# Marketing Valuation

In this series of articles I’m talking about the 5 steps of digital marketing attribution. Classification, Pathing, Attribution, Valuation and Optimisation. In the last article I covered the mechanics of attribution modelling, building on the detailed articles about pathing and channel classification. Today we’ll expand on how we estimate value for marketing channels having applied an attribution model.

In the last article we talked about how to value marketing channels from web tracking paths. Let’s take an example of a few paths along with the overall valuation which we would give to the channels from the last article.

```PPC -> SEO -> SEO -> CONVERSION
DISPLAY -> SEO-> PPC -> CONVERSION
PPC -> DISPLAY -> CONVERSION

Last touch)
SEO : 1
PPC : 1
DISPLAY : 1

Linear)
SEO : 1
PPC : 1.167
DISPLAY : 0.833
```

The assumption that this is built on is that all conversion events are equal. In reality this is very rarely the case.

Valuing a conversion event

Imagine that we’re running an online shop. Let’s go back to our earlier conversion events and look at the value of the conversions.

```PPC -> SEO -> SEO -> CONVERSION £100
DISPLAY -> SEO -> PPC -> CONVERSION £40
PPC -> DISPLAY -> CONVERSION £5
```

The different conversion values here are far from unusual, and in an ideal world we should factor this in when we value our marketing channels. To do this we simply multiply all channel contributions by the value of the conversion. This has a pretty dramatic impact on how much we value the channels in these example conversion journeys.

```Last touch)
SEO : £100
PPC : £40
DISPLAY : £5

Linear)
SEO : £80
PPC : £15.83
DISPLAY : £15.83
```

Some of you may be wondering whether or not this is really sensible, as perhaps on another day Display would have brought in the high value conversion. That’s absolutely true, and this process only works by looking at lots of conversions. If the value of conversions is evenly distributed across all channels, then a value-weighted model will return the same trend as a pure conversion model. However, there are plenty of real-world examples where prospects from different marketing channels go on to have different values to the business, in general you will learn more by factoring this in.

You can also use these measures to estimate the value of groups of keywords for PPC, or individual marketing partners for display. This helps you to spend more on the marketing activities that are bringing real value to your business, while lowering your spend in areas that haven’t currently proved their value.

If you have a low number of conversion events then you should also consider looking at things like Linear attribution models as they’ll help make the most of sparse conversion data.

Valuable Conversions as a boolean metric

If we have particularly skewed conversion values it can sometimes be useful to convert the conversion value into a valuable conversion metric. Imagine you’re selling cars that range in value from £200 to £2 000 000. Perhaps you make no profit on cars that cost less than £1000, and you only sell 30 or 40 cars per year at £100K+. If you use the price of the car as the conversion valuation then it’s probable that you’ll end up overweighting the channels that are involved in high value conversion.

Maybe all £100K+ conversions come from PPC terms like Ferrari, Lamborghini. The last thing you want to do is overweight these terms in PPC because it’ll strangle your mid-range value conversions, which are probably where the majority of your revenue comes from. To avoid this, introduce a metric called “valuable conversion” which is only set where a car is worth £1000 or more. By optimising to this new metric we take out any weighting for unprofitable conversions, and lower the contribution of the rare super-valuable conversion events.

If you’ve got some decent backend systems then you may be able to do better than looking at the value of just the first purchase. By looking at valuation across the entire lifetime of a customer you can gain insights about the types of marketing that lead to return visitors.

This isn’t something that works for all companies, but if your business model is based on long term client engagement then lifetime value is likely to show up some interesting trends in acquisition models. A good example would be an online grocery store which is aiming to signup customers that keep coming back for more.

The downside of lifetime value metrics are that you need to wait some time to evaluate whether your marketing is performing well based on this metric. That probably means you should only optimise to these metrics every few months rather than using them as realtime marketing performance measures.