, so there’s a bit of news in this post after all. Huzzah.