1. Assignment – 2
ROI of sponsored Search
Group 4:
Rohit Dhanwani
Anmol Khurana
Nitish Mathur
Ravali Nimmaneni
2. 1.Why is Bob’s ROI calculation of320% for sponsoredsearch advertisingfor branded keywordson
Google possiblyincorrect?Please explainyourreasons.
Assumption
Accordingto bob“once customerlandsonpage the probabilityof bookingis12% “ if we emphasise on
the wordswe wouldinferthatbobhas not consideredCTRashe isonlyconsideringthe conversionson
landingpage howeverif we gowiththistheorywe have tofindimpressionstogive correctcalculationIt
can be derivedCTR=Clicksonads/noof clicksonsponsored+organicclicksbut thisformulaignoresthe
people whoneverclickedonanysponsoredororganicsearch.
Hence we are assumingthatBob’s12% probabiltyof bookingiscorrectandwe are goingwiththe below
theory.
Bob’smathematical calculationforcalculatingthe ROIseemstobe correct at firstglance accordingto
himthe average probabilityof makingapurchase “P” is12%, average marginperconversionis 21$
resultinginaverage revenue perclickof .12X21= $2.52 where average costper clickis .60$ resultingin
ROI of 320% .
Afterdoingsome analysis ROIfigureswhichBobcalculatedseemstobe wrongbecause Bobdidn’ttake
one of the factors intoaccount about the people whowere actually navigatedtothe website using
sponsoredad to purchase.
There mightbe lot of people whointendto make purchase frombazaar.comandwouldsearchfor itand
because we are havinga sponsoredadtheywoulduse itto reach bazaar.com.
This meanspeople whowouldhave usedorganicsearchtoreach bazaar.comcouldalso be using
sponsoredadsto reachbazaar.com.Also, people whoare usingbrandedkeywordshave ahigher
conversionrate thanpeople intheir awarenessphase this couldbe areasonforhigherROI through
sponsoredads.
Accordingto the formulaused
Average revenueperclick=Probability x marginperconversion
Where ProbabilitythatBobcalculatedseemstowrong
Accordingto the data providedwe have twosearch engines:
Google
Bing
For bothwe have organic and sponsoredclickwhile forgoogle the sponsoredclickstoppedinthe ninth
week thishelpsus understandingwhere the calculationiswrong.
Accordingto the bingdata there isno suddenjumpinnumberof clicksforthe 10th
, 11th
and12th
week
whicheliminatespossibilityof SeasonalityFactor or any otherFestival Season etc.Whereasingoogle
data we have a suddenjumpinorganicclickswhen the sponsoredad’swere down,thismeansthat
mostof the people whoare usingnavigational keywordintheirsearchwill convert evenif theydon’tget
sponsoredads.
3. 2.What analysis can you do from the data to get a better ROI calculation for sponsoredsearch
advertisingfor branded keywordson Google?Please explainanyassumptionsyou make based on the
discussionin the case, conduct the proposedanalysis and report the ROI you have calculated.
We can performCausal analysistodetermine the differencesbetweenbothsearchenginesinthe 10th
,
11th
and 12th
week. We can note that there isa suddenjumpinfiguresof google organicsearcheswhen
theirsponsoredsearch wasdownwhereasnosuchchangeswere observedinBingindicatingtothe fact
that mostof the sponsoredsearchesdidmove toorganicresults.
To Analyze further
Let ustake the figuresof google sponsoredandorganicsearches
Average
Week Week 1-9 Week 10-12
Sponsored 31390 0
Organic 125,195 148,477
Total Average 156585 148477
The average trafficof sponsoredads plusorganicads by google forthe first9 weeksis156585
Whereaswhenwe are not havinganysponsoredadsin10th
, 11th
and 12th
weekthe average trafficon
thiswebsite is(150,188+148,658+146584)/3 = 148477
If we findthe difference between these trafficswithsponsoredandwithoutsponsoredads(156585-
148477) = 8108 whichmeansthisisthe trafficthat we lostfrom website withoutsponsoredkeywordsor
inotherwords thisis the actual traffic that we were gettingdue to sponsoredkeywords.
Firstlet’s calculate howBobhas concludedhisROI
Accordingto bob 32269 were the clicksthroughsponsoredadsand12% of these i.e. .12x32269=3872
clicksare those whichgenerate revenueorare our conversions,we know thataverage marginper
conversionis21$ hence total revenue throughtheseclicksis3872X21= 81312$. Now calculatingour
investmentthatiscost perclickwhichis .60$ perclick will be 32269X.60=19361$.
Hence calculatingourmargini.e.Revenue-investment81312-19361=61951$
So our ROIcomesup to (61951/19361)*100= 320%
Now whenwe knowthe actual clicksthat were drivenby sponsoredclicks ,letuscalculate the ROI
accordingly:Actual clicksthatare driventhroughsponsoredadsonlyare 8108 out of whichaccording to
bob12% has the probabilityof bookingso8108*.12= 973, now let’scalculate the revenue generatedby
these clicksthatwill be 973X21=20433.
Total costper click isstill the same i.e.32269X.60=19361 so our margindrops to 20433-19361=1072.
Calculatingthe ROInow(1072/ 19361) *100=5.53%.