The Vale of Tiers
Promises and Pitfalls of Tier Pricing
SSP 29th Annual Meeting
San Francisco
Douglas LaFrenier
Director, Publication Sales & Market Development
American Institute of Physics
Some caveats
• I am speaking from the perspective of
scientific research journals, with a mature
subscription base.
• All my sample data is from AIP journals
alone.
• I don’t pretend to know all the issues for
other types of publisher or even other
science publishers.
Pricing in the Print World
• With print, pricing was a simple function of
costs, number of customers, and desired
margins (profits).
• The “behavior” of the customer – the
library and its patrons – had nothing to do
with it.
• Scholarly publishers were much like hard-
goods manufacturers in that sense.
Pricing in the Print World, cont’d
• A “big” customer was one that subscribed
to multiple copies. E.g., Princeton
University for years had about 8
subscriptions to one AIP title, Applied
Physics Letters.
• But for many specialized titles, there was
essentially one copy for each institution.
So, from the publisher’s point of view, MIT
= Vassar = Arizona State.
Online is a vastly different world.
SumOfArticleDw nls2005
10,000
30,000
50,000
70,000
20,000
40,000
60,000
80,000
90,000
0
10012985
96002161
NA626465
10060329
M I114520
10011582
CO825792
NO799770
TO803278
GE790912
E 948124
10009359
10012580
93000625
10035092
TW392917
PE401250
10038858
10012806
AR166565
Top 100 AIP accounts range in downloads from 17,102 to 84,330.
SumOfArticleDw nls2005
10,000
30,000
50,000
70,000
20,000
40,000
60,000
80,000
90,000
0
10012985
NA62403
10046766
PE326250
10002258
CO825792
97005409
CA143559
10066556
10004223
10037588
10064191
10080010
10267583
10279292
AT724122
DR06600
IN189369
TE804019
But bottom 300 accounts have 1-10 downloads each. In fact, 33% of
our accounts have zero usage.
SumOfArticleDw nls2005
10,000
15,000
20,000
25,000
30,000
35,000
40,000
45,000
50,000
55,000
60,000
65,000
70,000
75,000
80,000
85,000
5,000
0
Acct 1
Acct 16
Acct 31
Acct 46
Acct 57
Acct 72
Acct 87
Acct 102
Acct 111
Acct 126
Acct 141
Acct 156
Acct 165
Acct 180
Acct 195
Acct 210
A cct 234
A cct 249
A cct 264
Acct 279
A cct 294
A cct 309
A cct 324
A cct 339
Acct 354
Selected mid-range accounts in each of AIP’s 6 tiers.
Distribution of usage
Ratio of accounts to downloads (2005 data)
60
50
40
30
% downloads
20 % accts
10
0
0-99 100- 501- 1k- 5k- 10k- 50k+
500 1k 5k 10k 50k
• 50% of all downloads come from the top 4% of accounts.
• Only 4.7% of downloads come from the bottom 74% of accounts.
Thus, the rationale for tier pricing.
• It’s a question of fairness:
– Should smaller institutions have to pay the same as
bigger, research-heavy institutions?
• It’s a question of value:
– Since the value derived by the smaller institutions is
so much less, aren’t they much more likely to cancel
(raising prices for remaining subscribers)?
• So, the real purpose of tier pricing is not so
much to “tax” the heavy users as to relieve the
burden on the smaller users.
Downloads are not everything.
• Strict usage-based pricing is unlikely for several
reasons:
– Utility-company-like pricing is appealing, but high-end
users would bust any library’s budget.
– We don’t want to discourage usage, see limits
imposed by the institution, etc.
– There can be plenty of usage without downloads –
e.g., read abstracts free, search database, use
current-awareness tools.
• Other measures might include authorship,
subscriptions, research activity, membership,
size, GDP, academic v. corporate, etc.
Problem #1
• Because there are more low-end users
than high-end users, it’s very hard to
minimize prices for smaller users without
clobbering big users.
– In a revenue-neutral scenario, high-use
accounts will have to pay a disproportionately
high price increase.
Scenario 1: Publisher has 1000 customers for a $500 journal and plans
a 6% price increase. Each customer now pays $530 and publisher
revenue is $530,000.
Scenario 2: Publisher establishes 3 tiers, plans a 2% and 4% increase
for lower users, and still expects $530,000 in revenue.
Tier Type No. % Incr. New price Revenue
T3 High 100
users
T2 Med. 300 1.04 $520 $156,000
Users
T1 Low 600 1.02 $510 $306,000
Users
Tier Type No. % Incr. New price Revenue
T3 High 100 1.36 $680 $68,000
users
T2 Med. 300 1.04 $520 $156,000
Users
T1 Low 600 1.02 $510 $306,000
Users
For the same revenue, T3 must contribute
$68,000, resulting in a 36% price increase
for high users!
The problem would be worse if one wanted to
actually reduce prices for low-end users.
• Is this why we don’t see the price-tiering
phenomenon among commercial
publishers?
– The total bill is just too big.
– High users can perhaps pay a one-time 20%
increase on $20k package, but not on a
$2MM package.
– (Bigger problem is likely that commercial
publishers have multiple disciplines, product
lines, customer types.)
In any case, it’s best to move high-end users
incrementally and avoid “sticker shock” in any one
year.
Problem #2: How to Tier?
• External measures
– Carnegie Mellon, JISC – but no direct foreign
equivalents
– Descriptive: Universities with PhDs, Liberal Arts
Colleges, Two-year Colleges, Corporations, etc. –
sometimes size-based as well
– Size-based (but need to define what FTEs are
counted)
– Research productivity (as defined by article output)
• Internal measures
– Downloads (and other activity?)
– Authorship, membership
– Subscriptions
– Intrinsic metrics of each journal, e.g., impact factor
Problem #3: The problem of the
cusp
• In many schemes, especially using
internal measures, there are no hard lines
between tiers. An institution could easily
fall just above or below a given threshold.
• Is the publisher prepared to handle an
appeals process? To review other data,
such as prior cancellations or budget
conditions?
• This is made worse by . . .
Problem 4: The problem of volatility
• Usage activity changes, with a potential
yo-yo effect on institutions near the cusp.
No one wants to be re-tiered every year.
• AIP has adopted two strategies for this:
– The “two-year rule”: Any change in activity
has to hold for two years before we reassign a
tier.
– The “significance rule”: Changes to our
assigned “research activity index” have to
exceed 20% to be considered significant.
Problem #4: The problem of administration
• You’re in for it now. This is a lot of work!
– Research, planning, validating,
communicating, responding . . .
Problem #5: Transparency and trust
• Publishers should not use tiering to jack up
revenue, but the potential exists. Who validates?
• For this reason, it’s very important to involve
the library community in your process.
Problem #6
• Who the heck knows what’s going on
anymore?
– With tiering, consortia licensing, multiple
product options, differing publisher policies (re
archiving, for example), negotiation, etc., it is
now impossible for one customer to compare
prices with another.
– Is it therefore going to be harder for librarians
to keep publishers honest?
Problem #7
• And, of course, none of this has anything
to do with actual pricing, that is, what
prices a publisher establishes for each
product and tier. The fairest tiering
scheme in the world could still result in
too-high prices.