CN111598631A - Cross-channel keyword price optimization method with effect target oriented - Google Patents
Cross-channel keyword price optimization method with effect target oriented Download PDFInfo
- Publication number
- CN111598631A CN111598631A CN202010646498.4A CN202010646498A CN111598631A CN 111598631 A CN111598631 A CN 111598631A CN 202010646498 A CN202010646498 A CN 202010646498A CN 111598631 A CN111598631 A CN 111598631A
- Authority
- CN
- China
- Prior art keywords
- consumption
- rcpa
- project
- keyword
- activity
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 230000000694 effects Effects 0.000 title claims abstract description 153
- 238000000034 method Methods 0.000 title claims abstract description 73
- 238000005457 optimization Methods 0.000 title claims abstract description 42
- 238000006243 chemical reaction Methods 0.000 claims abstract description 29
- 230000003247 decreasing effect Effects 0.000 claims abstract description 7
- 230000008569 process Effects 0.000 claims description 22
- 230000001603 reducing effect Effects 0.000 claims description 15
- 238000004364 calculation method Methods 0.000 claims description 9
- 230000009471 action Effects 0.000 claims description 6
- 230000007423 decrease Effects 0.000 claims description 5
- 235000013399 edible fruits Nutrition 0.000 claims 2
- 230000001351 cycling effect Effects 0.000 claims 1
- 239000000446 fuel Substances 0.000 claims 1
- 239000007788 liquid Substances 0.000 claims 1
- 230000002354 daily effect Effects 0.000 description 6
- 230000002596 correlated effect Effects 0.000 description 3
- 230000003542 behavioural effect Effects 0.000 description 2
- 230000000875 corresponding effect Effects 0.000 description 2
- 230000009467 reduction Effects 0.000 description 2
- 230000000717 retained effect Effects 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 235000015392 Sesbania grandiflora Nutrition 0.000 description 1
- 244000275021 Sesbania grandiflora Species 0.000 description 1
- 230000001174 ascending effect Effects 0.000 description 1
- 230000003139 buffering effect Effects 0.000 description 1
- 238000013075 data extraction Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000002222 downregulating effect Effects 0.000 description 1
- 230000003203 everyday effect Effects 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 230000008676 import Effects 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000012804 iterative process Methods 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 230000001737 promoting effect Effects 0.000 description 1
- 230000002441 reversible effect Effects 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 235000013599 spices Nutrition 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0241—Advertisements
- G06Q30/0273—Determination of fees for advertising
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0241—Advertisements
- G06Q30/0249—Advertisements based upon budgets or funds
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0241—Advertisements
- G06Q30/0273—Determination of fees for advertising
- G06Q30/0275—Auctions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0241—Advertisements
- G06Q30/0277—Online advertisement
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Accounting & Taxation (AREA)
- Development Economics (AREA)
- Strategic Management (AREA)
- Finance (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Economics (AREA)
- Marketing (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses an effect target oriented cross-channel keyword price optimization method, and relates to automatic optimization of search keyword prices for search engine marketing according to own conversion data of advertisers. A method for automatically optimizing the price of a search engine marketing keyword is disclosed. Aiming at the condition that one marketing item has a plurality of advertisement putting activities on marketing platforms of a plurality of search engines, the advertisement has self conversion data which cannot be collected by a channel. After a CPA target, a project overall budget and other parameters are specified, historical data and self-owned conversion data of a plurality of channels can be fused, statistical indexes of a plurality of layers are calculated, and then the keyword price of each activity is optimized by automatically increasing and decreasing advertisement expenses of a plurality of activities in combination with the CPA target and the overall budget requirement, the overall delivery effect is improved, and finally the expected CPA target is achieved.
Description
Technical Field
The invention relates to automatic optimization of Search keyword price of Search engine marketing (Search engine marketing) according to conversion data owned by an advertiser, in particular to automatic optimization of keyword pricing of marketing activities distributed on a plurality of Search engines according to total budget and CPA (cost Per action) targets set by the advertiser.
Background
Search Engine Marketing (Search Engine Marketing) is abbreviated as "SEM", and specifically refers to network Marketing or Marketing based on a Search Engine platform. When people use a search engine to retrieve information, advertisers transmit various marketing information to target users through an advertisement platform (SEM platform or channel for short) of the search engine. The demands of SEM advertisers are becoming more and more streamlined, and click behavior on the line is no longer being seen, but rather the actual effects after clicking, such as access depth, registration, ordering, etc., are being more focused. The advertisement placement effectiveness index starts from CPC (cost Per click) and turns to CPA (cost Per action) of effect guidance. The SEM platform of the search engine itself requires that conversion events must be platform trackable, and conversion events that are not trackable, often require manual import into the platform. This is difficult or even impossible for the advertiser. Methods and tools exist that combine single channel reporting data with proprietary data, count the effect (or quality) of keywords, automatically optimize prices, boost keyword ranking, or increase the number of conversions. But still the following problems cannot be solved:
1. the channels cannot be communicated and are in a fracture state. The release data of the keywords of one channel cannot be automatically applied to other channels, the initial price is manually optimized after historical data is evaluated by operators, the optimization process is slow and low in efficiency, and resources are wasted;
2. many advertisers wish to distribute budgets across multiple channels. They wish to optimize the budget allocated to these activities, ultimately achieving an overall CPA goal. The existing method can only adjust the price and budget according to the effect of a single activity, and can not uniformly allocate the budget across a plurality of channels and activities according to the overall CPA target;
therefore, a new method for fusing historical data and offline conversion data of multiple channels and automatically optimizing keyword pricing of the multiple channels according to CPA targets is needed, and the delivery efficiency of SEM advertisements is improved.
Disclosure of Invention
The features and methods of the present invention will be explained based on several embodiments. These examples are merely illustrative of specific implementations of the invention, and technical details are provided to clearly and completely illustrate the features and methods of the invention and not to limit the method of practicing the invention. All other embodiments, which can be derived by a person skilled in the art from the features and methods of the present invention, shall fall within the scope of protection of the present invention. The full scope of the invention is set forth in the appended claims.
1. Related concepts
For a detailed explanation of the present invention, we first define the relevant concepts. As shown in fig. 1, these concepts include: project, SEM platform/channel, campaign, and keyword. The specific definition is as follows:
SEM platform/channel
The SEM platform is also called a channel, is an advertisement marketing platform established by search engine manufacturers, and sells search keywords in a bidding manner. The domestic environment has Baidu, dog search, magic horse, must harmonize 360 and the like. As shown at 115 in FIG. 1 and 130, advertisers may open marketing accounts on multiple SEM platforms in the hope of maximizing reach.
Keywords (keyword)
Query keywords entered by a user in a search page. The SEM platform sells advertisement opportunities (text or picture advertisements are displayed on a search result page) corresponding to the keywords in a bidding mode. A set of keywords may be placed in multiple channels or SEM platforms.
Activity (campaigns) and project (project)
In the present invention, a campaign refers to a bidding strategy that an advertiser establishes for one or a group of similar advertisements on a certain SEM platform. These advertisements share a set of keywords, bids, and budgets that require bidding, as shown at 135. An advertisement may set up multiple campaigns (e.g., 125 and 115) on one SEM platform. We represent the set of active keywords by the name KSet. In the present invention, an item refers to a collection of activities (e.g., 100) distributed across one (e.g., 105 and 110) or more SEM platforms. Activities in the same project share budgets, typically manually deployed across multiple SEM platforms.
Keyword report
Each channel/SEM platform will time a multidimensional data report for campaign statistics. The invention mainly focuses on a keyword report counted by days, which at least comprises the following fields or dimensions:
1) date: date of keyword exposure generation
2) Activity ID: numbering of current activity
3) Key ID, the number of the current key, different keys having different numbers
4) Exposure amount: how many exposures are made during the day
5) Consumption amount: within the day, click the generated advertisement fee
The invention does not limit the format of the keyword report, and the format is only one example, thereby facilitating the subsequent explanation.
Transformation (action) and self-owned data
A conversion is a behavioral event in which an advertiser sees that an average user has become a customer or potential customer. The present invention is directed to behavioral events that cannot be automatically tracked by the channel, but which can be tracked and recorded by the advertiser's system, such as registration, payment, and incoming calls, among others. The self data refers to all conversion events which are automatically collected by the advertiser system through SEM marketing diversion;
the integrated self-owned data SEM platform data report is the basis of automatic keyword price optimization. Each advertiser may need to develop different integration rules.
Budget and consumption
A budget is an amount that an advertiser can spend allocated for a campaign or project. The consumption is the amount already spent. The present invention requires advertisers to allocate budgets by date and to deliver evenly every hour. For example, the total budget is 2400 dollars, with an average delivery of 10 days, a budget of 240 dollars per day, and a delivery of 10 dollars per hour. The following "budget" (whether actual or estimated) refers exclusively to a single day budget (referred to as a daily budget) if not specifically stated. Consumption is also the same, and refers to consumption by a single day.
CPA and rCPA
CPA (cost Per Action) is a performance assessment index, measured as the average consumption of a single transformation (Action). The specific definition is as follows:
CPA = total consumption/total number of conversions. (formula 1)
The smaller the CPA is, the better the dispensing effect is, and conversely, the dispensing effect is worse. rCPA is the reciprocal of CPA, and the smaller the rCPA is, the worse the dosing effect is, and conversely, the better the dosing effect is. In order to facilitate calculation, the rCPA is used as a main effect index. For convenience of description, the two indices will be represented by the lower case cpa and rcpa.
2. Common system architecture and optimization method
Fig. 2 is a common architecture of a marketing system that supports automatic bidding of keywords. The present invention is not limited to a specific system, and the present structure is only the explanation basis of the present invention, and the scope of the present invention is not limited thereto. As shown in fig. 2, the keyword and activity data pull module (210) periodically pulls activity and keyword related reports from a plurality of SEM platforms/channels (230) to the report database (215). The integrated owned conversion module (240) reads the report data of each activity, finds the converted data from the owned data (235) according to the rules, and merges the converted data into the report. The keyword price optimization module (220) independently optimizes keyword prices and sometimes activity consumption for each activity based on the integrated proprietary data. As shown by block (220), the keyword price optimization process for each activity is logically independent and unassociated. After converting the optimized structures into configuration parameters for the SEM platform, 225 synchronizes the configurations to the SEM platform (230);
FIG. 3 is a diagram of a common keyword price optimization method for a single campaign. According to the report with the conversion data, a plurality of indexes (315) including consumption, click rate, conversion rate and the like are counted for each keyword in the activity, and then according to the indexes, the price of each keyword is automatically adjusted (320). Generally, according to a predefined rule or a certain prediction formula, the price of the keyword with poor performance indexes such as click rate and conversion rate is reduced, the price of the keyword with good performance indexes is increased, and the budget of the activity may be adjusted. The method like fig. 3 can only adjust the price and budget according to the effect of a single campaign, and cannot uniformly allocate budget and optimize keyword prices for multiple channels and campaigns according to the overall CPA goal of the project. In addition, the released data of the same keyword on a plurality of channels cannot be shared and utilized.
3. Cross-channel optimization method for effect target guidance
FIG. 4 illustrates the nested structure of the project, activity and keyword three-class optimization processes of the present invention, which, although different, are not completely independent. Under budget constraints, with the effect CPA objective as a benchmark, the project optimization process dominates the optimization of multiple activities: the consumption of high CPA (poor effect) activities is reduced and the reduced consumption is allocated to low CPA (good effect) activities. Campaign optimization in turn drives optimization of internal keyword prices: reducing the price of high CPA (poor effect) keywords, reducing their consumption, assigning the reduced consumption to low CPA (good effect) keywords, and raising their price. The following are the main parameters required for the invention:
1)d: optimizing step length of the project, namely the consumption of each iterative process adjustment;
2)Δ: the key consumes the optimal step size and,Δmuch less thand；
3) MAX _ WEIGHT: maximum price weight of the keyword;
4) VMIN: a minimum threshold for moving average consumption. The VMIN of the item is different from that of the keyword, and the VMIN of the item is far larger than that of the keyword;
5) BASE _ spice: the base price of an activity, each activity may have a different base price.
The invention aims at a single cross-channel marketing project, and if a plurality of projects can be processed one by one or in parallel according to the method. Fig. 5 is the general structure of the present invention. Downloading daily reports of keywords from one or more channels at a fixed time point (within the recommended time period of 0-5 points) every day, integrating self-contained conversion data, and then providing the keywordsWords, activities, and projects are calculated and counted to generate a plurality of metrics (510). The present invention does not specify the particular method of collecting and integrating the transformed data, and different methods may be used in different situations. Optimization procedure, dependent on step size parameterdGradually increasing and decreasing the consumption of activity. Parameter(s)dFine tuning can be done according to different items. In most of our examples, we used 0.1% of the total consumption of the project. Such as the consumption of the item being 2000 dollars,dis 2-membered;
after the statistical indicator is ready, the batch is used (numbered by time sequence)tMark) step by step, optimizing the consumption distribution of the items, ensuring a good distribution of consumption between activities: consumption is positively correlated with effect. Reducing project total consumption per batchd: reducing consumption of multiple activities and optimizing keyword prices according to rCPA from low to high until total amountdIs used up (515). After the adjustment is finished, the total consumption of the project is improvedd: according to the rCPA from high to low, the consumption of a plurality of activities is promoted, and the price of the keyword is optimized until the total amountdAnd (4) the product is used up. If the CPA of the entire project is decreased (525), it indicates that there is room for optimization, the project status of the current batch (the indexes of project, activity, and keyword generated by each optimization) is maintained, and step 515 is repeated to start a new round of new optimization batch (step (a) ((525)t+1). If the CPA of the project is not reduced, indicating that there is no room for optimization, the state of the project is rolled back to the previous batch: (t-a state (527) of 1);
after the consumption distribution is optimized, if the CPA of the project exceeds the target value or the project consumption is too much and exceeds the budget, the whole consumption must be reduced, the price of the keyword is reduced, and the flow with poor effect is removed. As shown at 530 and 535, this is a function of step sizedThe process is gradually cut off until the consumption does not exceed the budget and the CPA is not higher than the target. Next, if CPA is below the target value and the project consumption is below the budget, it is stated that the flow can also be increased. As shown at 540 and 545, in stepsdAnd gradually increasing the consumption of the project until the CPA of the project is consistent with the target or the consumption is consistent with the budget. The above-mentioned optimization is completedAnd then, according to the optimization result, outputting the price of the keyword and the budget configuration of the activity to a synchronization module, updating the price and the budget configuration to a corresponding channel, and influencing the real delivery process on the channel. The next morning a new round of data extraction, calculation and optimization is started. The processing and calculation method of each step will be described in detail later.
Statistical index
This section explains the category and calculation method of the keyword, activity and project statistical index in section 510 of fig. 5.
Actual consumption of (rcost) And the actual conversion (raction)
The present invention assumes that the keyword report can be obtained on a daily basis. For simplicity of explanation, we use timing numberingtThe sequence marks the specific date after the project, activity, or keyword is online. If not on-linet= 0, the first day after log-on ist=1, the next dayt=2, and so on. We usercost[t]Andraction[t]respectively representing items, activities or keywordstThe actual daily consumption and the number of conversions, usercost_total[t]Andraction_total[t]represents a cutoff totAll the days ofrcostAndractiona summary of (1). In the following description, if the latest date (or current date) is referred to, we will omitt。
Moving average consumption: (ma_cost）
The average of the actual consumption of a plurality of days in succession in the recent past, smoothes the short-term fluctuations, estimates the medium-and long-term trend of the consumption. Adopted recentlyn>The moving average of actual consumption for =1 date is specifically defined as follows:
This approach is used for both the moving average consumption of items and keywords. In the following description, we will omit the latest moving averaget. The invention is not limited by the weightw i The setting method of (3) can be selected from various schemes. Hair brushIn one embodiment of the invention, a simple moving average method is used, i.e.w i =1/n. Another embodiment adopts the method of EMA (exposing Moving average), namelyw i =α *(1-α) i 。
Moving average rCPA (ma _ rCPA)
For the rCPA of the project and the keyword, a special VWMA (volume Weighted moving average) calculation method is uniformly adopted, and the specific definition is as follows:
Where VMIN is the minimum threshold for consumption and BASE is the initial rCPA value. In the following description, we will omit the latest moving averaget. The VMIN of the item is different from the VMIN of the key, and the VMIN of the item is much larger than the VMIN of the key. The BASE of the project is one>A constant of 0 is typically set to the reciprocal of the target CPA. The BASE of the key is a function. Through the BASE function, the release data of the keywords of one SEM platform can be automatically applied to other SEM platforms, the initial price of the keywords is optimized, and the release efficiency is improved.
BASE of keywords
Firstly, calculating the amplitude of the increase or decrease of rcpa of a certain keyword in an activity compared with the whole: rcpa _ lift. Is provided withxIs a certain activity that is the result of the activity,kis one of the keywords, and is specifically defined as follows:
On the basis of the above, for a certain activityaIs a key ofkThe BASE function is defined as follows:
Its core function is currently activema_rcpaOn the basis of the same key words for other activitiesrcpa_liftTo perform weighting correction.
Reference budget, reference consumption (cost), reference rCPA, and reference CPA targets
Reference consumption (cost) and reference rCPA are the main indicators used for the optimization method. Before the project optimization process begins, the reference consumption of the initial keyword is equal to the moving average consumption (ma_cost) The initial reference rCPA is equal to the moving average rCPA (ma_ rcpa). Movement ofaIs a weighted average of the reference consumption and reference rcap of its keywords, defined as follows:
ItempThe reference consumption and reference rCPA of are defined as follows:
Consumption (cost) and rCPA in the following description refer to reference consumption and reference rCPA, if not specifically noted. These indices are recalculated using equations (6) and (7) for each optimized batch.
The reference CPA target is the CPA value for which the project optimization process is directed, defined as follows:
The CPA target in the following description is a reference CPA target if not specifically noted. The reference budget is a budget value for the project optimization process, and is defined as follows:
If not specifically noted, the project budget in the following description is the reference budget, rather than the actual budget specified by the advertiser. The reference CPA target and the reference budget are fixedly calculated before the optimization starts and are not recalculated in the optimization process.
Effect area and effect area table
In order to alleviate the prediction fluctuation caused by small data quantity, each active keyword is classified into a group according to rcpa binning (data binning or data buffering): the "effect area". Each effect region having a low boundarylow_vHigh boundaryhigh_v. Adjacent effect areas a and b, satisfya.high_v=b.low_v. The keywords in each effect area are defined as a setKSetSatisfy any one of the keywordskϵ KSet satisfies:k.rcpa>low_vANDk.rcpa<=high_v. Ork.rcpa>high_v(if the current effect areahigh_vLargest among all packets). The invention is not limited to the method of box separation, and can be an equal division method or a non-equal division scheme.
Project and activity states
Each time the consumption of the activity is reduced or increased, some indexes of the keywords and the activity are updated, including: reference consumption (cost) Reference rCPA/CPA (rcpa/cpa) And price weight of keyword: (weight). These metrics are referred to as the active state, and the last active state is retained for each iteration. Similarly, each time the consumption of a project is reduced or increased, the reference consumption of the project is updated (cost) Reference rCPA/CPA (rcpa/cpa) And the status of all activities. We refer to these indicators together as the status of the project. Each iteration retains the state of the last item.
Reducing and increasing project consumption
The process of FIG. 6 "reduce project consumptiond"(610) is the process of reducing the consumption of multiple activities by rCPA in steps 515 and 535 of figure 5. The process of FIG. 6 "promotes project consumptiond"(650) is the process of boosting multiple activity consumptions by rCPA in steps 520 and 545 in fig. 5;
the input to process 610 is the total amount decrementedd. Record the total amount (615) and then take the rCPA minimum with cost: (Consumption) and down-regulates the consumption of the activity according to the credit and recalculates its consumption and rCPA (635). Finally, the remaining downward credit is calculated (640). If credit remains (620), then continue to look for the next lowest rCPA activity, down-regulating its consumption until credit is exhausted. Finally recalculating the consumption of the project and rCPA (643);
the input to process 650 is the total amount of promotiond. The total credit is recorded 655, then the highest rpca activity is accessed 665, the consumption of the activity is increased according to the credit, and the consumption and rpca are recalculated 675. Finally, the remaining liftable quota is calculated 680. If credit remains 660, then continue to look for the next highest activity of rCPA and up-regulate its consumption until credit is used up. Finally, the consumption of the project and the rCPA are recalculated (683).
Adjusting consumption of individual activities
Fig. 7 shows a procedure how the individual activity consumption is adjusted (reduced/increased). 635 and 675 in FIG. 6 invoke this process, reducing or promoting the consumption of a single activity. This process has two basic parameters:gthe total consumption amount of the descending/ascending of the activity,Δis the consumption adjustment step size of the effect region, compared to the active adjustment step size in FIG. 5dIs small. Similar to consumption distribution (515, 520, 525 and 523) of the optimization items, steps 715, 720, 723 and 725 gradually optimize consumption among the effect areas in multiple batches, so that the total consumption of the keywords in the effect areas is positively correlated with the effect of the consumption areas;
firstly, according to the rCPA of the effect area, the price and consumption of the keywords in the plurality of effect areas are sequentially reduced from small to large until the consumption step lengthΔIs exhausted (715). Then sequentially increasing the price and consumption of keywords of the effect area according to the increase and decrease of the rCPA until the keywords are used upΔUntil now. If the CPA of the entire campaign has decreased, indicating that there is room for optimization, the campaign status of the current lot is retained, and the process returns to step 715 to begin a new round of new optimization lot (t+1). If the CPA of the project is not lowered, indicating that there is no room for optimization, the state of the project is rolled back to the previous batchNext (t-state (527) of 1), continuing the following optimization procedure;
then adjust the total amount according to the activity consumptiongThe key word prices and consumptions of the plurality of effect regions are escalated 745 or reduced 740. The limit of each adjustment is fixed to be the consumption adjustment step length of the effect area at mostΔUntil the adjusted total amount of consumption is exhausted (750, 730).
Adjusting consumption of an effects area list
Fig. 8 illustrates how the keyword price and consumption of a plurality of effect zones within the list of effect zones of a single activity are adjusted according to the rpaa size of the effect zone, which is a refinement of 715, 720, 740, and 745 of fig. 7.The parameter is an adjusted amount. Firstly, acquiring an active effect area table (815), then gradually adjusting the prices and consumption of a plurality of effect areas (825, 830) according to the requirements of promotion or reduction, and recalculating the active consumption (840) after the adjustment is finished;
if it is a decrease (825), then the rCPA minimum (least effective) is taken first and the area of effect consumed is based onAnd adjusting the price and consumption of each keyword in the effect area, and then recalculating the attributive effect area of the keyword according to the rCPA adjusted by the keyword. Can not be adjustedUnder the condition of (1), circulating again until the adjustment is finished;
if it is raised (830), the maximum effect area of rCPA is taken first, based onAnd improving the price and consumption of each keyword in the effect area, and then recalculating the effect area to which the keyword belongs according to the rCPA adjusted by the keyword. Can not be adjustedUnder the condition of (1), circulating again until the adjustment is finished;
and finally recalculating the consumption of the activity and the rCPA based on the consumption of the keywords and the rCPA.
Adjusting bids and consumptions of single effect region keywords
FIG. 9 shows how to adjust the amount of money according to a certain adjustmentL(positive numbers represent promotion and negative numbers represent reduction), the price and consumption of keywords within a single effect region are reduced or promoted. Are refinements of 825 and 830 of figure 8.KSetIs the set of key objects of the current effect region,MAX _ WEIGHT is the maximum keyword bid WEIGHT. It requires two conversion coefficients:cost2priceandprice2rcpathe former calculates the consumption lifting amplitude according to the lifting amplitude of the price, and the latter calculates the variation amplitude of the rCPA according to the lifting amplitude of the price;
first according toLCalculating the magnitude of the adjustment relative to the total consumption of the current effect regionq(915) Positive numbers represent increasing and negative numbers represent decreasing. Then, for each keyword object in the effect areaKAdjusting the price weight(weight) And consumption of(cost) AndrCPA(920). For each keyword, the keyword is processed firstcostIs adjusted to<1>) And then reuse the consumption price conversion coefficientcost2priceCalculate a new price weight (<2>、<3>) Reuse of the conversion factor of the price rCPAprice2rcpaCalculating a new rCPA value according to the adjustment range of the price weight (<4>) And updating the price weight (b)<5>). After the adjustment is completed, the cost of all keys is used as the cost for calculating the current effect area (925).
Conversion coefficients cost2price and price2rpca
The consumption of the keyword is positively correlated with the price, and the consumption increase means that the price is increased. We use a single conversion coefficientcost2priceScaling the adjustment margin between consumption and price. Consumption adjustment amplitude multiplied by conversion factorcost2priceThe adjustment range of the price is the adjustment range. This factor controls the sensitivity of keyword price adjustment and the process optimization speed.cost2priceThe larger the consumption adjustment, the larger the adjustment range of price, and the more intense the market competition. The invention does not specify a coefficient value, and the implementer can optimize the coefficient according to the specific situation of the advertisement market. In general, the coefficient needs to be 1 or more. The market competition of one embodiment of the invention is not strong, namely 3.0 is obtained, and the market competition of the other embodiment is fierce, namely 1.0 is directly obtained;
in most cases, there is a negative correlation between the price of the keyword and the rCPA, with price escalation meaning that the rCPA will drop. We use a single conversion coefficientprice2rcpaAdjustment of conversion between price and consumptionAnd (4) degree. Price adjustment amplitude multiplied by conversion coefficientprice2rcpaThe amplitude of rcpa is adjusted in the opposite direction.price2rcpaThe larger, representing the same unit of price adjustment, the amplitude of the reverse direction adjustment of rcap. The invention does not specify specific coefficient values, which the implementer can optimize according to the ad market specifications. One embodiment of the invention automatically extracts the pairing data of the price and the rcpa from the historical release data and automatically calculates the optimalprice2rcpaThe value is obtained.
Budget for price and activity of output keywords
The 550 of fig. 5 exports and synchronizes the price and active budget configurations of the keywords to multiple SEM platforms. The price of the output keyword is equal to:
BASE_PRICE *weight(formula 10)
Where BASE-prime is the BASE PRICE for the activity,weightis the price weight of the keyword. Reallocating the daily actual budget according to the proportion of the reference consumption of each activity to the total reference consumption, which is specifically defined as follows:
Because how the daily actual budget for each activity is allocated when the project is first started does not generally affect the final result, the invention does not make provisions, and can be all the same, and also can differentiate the budgets among activities according to the actual delivery experience.
Drawings
FIG. 1 illustrates the relationship between underlying concept items, SEM platforms/channels, campaigns, and keywords
FIG. 2 is a common system architecture of a marketing system supporting automatic keyword bidding
FIG. 3 depicts a common keyword price optimization method for a single campaign
FIG. 4 illustrates the nested structure of the project, activity and keyword three-class optimization process of the present invention
FIG. 5 is a main flow of an effect target oriented cross-channel keyword price optimization method
FIG. 6 is a refinement of the method of FIG. 5, showing how the overall CPA of a single project is optimized by adjusting the consumption of multiple activities
FIG. 7 is a refinement of the method of FIG. 6, showing how consumption of a single activity may be optimized based on adjustment credits
FIG. 8 is a refinement of the method of FIG. 7, showing how the consumption of multiple effect regions within a single campaign may be optimized based on adjustment credits
FIG. 9 is a refinement of the method of FIG. 8, showing how the bids and consumptions of all keywords within a single effect region are adjusted according to an adjustment amount.
Claims (13)
1. In a system or environment supporting single marketing project advertising on a marketing platform of multiple search engines, a computer-implemented method for automatically optimizing keyword prices for each marketing campaign, after specifying CPA (cost Per action) objectives, overall budgets and other parameters, and ultimately achieving overall CPA objectives; the method is characterized in that the optimization is carried out at a fixed time point each day according to the following steps:
a) extracting daily keyword reports of a plurality of channels, integrating self-owned conversion data of advertisers, and respectively calculating and counting various statistical indexes for keywords, activities and projects, wherein the statistical indexes comprise:
1) calculating a moving average of consumption for the project and keywords: (ma_cost) And a moving average rCPA index (fma_rcpa)；
2) Initial reference consumption of a keyword: (cost) Equal to the moving average of consumption (ma_cost) The initial key reference rCPA (rCPA) is equal to the moving average rCPA metric (ma _ rCPA);
3) calculating a reference consumption (cost) and a reference rCPA index (rCPA) of the activity according to the result of 1);
4) calculating a reference consumption (cost) and a reference rCPA index (rCPA) of the item according to the result of 3);
5) calculating a reference budget and a reference CPA target of the project;
6) grouping all keywords of each activity by adopting a binning method according to the size of the reference rCPA to form a plurality of effect areas (keyword sets);
b) reducing the reference consumption of the project by using a fixed small amount, and then improving the reference consumption of the project by using the same amount; looping the decreasing and increasing steps until the reference CPA of the project no longer decreases;
c) after step b is completed, if the reference CPA of the project exceeds the reference CPA target or the reference consumption of the project exceeds the reference budget, the same method as the method for reducing the reference consumption of the project in step b is adopted, and iteration is carried out in batches by the same quota until the reference consumption of the project does not exceed the reference budget and the reference CPA is not higher than the reference CPA target,
d) after the step b is finished, if the reference CPA of the project is lower than the reference CPA target and the reference consumption of the project is lower than the reference budget, adopting the same method as the method for improving the reference consumption of the project in the step b, and carrying out batch iteration promotion by using the same quota until the reference CPA of the project is consistent with the reference CPA target or the reference consumption is consistent with the reference budget;
e) outputting the keyword price and the daily actual budget of the campaign to a marketing platform of the search engine.
2. The moving average consumption of step a of claim 1: (ma_cost) The calculation is calculated by the following formula:
whereinrcostIs the consumption of the liquid per day and,nis the number of days of putting in the tank,tis the current date of the day or week,w i is the weight.
3. The moving average rCPA indicator of step a of claim 1 (a) ((b))ma_rcpa) The following formula is used for calculation:
whereinVMINIs the minimum threshold for the amount of consumption,rcost_totalis cut off totThe actual consumption of the fuel is summarized,ractionis the actual number of daily conversions,raction_totalis a cut-offtSummarizing the actual conversion number; BASE is the initial rCPA value; the BASE of the project is one>0, the BASE of the key is a function, calculated according to the following formula: for a certain activityaIs a key ofkThe BASE function is defined as follows:
whereinrcpa_liftIs defined as:
whereink xIs referred to as movementxKey word ofk。
4. Reference consumption of activity of step a of claim 1: (cost) And a reference rCPA index (rcpa) The calculation method is as follows:
whereinaIs an activity of the person to be examined,a.Ksetis a home activityaThe set of the keywords of (1),k.rcpais the reference rcap of the keyword,k.costis the reference consumption of the key.
5. Reference consumption of the item of step a of claim 1: (a)cost) And a reference rCPA index (rcpa) The calculation method is as follows:
whereinaIs an activity of the person to be examined,a.rcpais a movementaWith reference to the rpac of (1),a.costis a movementaIs consumed.
6. The method of calculating a reference budget and reference CPA target for the project of step a of claim 1, comprising:
whereinrcpaAndcostis the project's reference rcap and reference consumption.
7. The method of reducing the consumption of the project references with a fixed small value unit in step b of the method of claim 1, characterized by: taking the lowest activity of the reference rCPA each time, reducing the reference consumption of the reference rCPA, and recalculating the reference rCPA; if the credit is not used up, the lowest activity of the reference rCPA is taken again, the operation is executed again, and the process is circulated until the credit is completely used up.
8. The method of claim 1 wherein the step b of the method of raising the consumption of the item reference by a fixed small amount is characterized by: taking the highest activity of the reference rCPA each time, improving the reference consumption of the reference rCPA, and recalculating the reference rCPA; if the credit is not used up, the highest activity of the reference rCPA is taken again, the operation is executed again, and the process is circulated until the credit is completely used up.
9. The method of reducing or increasing consumption of an active reference according to claims 7 and 8 is characterized by:
a) reducing the reference consumption of the fruit area by using a fixed small amount, and then improving the reference consumption of the fruit area by using the same amount; cycling until the active reference CPA is no longer decreasing;
b) after the step a is finished, if the reference consumption of the activity is to be reduced, adopting the same method as the reference consumption of the effect area reduced in the step a, and using a fixed small amount to perform iterative subtraction in batches until the total amount required to be reduced is reached;
c) after step a is completed, if the reference consumption of the activity is to be increased, the same method as the reference consumption of the effect area in step a is adopted, and the iterative increase is performed in batches by using a fixed small amount until the total amount required to be increased is reached.
10. The method of claim 9 for reducing the reference consumption of the effect region is characterized by: taking an effect area which is related to key word consumption and has the lowest reference rCPA each time, reducing the reference consumption and price weight of all the key words belonging to the effect area, and up-regulating the reference rCPA, then recalculating the reference consumption of the effect area, and recalculating the effect area to which the key words belong; if the credit is not used up, the operation is executed again by taking the effect area which is related to the key consumption and has the lowest reference rCPA, and the loop is executed until the credit is completely used up.
11. The method of claim 9 for increasing the reference consumption of the effect region has the following features: taking the highest effect area of the reference rCPA each time, improving the reference consumption and price weight of all keywords belonging to the effect area, lowering the reference rCPA, recalculating the reference consumption of the effect area, and recalculating the effect area to which the keywords belong; if the credit is not used up, the highest effect area of the reference rCPA is taken again, the operation is executed, and the loop is executed until the credit is completely used up.
12. The method of claims 10 and 11 adjusts (decreases or increases) the reference consumption of the keyword by the same adjustment amplitude as the required effect area reference consumption, the adjustment amplitude of the price weight is equal to the adjustment amplitude of the reference consumption multiplied by an amplitude conversion factor, and the adjustment amplitude of the reference rcap is equal to the adjustment amplitude of the price weight multiplied by another amplitude conversion factor.
13. The keyword price of step e of claim 1 equal to the base price of the campaign multiplied by the price weight of the keyword, the actual daily budget of the campaign calculated according to the formula:
where cost is the reference consumption of activity.
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010499063 | 2020-06-05 | ||
| CN2020104990631 | 2020-06-05 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111598631A true CN111598631A (en) | 2020-08-28 |
Family
ID=72182834
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010646498.4A Pending CN111598631A (en) | 2020-06-05 | 2020-07-09 | Cross-channel keyword price optimization method with effect target oriented |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111598631A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113506146A (en) * | 2021-09-10 | 2021-10-15 | 北京易真学思教育科技有限公司 | Data adjusting method and device and computer readable storage medium |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101208716A (en) * | 2005-04-26 | 2008-06-25 | 管理动力学有限责任公司 | A method of digital good placement in a dynamic, real time environment |
| CN102375810A (en) * | 2010-08-04 | 2012-03-14 | 阿里巴巴集团控股有限公司 | Estimation value feedback method and server for entire account search keywords |
| CN104252680A (en) * | 2013-06-28 | 2014-12-31 | 麦奇数位股份有限公司 | Keyword automatic pricing method and search engine marketing system |
| KR20170037760A (en) * | 2015-09-25 | 2017-04-05 | 주식회사 에코마케팅 | System and method for bidding automatically using correlation of bidding related data in keyword advertisement, and bidding management server |
| CN106960359A (en) * | 2017-02-10 | 2017-07-18 | 上海智子信息科技股份有限公司 | A kind of full-automatic bid optimization method and system calculated based on streaming |
| CN107481062A (en) * | 2017-08-21 | 2017-12-15 | 小草数语（北京）科技有限公司 | The distribution method and device of advertisement putting budget |
| CN108269123A (en) * | 2017-12-29 | 2018-07-10 | 广东神马搜索科技有限公司 | Control method and device are launched in advertisement |
| CN108629626A (en) * | 2018-05-08 | 2018-10-09 | 多盟睿达科技（中国）有限公司 | A kind of search advertisements intelligence trustship method and system based on AI big datas |
| CN109636491A (en) * | 2019-01-25 | 2019-04-16 | 西窗科技(苏州)有限公司 | A kind of optimization method and device that search engine advertisement keyword is launched |
| TW201933229A (en) * | 2018-01-25 | 2019-08-16 | 優愛德股份有限公司 | System and method for automated allocating advertising budgets |
-
2020
- 2020-07-09 CN CN202010646498.4A patent/CN111598631A/en active Pending
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101208716A (en) * | 2005-04-26 | 2008-06-25 | 管理动力学有限责任公司 | A method of digital good placement in a dynamic, real time environment |
| CN102375810A (en) * | 2010-08-04 | 2012-03-14 | 阿里巴巴集团控股有限公司 | Estimation value feedback method and server for entire account search keywords |
| CN104252680A (en) * | 2013-06-28 | 2014-12-31 | 麦奇数位股份有限公司 | Keyword automatic pricing method and search engine marketing system |
| KR20170037760A (en) * | 2015-09-25 | 2017-04-05 | 주식회사 에코마케팅 | System and method for bidding automatically using correlation of bidding related data in keyword advertisement, and bidding management server |
| CN106960359A (en) * | 2017-02-10 | 2017-07-18 | 上海智子信息科技股份有限公司 | A kind of full-automatic bid optimization method and system calculated based on streaming |
| CN107481062A (en) * | 2017-08-21 | 2017-12-15 | 小草数语（北京）科技有限公司 | The distribution method and device of advertisement putting budget |
| CN108269123A (en) * | 2017-12-29 | 2018-07-10 | 广东神马搜索科技有限公司 | Control method and device are launched in advertisement |
| TW201933229A (en) * | 2018-01-25 | 2019-08-16 | 優愛德股份有限公司 | System and method for automated allocating advertising budgets |
| CN108629626A (en) * | 2018-05-08 | 2018-10-09 | 多盟睿达科技（中国）有限公司 | A kind of search advertisements intelligence trustship method and system based on AI big datas |
| CN109636491A (en) * | 2019-01-25 | 2019-04-16 | 西窗科技(苏州)有限公司 | A kind of optimization method and device that search engine advertisement keyword is launched |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113506146A (en) * | 2021-09-10 | 2021-10-15 | 北京易真学思教育科技有限公司 | Data adjusting method and device and computer readable storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108269123B (en) | Advertisement putting control method and device | |
| CA2572582C (en) | Adjusting ad costs using document performance or document collection performance | |
| CN111651722B (en) | Risk assessment method and platform for advertisement putting effect of advertiser | |
| JP2010512604A (en) | Bid optimization in search engine marketing | |
| AU2005242763B2 (en) | Facilitating the serving of ads having different treatments and/or characteristics, such as text ads and image ads | |
| US20150134462A1 (en) | Dynamic bid price selection | |
| CN111263319B (en) | Method and device for reminding balance of mobile phone charge | |
| US20070027761A1 (en) | Application program interface for customizing reports on advertiser defined groups of advertisement campaign information | |
| US20050149396A1 (en) | Online advertising system and method | |
| US20080270223A1 (en) | System and Method for Creating and Providing a User Interface for Displaying Advertiser Defined Groups of Advertisement Campaign Information | |
| CN104252680B (en) | Keyword automatic pricing method and search engine marketing system | |
| WO2010088508A2 (en) | Conversion crediting | |
| CN106960359A (en) | A kind of full-automatic bid optimization method and system calculated based on streaming | |
| WO2009158095A2 (en) | Systems and methods for utilizing assist data to optimize digital ads | |
| WO2013112305A1 (en) | Advertiser modeling | |
| CN106296247A (en) | The online sort method of network information resource and device | |
| CN111899041A (en) | Information delivery processing method, information delivery device, information delivery equipment and storage medium | |
| CN108564404B (en) | Method and device for predicting return on investment of advertisement | |
| CN111598631A (en) | Cross-channel keyword price optimization method with effect target oriented | |
| CN110490649B (en) | Uniform speed advertisement putting method and system | |
| CN107025575A (en) | Brand client KPI predictor methods in sequencing advertisement putting | |
| CN116089401B (en) | User data management method and system | |
| US20210049648A1 (en) | Systems, methods and articles for automated auction-based marketplace for content | |
| CN109685197A (en) | A kind of optimization method based on the power supply set meal for improving particle swarm algorithm | |
| JP2012504270A (en) | Method and system for managing quality of advertising documents |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20200828 |
|
| WD01 | Invention patent application deemed withdrawn after publication |