REDACTED FOR PUBLIC FILING
EXHIBIT
PSX00204
GOOG-DOJ-04372072
Message
From:
B
ene
d
ict
Gomes
Redacted@googl
e.com]
Se
nt
: 3/23/2019 3:35:16
PM
To:
Nick
ha
CC
:
Abhinav
Taneja
Redac
ted@google
.com]
Subject
: Ads cy
Hey - here's what I want to send to Pragh prior to the meeting. W dyt?
Not helpful? I have to admit that
I'm
feeli ng annoyed both personally and on behalf
of
the team. This may be a
useful framing too. Maybe worth turning into a doc?
ben
Definitely the main topic. I think there are many things being done, and there is misunderstanding between the
teams that we need to sort out.
- I think your team somehow feels that we don't care about usage. This is
jus
t not true. Not true. And this is
causing a lot of angst on both sides. Almost
all
the team now does (a pivot I got Amit to start a few years ago)
is growth motivated. We do care very deepl
y.
Saying that we don't feels very strange to my team.
- most headcount (non assistant) for search has gone into projects
tha
t are growth oriented.
To
the point I worry
that we are really not investing in research or speculation adequatel
y.
However, our ability to predict growth
associated with work we do
in
the short term (months) is still limited. The longer term (years) seems like there
are very clear correlations between growth and investment and vice versa (no .investment leading to declines).
- we have spun up efforts with your team to look at
- desktop (by porting over mobile UI)
- UI tweaks to improve access points (0 query suggest, query box size etc). suggest ranking etc.
- latency efforts which are a high priority already but will get more attention.
- the explore team is working with the goal
of
increasing user journey length.
- Aside from latency, we have historically not been able to move queries
in
the very short term
in
a meaningful
way. Some suggest changes do so, and some system changes do so.
- there is angst around the metrics - I think this angst is excessive since all good metrics will be correlated - but
we are moving to a metric that will hopefully keep both teams happier and more in sy
nc.
- there are issues around forecasting that have not been fully resolved.
We can improve *engagement* in the short term. I know ways we cou
ld
do this. T
hi
s is the *equivalent
of
rpm
heroics*, but that does not he.Ip you! But we don
't
have the levers (or muscle) to increase queries in the code
yellow way. To a certain extent we wi
ll
develop a bit
of
this with a tweaks type approach to entry points, even
more focus on latency etc. But imho this is
just
far more limited than increasing engagement.
We could increase queries quite easily in the short term in user negative ways (turn
off
spe
ll
correction, turn
off
ranking improvements, place refinements all over the page).
If
we, as a company, want to go there we should
discuss that. It is possible that there are trade offs here between different kinds
of
user negativity caused by
engagement hacking. But I will say that I
am
deep
ly
deeply uncomfortable with this, and
I'd
be surprised
if
the
ads team wants this. The nature
of
how you would easily increase queries is a key reason I do
n'
t like queries as
an end metri
c.
The easy ways are almost all bad.
Ha
ving queries as a metric will, in my opinio
n,
have a subtly
bad effect as a launch metric even if we 'decide not to do the bad things'.
My claim is that our best defense against query weakness is compelling user experiences that make users want
to come back. This is why I care so much about organic shopping info as a driver
oflong
term query growth - it
Redacted