Marin Bidding

The Challenge

Are the limitations of publisher bidding holding you back?

Are you only advertising on Google? Didn’t think so. Don’t you want a bidding solution that works across publishers and channels? There are many factors that impact your performance as a business that are outside the auction – think ratings, social media buzz, and new product launches. Shouldn’t the machine learning calculating your bids know about these things? Are you optimizing to all downstream events, or just “leads”?

And should you be thinking about what is Google doing with your data? Do you want to share all of your conversion and profitability data?

Maybe it’s time for a better approach?

Marin Bidding unlocks performance across publishers

Marin’s automated bidding translates your business objectives into the appropriate pay-per-click bid. Marketers can choose from various goals, including spend, volume, or efficiency.

Marin bases its bidding decisions on your ‘source of truth’ for conversion data, with flexible support for customer lifetime value, call tracking, and offline conversions. Initially, Marin brings in 12 months of historical data so we can start bidding accurately from day one.

The main differentiators of Marin’s bidding solution is our focus on transparency, control, customization, and long-tail optimization.

Flexible setup with Strategies

Strategies in Marin are groups of campaigns with a common business objective, including spend, ROI, or impression share. Strategies work across accounts and publishers to make managing your performance easy. A typical brand would have 5-10 strategies, but there is no limit on the number of strategies or campaigns per strategy.

Set your objective and Marin does the rest

Once campaigns are mapped to a strategy, all you need to do is provide the business goal for that Strategy. Strategies can target one or more of the following:

Efficiency (ROAS or CPA)
Volume (Clicks)
Engagement (page views, time on site)
Awareness target (impression share)

Marin calculates the optimal bid for each object based on the predicted performance of each campaign. See below for how we handle seasonality, data sparsity, and many other aspects of the algorithm.

We aim to give you the controls you need to customize the algorithm for your business. Marin offers Bidding Rules to control the algorithm's changes and Dynamic Actions to incorporate external signals into the calculations.

Track the entire customer journey with conversion types

Marin Bidding accommodates businesses that track multiple conversion events across the customer journey. These events can be online or offline and have different values. By tracking the entire customer journey, we’re able to optimize the bids to the business value rather than a blended average. For example, using different values for new and existing customers or stages in a conversion funnel.

Preview bids before they are sent

Some users run their bid strategies in preview mode shortly after setting up a folder or when they significantly change a folder. Preview mode allows users to QA the bids Marin is calculating before setting them live and allowing Marin to push bid changes automatically.

The power of Smart Bidding for all publishers

While they both have their limitations, the bidding tools on Google and Facebook are pretty good. The other performance marketing publishers, including Amazon, Apple Search Ads, and LinkedIn, are further behind. With Marin bidding, you have the power of Smart Bidding across all of the platforms.

Marin Bidding solves the optimization problems you face daily

Determine the right level of investment with Forecasting

Marin’s forecasting tool is a fully transparent, predictive model that allows advertisers to visualize the return at different spend levels. You can see the effects of spending more or less over the month on metrics like clicks, cost, conversions, profit, and gross profit.

Forecasting is handy for seeing the effects of a budget reduction before committing to it – or making the internal case for a budget increase.

Optimize to actual business value across the complete customer journey

For longer sales cycles, Marin's full-funnel bidding uses upper-funnel conversions for responsiveness while factoring in final sales values for accuracy. Better data leads to more efficient campaign and revenue growth without losing responsivity to recent market changes.

Incorporate first-party signals into bidding algorithms

Publisher bidding systems don’t know about your business and environment, but your bidding algorithm should be able to ingest external data and factor it into its predictions. Examples include:

Inventory data
Other marketing activities
Weather, stock, interest rates
SEO performance

Dynamic Actions uses these external signals to apply positive or negative boosts to the bids. The boost adjusts the bid calculated by the algorithm.

Coordinate SEO and SEM campaigns

Harness the power of SEO data to build flexible rules that adjust bids based on trends in this contextual data. Marin links SEO ranking to our bidding algorithms for improved optimization. For example, a client could automatically bid down PPC keywords in organic position one and, conversely, bid up on PPC keywords when the corresponding organic keywords have a low ranking.

Advertisers can apply boosts to a group of objects using dimension tags. Let’s say an advertiser wanted a group of campaigns to generate more conversions, they could tag those campaigns with a dimensions tag and apply a boost of 30% to any object with that dimension tag.

Target an awareness threshold

Awareness Targeting as a bid strategy allows advertisers to have Marin automatically bid to maintain a preferred impression share or impression rate for a given set of keywords. Advertisers can choose from multiple metrics and set different goals / targets by device (desktop, mobile, or tablet) – or simply leverage one visibility target across all devices.

Automatically adjust bids based on day or week and time of day performance

Marin can automatically adjust bids according to performance variance by day of the week and (for Google campaigns) by hour of day, optimizing for seasonality and maximizing ROI.

Additionally, users can leverage Marin’s Scheduled Actions at the creative level to pause or resume ads at any date or time. This additional control allows advertisers to save time by easily scheduling promotional updates or ad copy changes for a future date or time.

Boost bids for upcoming promotions

When your conversion rates spike due to sales or other promotions, most bidding systems won't adjust quickly enough. Put your thumb on the scale with scheduled boosts that increase your bids during the sale and then revert back once it has ended.

Coordinate bidding across multiple brands

Channel partners or companies with multiple brands often want a specific brand to appear at the top of the auction when they’re bidding on the same or similar terms. Marin automates this coordination to ensure compliance and avoid the brands unnecessarily bidding against each other.

Respond to quickly changing conversion value

Some companies, especially in the lead generation space, have a value per conversion that can quickly change, even throughout a day. This variability wreaks havoc on a bidding system relying on a stable value per conversion. With Marin, you can automatically adjust the value per conversion programmatically, so the bids are based on the current value per lead. The result is less wasted spend and fewer missed opportunities.

Getting under the hood with the algorithm

Highly reactive AND effective on the long tail

Marin Bidding can handle high-volume and long-tail separately. For high-volume objects, bids are based on that object's most recent data. For the long tail, Marin Bidding expands the historical lookback to include data from similar objects and uses Bayesian blending to calculate accurate bids. Such an approach provides a demonstrable lift with a unique model for low-volume keywords that result in intelligent bids, even for keywords with zero clicks. Marin starts bidding on new ads immediately without a burn-in period.

Frequent calculations to keep up with a changing landscape

Marin bids between one and six times daily, depending on the use case.

Adjustments for device, audience / demographic, geography, time of day, intent, and many other factors are applied in real time during Google’s auction process.
Changes to Strategy targets calculate new bids immediately
Impression share / rate targeting strategies calculate and traffic on average twice per hour
Economic-based and budget allocation Strategies re-calculate and traffic within two to three hours on average (five times per day).

Rapidly adapts to seasonal changes

Marin Bidding proactively predicts the impact of day, week, month, and year seasonality to eliminate missed opportunities and wasted spend.

When a user knows there will be a significant change in performance, for instance, the start of a traditionally busy period or something unforeseen like a competitor cutting their prices, a manual boost can adjust the algorithmic bids. Boost is a rule that lowers or raises the Marin calculated bid by a percentage to allow users to immediately manipulate the algorithm to account for contextual factors. The boost percentage reflects the expected change in performance.

Tune the algorithm to match your business

The above settings are enough for most advertisers; however, additional customization enables bidding to respond to specific use cases. Our data science team works with you to fine-tune the algorithm.

Excluded dates with unusable data

You can exclude specific dates from consideration when calculating a bid if there has been a period of unusual performance that would skew the bid calculations or if you have a data integrity issue.

For example, an advertiser might use excluded dates:

When a massive one-time TV promotion runs, which may have an unusual effect on paid search conversion rates.
If someone accidentally removed tracking pixels from the site, resulting in dates with missing or partial data.

‍

Adjust for latency in the purchase cycle

Marin’s team will work with the advertiser to identify the proper bid strategy based on the latency between click, initial conversion, and any subsequent latent conversions.

Suppose the sales cycle is short; for example, 90% of purchases happen within three days of completing a lead form. In that case, advertisers can exclude a rolling time frame from bid calculations (i.e., three to seven days) to ensure complete performance data influences the bid calculations.

Manually override calculated bids

Clients can use bid override to take back manual control of specific keywords if needed. Bid override is a feature that enables manual bidding, allowing users to temporarily or permanently exclude keywords from the folder bid calculations.

Users can set specific time periods for these manual bids to remain in place or set indefinite overrides.

Reduce risk with bidding rules

Marin allows users to constrain the algorithmically calculated bids. Unlike competing solutions offering separate rules-based and model-based solutions, these adjustments are fully integrated with the algorithm.