Not all leads are created equal. And neither are marketing campaigns. Particularly when it comes to lead gen. By optimising to better lead gen KPIs, ROI improvements of 2X and above are easily achievable. We lift the lid on how.

Kinase has been fortunate to work with a wide range of lead gen businesses in verticals including:

Home improvement (fitted wardrobes and shutters)
Services (estate agents and recruitment)
Luxury Holidays

What they all have in common is simple: a desire to drive better leads, which convert at a higher rate and ultimately generate more revenue.

But at the moment a lead is generated, it’s hard to tell which will be profitable and which will not. It’s even harder to focus advertising spend on the strongest leads. “Maximise leads” bid strategies will generate a lot of leads - but they’ll include an increasing composition of weak prospects. They may be cheap, but not good value for money.

This article will explore:

Why this matters
What’s changed in how ad accounts are optimised
How to measure and upload better KPIs to capitalise on modern algorithmic optimisation

Why is this important?

Distinguishing the most valuable leads has always been important, in order to set the right budgets, allocate the right bids, and avoid wasting valuable salesperson time on non-converting leads.

Recently it has become even more important. In the past we have been able to double a client’s ROI at the keyword level (by eliminating search terms with a high ‘click to lead’ rate but a poor ‘lead to conversion’ rate). But as Google takes more control over search terms and bids, we must now dig into the data layer to make such dramatic improvements.

Machine Learning algorithms deployed by Google and Meta are extremely efficient at optimising to the given KPI.

If the KPI you give is ‘maximise leads’, that is exactly what it will do. It will hit your CPL target regardless of whether those leads are good or worthless.

In fact, you might have noticed lead quality decline. Without a signal indicating which leads you want, your campaign won’t just optimise toward any lead regardless of quality, the AI will actively prioritise the worst leads because they tend to be the cheapest.

New Thinking

The onus is on advertisers to inform the algorithms which leads they value. Only then can the full power of modern marketing be harnessed as the engines learn to correlate the vast quantities of data they have access to (demographics, search history, etc), with the desirable outcomes that are fed in.

What are these outcomes? They will vary considerably between businesses. Often a financial transaction (which may occur offline) is the final positive result and feeding this in can be valuable to optimisation. However due to time lags or limited volumes, this is rarely enough by itself to give the algorithms sufficient learning data for fully efficient optimisation.

Better KPIs

The right data to pass back for optimisation will vary as widely as business models do. When working with clients to select the right data, we consider the following:

Final Outcome

Select a metric that represents the desired final outcome as closely as possible. A metric with an imperfect but good correlation is a lot better than nothing. Completing the first steps in the purchase journey is crucial - and once leads get stronger and profit increases, getting the budget to push further steps will be much easier.

Data Volume

The algorithms need enough data to build a predictive model of which leads will prove valuable. Where the volume of completed transactions is too low, completion of a qualifying step could work.

Time Lag

The faster the signal of lead value is passed back, the more optimisable it is in real time. With long paths to conversion (say a month or more) passing more timely signals will be beneficial.

All three must be traded off against each other to find the right KPI. This may also evolve over time, for instance as measurement improves or marketing volumes build.

It may also be that a hybrid metric built from the different signal strengths of available touch points needs to be built in order to drive the right combination of leads. For example, a direct phone call, an email sign up and a brochure request might have radically different conversion rates to your goal, but if each is weighted correctly, optimisation can focus on the conversions behind how the lead comes in.

If your business is lucky enough to have large volumes of converted leads within a short duration from the initial enquiry, transaction value is likely sufficient. If not…

Other Metrics

What is a good predictor of a lead which will convert within your business?

Form Completion

If it can be extracted from the data that the customer submitted upon lead creation (say, a combination of the product they are interested in and their budget) an automated score might be passed back immediately

Lead Scoring

Perhaps a sales agent can quickly assign a probability upon contacting the lead. (Even informing the system that the contact details provided were invalid is a powerful signal not to optimise for those sorts of leads).

Path to Conversion

Are there key moments in the sales journey that indicate a lead moving toward a successful conclusion. E.g. a brochure download, sample request or completed design sitting? These are all valuable.

Of course, more than one metric can be measured to get a combination of immediacy and correlation. And it’s important to track customers through each stage, in order to predict future closed sales and improve the modelling and optimisation over time.

Independent verification via geo-testing or Media Mix Modelling is also highly recommended.

Other Considerations

One of the best parts of working with lead gen businesses is that each is unique, and there are often specific requirements which can be factored into optimisation.

To provide a flavour, one is localised supply and demand. For instance where sales agents need to make a physical visit and have a geographical footprint that they cover. Here the value of a lead to a business varies not only based on external considerations, but also based on agent availability (there’s no point paying to send high quality leads to an agent with no capacity to service them). Incorporating this sort of 1st Party data takes optimisation up another notch.

How to Pass It

Once the improved metrics have been defined, they need to be passed back to the engines via a privacy-safe data integration that allows Google/Meta to tie it back to their users. It also requires integrating into reporting, budgeting and so on, which is a whole separate topic.

Impact

The more sophisticated other advertisers get at skimming off the most valuable leads, the greater the penalty to businesses which optimise to cruder metrics like CPL, have scoring that poorly correlates to outcomes, or have overlooked volume/latency considerations.

But even the early adopters working with Kinase have seen considerable benefits from leveraging modern algorithmic optimisation in a way that ecommerce retailers largely take for granted, with performance gains typically exceeding 50%.