Client: Pintler Group

Intro & App Logic

All of this analysis and writing is done in a format called R Markdown, which lets me continue to make edits; nothing here is set in stone. So if there are any other validation checks, visualizations, or other things you’d like to see, I can make that happen.

1) I’ve run 5000 simulations of the input combinations

2) The possible inputs considered are:

  • Total budget ($18000 - $22000)
  • Website quality (1-5; which is mapped to a conversion rate range spanning from .35% up to 3%)
  • Website visits (0 - 99900)
  • Funnel Area (aka low/med/high funnel)
  • Practice Type (type of clinic)

3) The rules built into the app are:

  • Website quality is mapped to a site-specific conversion rate that helps determine the rate at which visitors convert into new inquiries.
  • Each mktg channel is mapped to a channel-specific conversion rate.
  • The overall conversion rate for each mktg channel is the average of the channel-specific rate and the site-specific rate. This mimicks the logic that the conversion rate of visitors to a site depends both on where they’re coming from (the mktg channel) and the quality of the site they land on.
  • Mktg channel CPC and CPM depend on the practice type
  • The amount of the total budget allocated to each mktg channel depends on the funnel area (low, medium, or high).
  • Website visitors are determined by the total number of clicks for all mktg channels at a given budget. Clicks are determined by channel budget and CPC.

4) The outputs are a dataframe with all the inputs for each simulation shown, plus:

  • Mktg channels (for each channel: CPC, CPM, adjusted CPC, adjusted CPM, channel conversion rate, overall conversion rate, percent of budget allocated, predicted impressions, predicted clicks, predicted conversions)
  • Total Youtube views

5) I’ve filtered the results down to just the Dental Clinics (30% increase), PT Clinics (0% increase), and Addiction Services (15% increase) because they represent the extremes of the CPC/CPM adjustments based on practice type.

Check the setup

No results here, these are just the setup variables for each channel in the app.

Channel Low funnel: $ of budget Mid funnel: $ of budget High funnel: $ of budget Min CPM Max CPM Min CPC Max CPC Channel Conv Rate
Google Search 30 20 10 30.00 39.00 4.00 5.20 0.08
LinkedIn 3 0 0 40.00 52.00 6.00 7.80 0.07
Facebook 10 15 20 7.00 9.10 2.50 3.25 0.04
Spotify 5 5 5 15.00 19.50 3.00 3.90 0.04
YouTube 10 25 20 7.00 9.10 5.00 6.50 0.04
Instagram 7 5 10 10.00 13.00 2.75 3.58 0.03
Reddit 5 0 0 13.00 16.90 2.00 2.60 0.03
Google Display Ads 0 0 5 3.00 3.90 3.00 3.90 0.01
TikTok 0 0 0 14.00 18.20 1.50 1.95 0.01
Radio Ads 0 0 0 20.00 26.00 25.00 32.50 0.00
TV Ads 0 0 0 50.00 65.00 50.00 65.00 0.00
Design Services 10 5 15 NA NA NA NA NA
Video Creation 20 25 15 NA NA NA NA NA

Results Table by channel

Spend Summary - Channel
Tactic Avg Overall Conv Rate Average Total Budget Average Channel Budget Avg Impressions Avg Clicks Avg Conversions Avg CTR (%) Avg Cost Per Lead ($)
Google Display Ads 0.017 19,987.87 349.70 102,858.33 102.86 1.78 0.10 196.10
YouTube 0.032 19,987.87 3,697.24 464,736.55 650.63 21.05 0.14 175.63
LinkedIn 0.047 19,987.87 190.01 4,197.77 27.99 1.32 0.67 143.58
Instagram 0.027 19,987.87 1,475.76 130,077.24 473.01 12.93 0.36 114.14
Spotify 0.032 19,987.87 999.39 58,683.79 293.42 9.49 0.50 105.32
Facebook 0.032 19,987.87 3,031.20 381,356.35 1,067.80 34.54 0.28 87.76
Google Search 0.052 19,987.87 3,931.53 115,452.67 865.90 45.31 0.75 86.76
Reddit 0.027 19,987.87 316.68 21,527.02 139.93 3.82 0.65 82.94
Design Services 0.025 19,987.87 2,015.46 NaN NaN NaN NaN NA
Radio Ads 0.012 19,987.87 0.00 0.00 0.00 0.00 NaN NA
TV Ads 0.012 19,987.87 0.00 0.00 0.00 0.00 NaN NA
TikTok 0.017 19,987.87 0.00 0.00 0.00 0.00 NaN NA
Video Creation 0.025 19,987.87 3,980.90 NaN NaN NaN NaN NA

Key Metrics by channel

CPL

Conversion Rate

CPL by practice type

Average cost per lead ($) for all marketing channels for each practice type (average total budget / the total of each channel’s average conversions)

## # A tibble: 3 × 2
##   practice_type      overall_avg_cpl
##   <chr>                        <dbl>
## 1 Addiction Services            156.
## 2 Dental Practice               175.
## 3 Physical Therapy              135.

Budget v Impressions

Does the association between total budget and impressions look realistic? Here, the simulations are plotted as dots, and I’ve also highlighted two practice types — PT clinics and Dental clinics — in order to show the effect of the CPC/CPM multiplier you’ve outlined. PT clinics enjoy a 0% multiplier on CPC and CPM, whereas Dental clinics face a 30% multiplier.

The second chart shows the same breakdown of budget v impressions, but this time grouped by practice type.

Conversions

Next up are conversions. How do these look?

These point charts are a little harder to read than others like boxplots that aggregate the data a bit more, but i wanted to include these here in the beginning because they can help give you a raw sense of how the app results are behaving.

For instance, the lowest simulation at $22k budget got just under 100 conversions, while the highest at that budget level got just under 250 conversions — and the website quality explains a lot of the story as to why these two results are so different.

We can also visualize the conversions by themselves, to have a better look at just that distribution.

Total new inquiries from paid mktg + organic

Here I think we need to set up the logic for baseline / organic impressions and conversions in order to be able to calculate this.

# code tbd

Total YouTube views

(youtube spend / .03)

How’s this look?

Impressions, clicks, and conversions by channel

Impressions

How do these distributions look? The shapes of the distributions are due to how the percent allocations are set for each. LinkedIn and Reddit, for instance, only ever get a small amount of budget in the low funnel scenario. Their distributions are tightly clustered and then tail off very quickly. Google Display Ads, on the other hand, get a more substantial amount of the budget at all funnel levels, hence its distribution of possible impressions more often spans a wider range.

Big thing to look at here is if these distributions make intuitive sense based on what you know of each channel’s performance (e.g. can Youtube ever actually reach 3 million impressions?)

Clicks

Same question for clicks.

Conversions

Same question for conversions.